MAFL 2010 : Round 10 (Coming Soon to a Browser Near You)

Well normally I'd be posting this week's pre-round blog, but, as I type this, the Saints' alleged player sex scandal is still unresolved, at least to the satisfaction of the TAB Sportsbet bookie. As a consequence all St Kilda v Adelaide markets are suspended.

So, we've no ability to back the Saints, which we probably will both on the head-to-head and line markets once we're able. I'll wait 24 hours and see what happens.

 

MAFL 2010 : Round 9 Results

For the third week in succession, those with the Recommended Portfolio were let down by the Saints' performance in the last game of the round. In Rounds 7 and 8 it was the Saints' ineptitude that caused the disappointment; in Round 9 that ineptitude was nowhere to be seen when Investors most needed it.

Not that the weekend's losses can be entirely attributed to the Saints' victory on Sunday evening. Collingwood, the Roos, Sydney and Carlton all did their bits too, offset only - and not completely - by Essendon, Melbourne and Adelaide victories.

This week I'm trialling a new way of presenting the various pieces of information about wagers, tips, team ratings and margin prediction. I call it the MAFL Dashboard.

In the first part of the MAFL Wagering section you can see that only one of the six Funds is currently priced above its buy-in price of $1, and you see that three Funds made money this round (Prudence, Shadow and Heuristic-Based) and the other three Funds lost money. You can also see each Fund's season long performance statistics and price history.

Below this you'll find the same information for the various portfolios. You'll see that the Recommended Portfolio dropped 1.3c this week, the MIN#002 Portfolio rose by 1.9c, and the MIN#017 Portfolio dropped 2.6c. This leaves all portfolios in varying states of unprofitability.

Next is the MAFL Tipping section, which holds all the regular information about the performance of the 13 MAFL tipsters. Short-Term Memory I and Easily Impressed I currently head our tipping competition on 52 from 72 (72%). BKB continues to struggle, scoring just four this week (the Tipping Performance - Last 5 Rounds section has this information), leaving it now six tips behind the leaders.

The column headed Return (Home) records the profit or loss that would have accrued to someone level-stake betting on the tips of the relevant tipster whenever that tipster predicted a home team victory. So far this season, following that strategy with the predictions emanating from Short-Term Memory I would have been the most lucrative, yielding a profit of 9.63 units. Adopting the same strategy with Shadow, as the Recommended Portfolio has been via Shadow's own Fund and via its control of the Heuristic Fund, has yielded a profit of 6.18 units. Incidentally, Shadow made money this weekend and so will retain control of the Heuristic Fund for at least another three rounds.

In the right-hand portion of this section there is information about the recent tipping form of each tipster, as evidenced by its accuracy in each of the last five rounds. The two tipsters currently leading the overall tipping competition also have the (equal) best performances of all tipsters across rounds 4 to 9. In these five rounds they've tipped 28 winners.

Underneath the MAFL Tipping section is the MARS Ratings section. This week, seven teams moved up or down the MARS Ratings ladder. Largest amongst these moves was the two-rung climb recorded by the Hawks based on their 60-point victory over the Blues, which earned them 5.6 ratings points. Currently, the seven highest-ranted teams are also the top seven from the competition ladder, albeit in a different order. The eighth-highest rated team is Brisbane, who sit eleventh on the competition table. Port are eighth on the ladder, but are rated twelfth.

In the chart on the right of this section you track the ratings trajectories of each team.

The final section is the Margin Predictors section and it records the performance of the five MAFL Margin Predictors. BKB still has the season's best margin predictions, whether you rank based on Mean Absolute Prediction Error or on Median Absolute Prediction Error. LAMP is now second on Mean APE while ELO is second on Median APE. The chart on the right shows each Margin Predictor's Mean APE as at the end of every round of the season.

I've not included HELP's data in the dashboard since its unspectacular performance seems unworthy of considered attention.

This week HELP correctly predicted 5 of the 8 line results and moved to a 50% win-loss record for the season. Its very high levels of confidence in all of its predictions meant that its probability scores ended up sub-naive for the second successive week.

In the next day or so I'll be posting another blog in which I'll introduce one more dashboard, this one providing information about team performances.

MAFL 2010 : Round 9

Just when you get into a routine, someone has to go and change something.

This week the TAB Sportsbet bookie decided to post the line markets a day early, leaving me with the dilemma of deciding whether to follow the time-honoured strategy of placing bets around noon on Wednesday or, instead, to follow the other time-honoured strategy of placing bets just after the line markets go up. In the past, save for those weeks where there's been a Thursday game, these two events have coincided.

Worse was the fact that only half the line markets were posted, so I couldn't make all the bets I wanted to anyway.

Well the bets have been placed though we're still waiting for four more line markets to be posted, three of which we'll be wagering on. (I've included these wagers in the MAFL Wagers summary and in the Ready Reckoner, though the latter might need to change if we don't secure a price of $1.90 for the three line market bets we want to make.)

(Update: It's now Thursday evening and we've placed the three remaining line bets. On two of the bets we've secured more favourable terms than I expected: backing Adelaide we wound up getting 7.5 points start rather than 6.5, and backing Sydney we received a price of $1.95 rather than $1.90. In the other bet we took West Coast +6.5 points but only received a price of $1.80. The Ready Reckoner below now reflects these wagers.)

Investors holding the Recommended Portfolio have 23 wagers this weekend - including the three wagers-in-waiting - just one fewer than they had last weekend. New Heritage has six of those wagers, Prudence and Hope have two each, Shadow and the Heuristic-Based Funds have three each, and ELO-Based will have the remaining seven.

In total, Investors with the Recommended Portfolio will have about 19% of their funds at risk this week, again only a tad less than they risked - and significantly lost - last weekend.

In aggregate, Essendon carries the majority of the week's wagers. They're priced at $1.28 and are playing Richmond, a team that appear to have decided to act as this year's quasi-bye, thereby allowing other teams to become accustomed to what they'll face next year when the Gold Coast joins the competition. Hawks fans might have a somewhat different view, of course, as their team went perilously close last weekend to dropping points to the bye.

Other bets are at prices ranging from $1.36 to $2.25, so none of the funds at risk can reasonably be tagged "best forgotten".

Here's the detail:

For those with the Recommended Portfolio, the Essendon v Richmond clash, by virtue of the shedload of cash on the Dons, represents the largest swing between best and worst outcomes. It also represents by some considerable margin the largest downside risk.

The Adelaide v Lions game carries the second largest swing between best and worst outcomes, which it does because it promises the largest upside should the Crows win and cover the spread but because it also threatens the second largest downside should the Crows lose.

Here's the Ready Reckoner in full (subject to amendment [Update: now duly amended] once the three remaining line bets are placed):

For the tipsters it's a weekend of moderate agreement, with unanimity in just one of the games.

  • The Pies are 11-4 favourites over Geelong. The Cats can count BKB, Chi and the Easily Impressed brothers amongst their supporters, though BKB looks likely to have this game as one of its four - yes, four - Games of the Round, and no margin-tipper is predicting a victory margin in excess of 8 points.
  • The Dogs are 14-1 favourites over the Roos, with only Home Sweet Home tipping the upset. All five margin-tippers are on the Dogs with margins ranging from 13 to 24.5 points.
  • Fremantle are 9-6 favourites over the Swans, though all the margin-tippers are predicting a Swans victory. BKB will have this as another Game of the Round.
  • Essendon are the week's only unanimous favourite. They face Richmond and are tipped to win by margins from 16 to 34 points.
  • Port Adelaide are 12-3 favourites over the Dees. Three of the five margin-tippers - LAMP and BKB being the exceptions - are tipping Port victories, including ELO, who is tipping Melbourne by just 1 point, making this its Game of the Round.
  • Adelaide are 10-5 favourites over the Lions. This is the Game of the Round for Chi (tipping a Crows victory by 2 points), LAMP (tipping a Lions victory by 1 point) and BKB (tipping a Lions victory by 6.5 points)
  • Carlton are 11-4 favourites over the Hawks and are tipped by all five margin-tippers with margins ranging from 10 to 27 points.
  • St Kilda are 9-6 favourites over the Eagles. This is the final Game of the Round for BKB. None of the margin-tippers are suggesting that either team will win by a sizeable margin - the largest tipped victory margin for St Kilda is 13 points and for the Eagles is 6 points.

The average number of dissenting tipsters per game is, therefore, about 3.5. This fact, coupled with the bookie's generally miserly points start offerings, suggests that this will be a difficult round to tip. Only the Dons and the Dogs might reasonably be considered 'bankers' and four, maybe five of the remaining six contests have a coin-tossy feel to them.

Whilst seasoned judges might be expressing high degrees of uncertainty about the head-to-head results, HELP's expressing near-certainty about the result of all eight games when adjusted for points start.

Remarkably, all eight HELP predictions are for the away teams. In contrast, the ELO-Based Fund will be making seven wagers on home teams. I think we all know which tipster we want to be incorrect more often.

This week I'll finish by providing my short version of the competition ladder, from which I'd like to highlight the Saints' performance. Despite losing three of their last four games they remain in the top eight. The table shows that this is not because of their points-scoring ability (for which they're ranked equal fourth last) but is instead attributable to their defensive skills: they've allowed only 583 points in their eight games, including only 10.5 goals per game.

 

MAFL 2010 : Round 8 Results

Wow. It feels like I'm about to compose a eulogy.

Silly of me really, in hindsight, to hope that the Saints - who, I remind you, should probably be the reigning premiers - could add a mildly cheery note to an otherwise gloomy weekend's wagering. Instead, their failure merely compounded the misery that saw those with the Recommended Portfolio turn a profit in just three of the eight contests, wiping 11.2% off the value of their Investment.

Investors with other portfolios fared no better, MIN#002 dropping 11.8% and MIN#017 dropping 23.2%, leaving them, as those with the Recommended Portfolio, solidly in the red.

Still, it's only Round 8, so there's plenty of time to forgive and forget.

Here's the summary of each portfolio's performance this week and for the season as a whole:

So far this season I've not provided a graph of Fund performance, but with every Fund now having been active for at least three rounds, now seems like a good time to produce one, so here it is.

I do wish that those lines looked a little less cygnine, but they are what they are, I suppose.

Here are the game-by-game details of the weekend's wagering:

For the third week in succession just five favourites took the competition points - a phrase that's at least still synonymous with victory in the AFL if not the NRL - leaving BKB now five games behind our tipping leaders on 47 from 64 (63%): Easily Impressed I, which bagged seven this weekend, and Short Term Memory I, which bagged six.

Easily Impressed II is in clear third, on 45 from 64 (70%), and Follow the Streak is in fourth, on 44 from 64 (69%).

Home teams had yet another poor showing this week, causing Home Sweet Home to register just a single correct tip. It might have been two from eight but, there being no true home team in the Dogs v Swans game, the Home Sweet Home heuristic goes with the team higher on the ladder, which in this case was the losing Swans and not the notionally at-home Dogs.

The Heuristic-Based Fund lost money this week, so Shadow is now just two loss-making rounds away from surrendering control of the Fund to another heuristic. Based on current performance, the heuristic taking over the Fund would be Short Term Memory I.

Though losing this week, the Heuristic-Based Fund has recorded a 29% increase for the season to date. As the tables below demonstrate, much of this return has been due to Melbourne, Carlton and Fremantle, and only four teams have been loss-making for this Fund (Hawthorn, Port, St Kilda (barely) and the Dogs).

The New Heritage Fund has suffered from untimely losses perpetrated by seven of the teams, most painfully from those of St Kilda, Brisbane and Melbourne. Prudence has lost money on six teams, and most of all on St Kilda, Fremantle and Melbourne.

Melbourne, Port, St Kilda and the Dogs have been the sole value destroyers for the Shadow Fund, while the Lions, Fremantle and Melbourne have provided a similar service to the Hope Fund. ELO-Line has lost money on six teams, with the Saints inflicting the greatest net loss.

You might have noticed the names Melbourne and St Kilda cropping up regularly in the last few paragraphs discussing Fund losses. Small wonder then that they're prominent when we look at team performance by portfolio.

Investors with the Recommended Portfolio can thank Sydney, the Roos, Collingwood and Geelong for generating the greatest increments to their wealth, and can blame a slew of teams including the Saints, Lions, Dees and Freo for underperforming at inopportune times.

The MARS Ratings are particularly interesting this week, as five teams - Geelong, Collingwood, the Dogs, St Kilda and Carlton - have now established a sizeable ratings point lead on the rest of the field.

Over the course of the weekend, six teams changed position on MARS ratings, three recording rises and three recording falls. Essendon and West Coast enjoyed the weekend's only 2-spot rises, and Port Adelaide and Melbourne suffered the weekend's only 2-spot falls.

The Saints have dropped almost nine ratings points in two weeks and now find themselves in fourth spot and in danger of dropping to fifth should they have another poor showing against West Coast and should the Blues perform well against Hawthorn.

On margin predicting, BKB has retained its sub-30 MAPE and its preeminent position on MAPE and on Median APE rankings.

LAMP is in second spot on MAPE, though it's last on Median APE, a full point behind Chi. ELO holds second spot on Median APE, but trails BKB by 1.5 points and needs four games of sub-28 point performance to drag its Median APE down to BKB's current level.

HELP had a horrible week on line tipping and probability estimation, picking just two line-winning teams and scoring at a sub-naive level on all three probability scores, mostly because three of its four most confident tips proved to be incorrect.

This week I'll finish what's already been a long blog with a review of teams' quarter-by-quarter performances.

No team has won all eight of any single quarter. The best performance in a quarter has been the Dogs in second quarters where they've won seven of eight and registered a 176.6 percentage.

Two teams, however, have lost eight from eight for a particular quarter: Richmond have lost all eight third terms, and Hawthorn have lost all eight final terms.

Summing across all four quarters, Collingwood now have the best record of all teams, having won 22 of 32 quarters and drawn 1. Geelong are next best (on 22 from 32) and then Carlton (on 20 from 32).

Richmond, by some distance, have the worst performance. They're averaging less than a single winning quarter per game.

MAFL 2010 : Round 8

Some weeks there's a period where I'm 5 years old again and I hold my breath and close my eyes while I run the Fund scripts and partly fear, partly hope for the worst. This week that apprehension was tied to just how much we'd have riding on the Tigers who were being offered by TAB Sportsbet at $4.50 and taking on the Hawks at the G.

When I peeked and exhaled I found that the Funds commanded ... no head-to-head bet on the Tigers. We did have a 5% line bet though, but with 30.5 points start, I'll take that as a reasonable proposition.

In total, those with the Recommended Portfolio have an even two-dozen wagers this week, five of them struck by Prudence and representing about 15% of that Fund, four from each of Shadow (20%), Heuristic-Based (20%), New Heritage (38%) and ELO-Line (20%), and three from the Hope Fund and representing about 11% of that Fund.

In total, Investors with the Recommended Portfolio have about 21% of their total funds at risk. I tell you, money in that portfolio just turns its little tail off.

Only one of the week's bets is at a price in excess of $2, and that's Hope's 5.5% wager on the Lions at $3 facing the Cats but playing at home at the Gabba. Frankly, I like those odds. The remaining wagers are at prices from $1.36 - on Melbourne, of all teams, head-to-head facing the Eagles - up to $1.90, on Fremantle, head-to-head playing the Pies, and on all four line bets.

Here's the detail:

For those with the Recommended Portfolio, the round's last game, which is the Saints v Dons clash, is the game that represents the largest swing between best and worst outcomes. It also represents the largest downside risk as a loss would set 4.8% of the Portfolio ablaze. At the other end of the spectrum, the Lions v Cats game offers the greatest upside. If the underdog Lions win, the Recommended Portfolio will leap by 2.65%.

So then, what other threats and promises does this week's Ready Reckoner enfold?

For those holding the Recommended Portfolio it's another weekend where the outcome of most games carries a wealth swing of 2.5% or more. In consequence it's another weekend of baldly partisan viewing.

The tipsters are broadly in agreement for yet another week, although in none of the games is there unanimity amongst them.

  • The Pies are 13-2 favourites over Freo, with only Home Sweet Home, HAMP and LAMP separate from the herd. LAMP is by no means overwhelmingly committed to the home team Freo, however, and has them winning by just 5 points, making this its Game of the Round. The bookies too are only weak supporters, though they're on the Pies, and look set to have this as one of their two Games of the Round.
  • The Dogs are 12-3 favourites over the Swans. All five margin-tippers are on the Dogs with margins ranging from 8.5 to 25 points.
  • Melbourne are 12-3 favourites over the Eagles. In this game, predicted Eagles victory margins range from 11 to 23 points.
  • Geelong are 13-2 favourites over the Lions. HAMP has the Cats winning by just 6 points, making this its Game of the Round.
  • Adelaide are 8-7 favourites over the Roos. All the margin-tippers are on the Roos, however, including ELO, who is tipping the Roos by 2, making this its Game of the Round.
  • Hawthorn are 13-2 favourites over the Tigers. Chi is one of the two tipsters predicting a Tigers win and he has them winning by just 2 points.
  • Port Adelaide are 11-4 favourites over the Blues. Chi's on the Blues but tips just a 1-point win, so this is his Game of the Round. It also looks likely to be the bookie's other Game of the Round, with Port set to be offering just 6.5 points start.
  • St Kilda are 13-2 favourites over the Dons, though heaven knows why after the way they played on Monday night. The margin tipsters are particularly confident and are prophesying margins ranging from 17.5 to 31 points. Let's hope they're right.

Average starts this week are again quite low. All but one of them are 18.5 points or fewer, the highest being the 30.5 points start that the Tigers enjoy.

Finally, to HELP, who's looking to consolidate last weekend's improved showing (whatever that oft-quoted sporting reference really means).

There's some pluck in them there predictions, with Melbourne, the Roos, Hawthorn and St Kilda all confidently predicted to cover their respective spreads. Pluck comes, as ever, stapled to risk and reward, however, so HELP faces a roller-coaster weekend of probability scoring.

MAFL 2010 : Round 7 Results (Final)

Well at least we weren't left hanging until the end.

There are games that you witness where history's page is rendered inaccurate by the word "loss". When teams play as the Saints did tonight another column is required on the competition ladder, to sit alongside "W" and "L", let's call it "F", for "Failed to Compete" (although other designations also spring to mind).

About the only positive I can extract from the game is that, being played on a Monday, it's just four days distant from the next, hopefully memory-erasing wagering opportunity.

Here's what it all means for Investors:

In the end, despite the disappointment of Monday night, Shadow - and hence the Heuristic-Based Fund - made a small profit for the round, so Shadow will retain control of the Heuristic-Based Fund for at least another three rounds.

Here are the full details of the weekend's wagering:

On tipping, just five favourites were again victorious this week, which leaves BKB now an astonishing 4 games off the pace being set by Short Term Memory I on 41 from 56 (73%). Second and third spots are now filled by the Easily Impressed duo, both on 40 (71%), who've demonstrated, along with Short Term Memory I, that, for this season at least, about all you need take notice of is what happened last weekend.

Shadow and Silhouette are, however, lurking near the top and appear set to add another year of consistent tipping to their already-impressive historical record.

(I note that, at this point in the season, every one of the "unsophisticated" non margin-tipping heuristics is in the black on level-stake wagering.)

MARS ratings saw an extraordinary number of position swaps this week, including the replacement of the Saints by the Pies in the top 2, single-rung ladder climbs by Carlton, Fremantle, Adelaide, Port Adelaide and Melbourne, and falls by Sydney, the Lions, Hawthorn and the Roos.

Richmond, at 955.4 ratings points, are now just half a dozen or so solid performances away from surpassing Melbourne's end-of-2008 rating of 944.7.

On margin predicting, BKB is still the only margin-tipper with a sub-30 mean APE (though it was also the only margin-tipper to record an increase in mean APE this weekend). As well, BKB and HAMP remain as the only margin-tippers with sub-30 median APEs.

Overall, BKB still leads on both mean and median APE, while LAMP remains second on mean MAPE and HAMP remains second on median APE.

HELP performed a little more creditably this weekend, correctly predicting 5 of the 8 line betting results and providing probability assessments sufficient to drag its season-long probability scoring a little closer to naivety.

Despite two of the teams we'd wagered on this weekend being rundown in the final term, this has not generally been a season for final term reversals. Only 10 of 56 (18%) teams have won after trailing at three-quarter time, and none of them trailed by 3 goals or more at the final change.

Further, only 18 of 56 (32%) teams have won after trailing at half-time, and just 17 of 55 (31%) have won after trailing at the first change (in the other game the teams were tied at quarter-time).

MAFL 2010 : Round 7 - Results So Far

The round's not yet over as most Investors still have a pecuniary interest in Monday night's Saints v Carlton clash, so this'll just be a short blog to update Investors.

Adelaide's final-term blitz on Sunday evening was welcome relief - apparently there's rarely any other sort if we're to believe TV news anchors - for all Investors, with the profit from the Crows' 50-point victory leaving those with the Recommended Portfolio up 0.64% on the round and 3.24% on the season. MIN#002 and MIN#017 also benefited. MIN#002's portfolio is up 2.25% on the round and now 0.05% on the season, and MIN#017's is down 4.43% on the round and 3.87% on the season.

MIN#002 is the only Investor with no stake in the St Kilda v Carlton game, so he at least can watch the game without the mood-altering effects of a wager. There'll be no such enjoyment though for those with the Recommended Portfolio or for MIN#017 as they all have significant stakes in the outcome.

If St Kilda wins and covers the 15.5 point start they're giving, the Recommended Portfolio will rise by a further 1.63%. If they win but fail to cover the spread then the increase is only 0.68%. And, finally, if the Saints lose, then the Recommended Portfolio will drop by 3.13%, virtually eliminating the season's profit to date. For MIN#017 all that matters is whether the Saints win or lose. A win will trigger a 4.66% increase in portfolio wealth and a loss will inflict a 10.36%-sized hole.

(By the way, I discovered a glitch in the Ready Reckoner calculations over the weekend, so the return for the Recommended Portfolio provided in this blog is slightly higher than what you'd get if you added up the relevant numbers from Friday's table. Errors are always so much more forgivable when they're beneficial, aren't they.)

MAFL 2010 : Round 7

I confess I didn't realise quite how active the six MAFL Funds would be this year once they were simultaneously let loose on the market. This week, their innate tendency to wager has been cunningly exploited by TAB Sportsbet in establishing seven of the home teams as favourites.

So, we have seven wagers from New Heritage (totalling almost 73% of the Fund), six from Prudence (for 25%), five from ELO-Based line (also for 25%), three from Hope (for about 10%), and two each from the currently joined-at-the hip Heuristic-Based and Shadow Funds (for 10% in both cases). In total, Investors with the Recommended Portfolio have about 27.5% of their total funds at risk.

The riskiest of the head-to-head bets is Hope's wager of 5.91% on the Dees, this week's only home team underdog. Other head-to-head bets include a slew on the Pies at $1.20 taking on the Roos, and significant amounts on both the Cats and the Crows, both at $1.25.

In four of the five line bets made by ELO-line we're receiving starts. In the fifth we're taking Melbourne with 24.5 points start.

Here's the detail:

For those with the Recommended Portfolio, the Dees v Dogs game represents easily the greatest upside, and the Crows v Tigers game represents (narrowly) the greatest downside, but also the largest difference between upside and downside.

Herewith the week's Ready Reckoner:

It's an attention-grabbing weekend indeed for a Recommended Portfolio-holding Investor when the smallest swing between the best and worst results for any of the eight games is over 2% of his or her total funds.

This week there's broad agreement amongst the tipsters for four of the games, but considerable disparity amongst them for the other four:

  • The Dogs are 14-1 favourites over the Dees, with the Home Sweet Home the lone dissenter. Chi, who tips the Dogs by 4 points, has this as his Game of the Round, and HAMP, who tips them by 10 points, has this as one of his two Games of the Round.
  • Port Adelaide are 8-7 favourites over the Dons, though the margin-tippers are all on the Dons. Our top 5 tipsters, however, are all on Port. BKB, who I expect will offer the Dons at -6.5 points, will have this as one of its two Games of the Round.
  • West Coast are 8-7 favourites over Hawthorn, and the margin-tippers have again voted as a bloc, in this game on the Eagles. This game would appear to be the uber Game of the Round as it represents ELO's and LAMP's Game of the Round and BKB's and HAMP's second Games of the Round.
  • Collingwood are unanimous favourites over the Roos. Tipped margins range from 26.5 to 35 points..
  • Fremantle are 9-6 favourites over the Lions. Again we have all margin-tippers selecting the same team, which here is Brisbane, and again we have all 5 top tipsters opting for the other team.
  • Geelong are 8-7 favourites over the Swans. The margin-tippers all select the Cats, with predicted victory margins spanning the range from 18 to 28 points.
  • Adelaide are unanimous favourites over the hapless Tigers. Predicted margins range from 24.5 to 33 points, which is about the deficit that the Tigers faced shortly after the centre-bounce last weekend.
  • St Kilda are 12-3 favourites over the Blues, though the tipsters supporting the Blues are each tipping at around 70%, so the Blues are not without a chance.

This week the starts on offer suggest that the matchups aren't as even as last week's - the Cats v Tigers game aside - but we can still look forward this weekend to a round where four of the games have starts of fewer than 16 points, and the remainder all have starts of fewer than 27 points. (Fewer or less? I always find this hard to decide for line betting. Points are undoubtedly countable, so that's one vote for 'fewer', but the bookies offer half point starts, so does that make points a mass noun?)

Anyway, to the HELP predictions.

Again nothing brave this weekend from HELP, especially in relation to the Lions and Saints tips, the probabilities for which I've had to show with a decimal to differentiate them from sheer naivety.

MAFL 2010 : Round 6 Results

Regression to the mean is a bit like unprompted advice from a friend: not always welcome.

But a weekend of regression it ultimately was for Investors with the Recommended Portfolio, as their last-chance Eagles failed to capitalise on their early-game form and eventually went down to the Dockers by over six goals.

With MIN#002 losing also, the only Investor to show a profit on the weekend was MIN#017 who made a 0.55% gain.

Here are the details for all Investors:

Overall, Investors with the Recommended Portfolio profited from four games and made losses in the other four. The most disappointing loss, I felt, was the one on the Dogs, who looked set to take the competition points before a scrappy final term in which they kicked 0.2 to the Saints' 3.4 saw them fall just 3 points short.

Geelong, mercifully, fully justified their $1.01 price by obliterating the Tigers at Kardinia. So comprehensive was their victory, in fact, that they broke the scoring worm on the www.afl.com.au site:

(Apart from the Cats' escaping worm-orbit in that screen capture above, there's also something going on with the axis scaling, which seems to suggest that the Cats are in front by more than 150 points when their lead is actually 'only' 108. I think this is an example of what programmers would refer to as an 'edge case', basically a rare but pernicious eventuality that a programmer must recognise and code for.)

Here are the full details of the weekend's wagering:

On tipping, just five favourites were victorious, leaving a little room for other tipsters to outperform BKB, which Short Term Memory I, Silhouette and Consult The Ladder duly did in each bagging six.

That leaves Short Term Memory I in the outright lead on 35 (73%), and four tipsters - Silhouette, Consult The Ladder, Easily Impressed I and Follow The Streak - just one tip behind on 34 (71%).

It's a bizarre season indeed when members of the Easily Impressed and Short Term Memory heuristic families are vying for tipping supremacy.

Chi continues to deliver the sort of tipping performances that leave me glad that I retired the Chi Fund.

This week the Heuristic Fund dropped 2.2%, so Shadow is now just two further weeks of losses away from losing control of this Fund which, as things currently stand, would be handed over to Short Term Memory I. That'd be fun - having our money entrusted to an algorithm that is the statistical equivalent of a goldfish.

On MARS ratings, Geelong did enough, and the Saints didn't do quite enough, such that positions 1 and 2 have been reversed, with the Cats taking a 1.6 rating point lead. The only other changes were amongst the teams in the bottom third of the ratings ladder, the most dramatic of which were the Dons' rise from 15th to 12th, and the Dees' fall from 11th to 14th.

On margin predicting, BKB is now the only margin-tipper with a sub-30 mean APE, and is one of only two margin-tippers - HAMP being the other - to have a sub-30 median APE.

Consequently, BKB leads on both mean and median APE, with LAMP second on mean MAPE and HAMP second on median APE.

At this point in the season I'll confess to being surprised at the strong performances of HAMP and LAMP but, each being based solely on bookie prices, I guess they should be doing reasonably well given the even stronger performance of BKB.

I'm far less surprised with the performance of HELP, which managed just three correct line tips this weekend and essentially stood still in terms of the three probability score metrics. Richmond's failure to win on line betting despite a ten-goal start was particularly damaging to HELP's probability scores as it had attached an 80% probability to Richmond's preventing such an embarrassment.

The Eagles' poor showing in the final term that doomed many Investors to an unprofitable weekend was their sixth losing final term of the season. They now have an appalling 44.3 percentage in final terms but, amazingly, this is only the second-worst performance in the competition, eclipsed by the Crows' 40.7 percentage.

Only two teams now have a term that they've failed to lose all season. They are the Swans and the Dogs, who've each won all six of their second terms.

 

MAFL 2010 : Round 6

It's as if someone flicked a switch, which in a fashion someone did.

All six Funds have taken up their option to participate in the weekend's wagering, many of them with an alacrity that can only be exhibited by something inanimate that has no sense of the terror that can be induced by a large wager on a team at $1.01.

Hope is the only Fund to show any restraint, chancing just a single wager - its first for the season - of around 1.6% on Sydney at $1.55 facing the Lions at the SCG. New Heritage has waded in like an adrenalin junkie coming home from an enforced 6-week stay at a Buddhist retreat. It's found 6 wagers to its liking, these totalling almost 41% of the Fund and including a 13.5% tilt at the Cats paying $1.01, an almost equally futile 0.1% wager on the Pies at $1.55, and a 3.1% bet on the Crows at $2.35.

The ELO-Line Fund has been almost as profligate, dropping 30% on 6 wagers (one of which - 5% on the Roos - I've yet to make as the market isn't currently available). Mercifully, it's avoided taking the Cats giving 60.5 (yes, over 10 goals) start to the Tigers.

Prudence, Shadow (and hence the Heuristic-Based Fund) have all outlayed about 15% of their respective Funds, in Prudence's case on four wagers, and in Shadow's and the Heuristic-Based Fund's, on three.

Here's the detail (in a slightly larger font this week):

Unsurprisingly, this cacophany of wagers puts a considerable chunk of total Funds in play this weekend, including those of MIN#002 and MIN#017, whose money is at risk for the first time this season. Welcome.

For those with the Recommended Portfolio, the Sydney v Brisbane game represents the largest gap between exaltation and despair, the difference being almost 7.5% of total Funds. The Geelong v Richmond game is next largest, but in that game the upside doesn't justify the label "exaltation" and the downside is the poster-child for "despair".

The remaining details of what result mean what profit and loss are in the Ready Reckoner, which appears below:

As it's been a while since you've had to worry about not just which team wins but also by how much they win, I'll just briefly go through how the Ready Reckoner works.

Consider the second row of the Recommended Portfolio part. It shows that if the Roos win, Investors' net MAFL worth increases by 1.64% (in this case the head-to-head and the line bets win).

If, instead the Roos lose head-to-head but Port fails to cover the spread, then the figure in the third column, representing the case where the home team (the Roos) win head-to-head but lose on handicap, pertains. It shows that Investors will lose 0.90% in that instance, the result of the head-to-head bets losing but the handicap bet winning.

The final column shows the result if the Roos lose head-to-head and also lose on handicap. In that case Investors will be down 1.8% for the game.

Next we look at the tipsters, who this week agree more often than not about the likely outcomes of each contest.

  • The Dogs are 13-2 favourites over the Saints, Consult The Ladder and Silhouette being the only holdouts.
  • Melbourne are 11-4 favourites over the Roos, though 3 of the 5 margin-tippers are on the Roos. All of the margin predictions are for a victory by either side of 9 points or fewer and this is one of BKB's two Games of the Round.
  • Port Adelaide are 11-4 favourites over the Crows, and again 3 of the 5 margin-tippers are on the less-favoured team. Both BKB and LAMP have this game as one of their two Games of the Round.
  • Hawthorn are 10-5 favourites over the Dons. Yet again the margin-tippers are split 3-2 in favour of the less-favoured team. ELO predicts that the Dons will win by just a point and therefore has this as its Game of the Round.
  • Sydney are unanimous favourites over the Lions. Predicted margins range from 12.5 points to 17 points, so no tipster is foreseeing a blowout.
  • Geelong are unanimous favourites over the Tigers. In this game, though, a blowout is seen as a distinct possibility. The bookies are giving the Tigers 60.5 start, which is one of the largest starts amongst my records. The only larger start was the 75.5 points that the Cats gave the Dees in Round 3 of 2008 (which, by the way, the Dees covered easily, going down by just 30 points).
  • Collingwood are 11-4 favourites over the Blues. Once again the margin-tippers are 3-2 to the underdogs. Chi has the Blues winning by just 4 points and so has this as his Game of the Round. LAMP has the Pies winning by 1 point and has this as its second Game of the Round.
  • Fremantle are 12-3 favourites over the Eagles. HAMP tips the Eagles to win by 1 point and has this as its Game of the Round.

The plethora of Games of the Round is a reflection of the overall narrowness of expected victory margins. Apart from the all-feline Cats v Tigers matchup where the Cats seem destined for a percentage-building massacre, no line market is offering a start greater than 15.5 points. That should make for a cardiovascularly taxing weekend.

Finally, to the HELP predictions.

HELP's predictions this week are just a little more risky than last week's. Two of the predictions are only barely non-naive - the first two - and five more have associated probabilities ranging from 58% to 63%. Only the 80% prediction for the Tigers to lose by less than 60.5 points will have any significant effect on HELP's average probability scores.

MAFL 2010 : Round 5 Results

That was a weekend that could easily the unwary into thinking there's something in this statistical modelling stuff.

Six wins from seven bets, with two of the wins coming from seriously unfancied teams, lifted the Heuristic Fund by almost 30% and left it up almost 41% on the season. Investors with the Recommended Portfolio are now, therefore, up over 4% on the season.

The only bet we lost was the one on the Hawks, who just couldn't quite do enough and went down by two straight goals to the Roos. In the only game in which we didn't have a starter, Port Adelaide, who were $3.75 shots mid-week and who were playing at home and so fair-game for the Heuristic Fund or the Hope Fund, outlasted the Saints at Footy Park.

But I'll take what we landed and just smile at what we might have missed.

Here are the details of the weekend's wagering:

Home Sweet Home and Short Term Memory I recorded the weekend's best tipping performances, each bagging 7 from 8, and five other tipsters, the Heuristic Fund's current guardian, Shadow, among them, scored 6.

That leaves three tipsters as joint-leaders of our tipping competition, Easily Impressed I, Short Term Memory I and Follow The Streak, all on 29 from 40 (72.5%). With only four favourites collection the competition points this round, BKB had another mediocre return and now sits two tips off the pace on 27 from 40.

Short Term Memory I now leads the race to be the next guardian of the Heuristic Fund, though with Shadow winning again this week it'll be at least Round 9 before that can happen.

With Round 6 upon us, Shadow will start trading as a Fund in its own right next weekend, as will New Heritage, Prudence and the ELO-Line Funds. So, Investors with the Recommended Portfolio will be doubly exposed to Shadow's prowess for the next 3 rounds at least - once via the Heuristic Fund and again via the Shadow Fund.

On MARS ratings, losses by St Kilda and Geelong cost both teams ratings points but left them in positions 1 and 2, albeit with a smaller lead over the pursuing duo of the Pies and the Dogs. Sydney and Carlton both pushed up a ranking spot, overtaking a plummeting Brisbane in the process.

Fremantle round out the top eight, and have now almost a 5-point lead over the ninth-placed Hawks. Melbourne are the weekend's biggest improvers. They leapt four spots into eleventh, a just reward for their 50-point victory over the Lions.

Margin prediction proved hazardous again this weekend. HAMP produced the round's best MAPE (32.4), a fraction ahead of ELO (33.3) and Chi (33.5). BKB, in a rare show of weakness, came last amongst our margin-tipsters with a MAPE of 35.6.

However, across the entire 5 rounds, BKB still leads both on mean and median APE. HAMP is now 2nd on both measures, with LAMP is 3rd on median APE and ELO 3rd on mean APE.

HELP scored marginally better than a naive tipster this week, correctly selecting 5 of the 8 line winners to take its aggregate record to 21-19, and accruing probability scores a fraction higher than would have been accrued by a tipster assigning a 50% probability to every tip.

Its season-long performance is still sub-naive.

This week the 16 teams combined to register as many goals as they did behinds, 187 of each, which makes the season's conversion rate now just 50.3%. This is, as those of you who've read my three earlier blogs on Modelling AFL Scoring (Part I, Part II, and Part III will know, significantly below the long-term average of 53.64%.

Adelaide and Essendon are doing most to drag the average conversion rate down. The Crows have kicked 41.59 to be at 41%, and the Dons 54.67 to be at 44.6%. St Kilda, with 70.40 (63.6%), and the Swans, with 78.64 (54.9%), are doing most to raise the level.

Round 6 is next, and that's when all the Funds are in play.

MAFL 2010 : Round 5

Well I did expect the level of wagering activity to ratchet up this weekend compared to last, but I thought Hope would be contributing to at least some of that increase.

That's not the case though. Hope has reviewed the market offerings and decided to return to slumber for at least another week; Shadow has looked at the same market and taken out its metaphorical knife and fork, rediscovering its Round 3-style appetite for the punt.

While Hope has passed on all eight contests, Shadow has plunged on the home team in seven, rating only Port Adelaide as home team no-hopers for the weekend. Here's the detail:

So, for the first time this season - possibly ever - Investors find themselves with an interest in the outcome of games across four consecutive days. Shadow's wagers across these four days traverse the landscape from a why-bother bet on Fremantle at $1.03 facing Richmond to a why-the-heck bet on Carlton at $4.60 taking on the Cats.

In between there are four more bets at should-win prices ranging from $1.10 to $1.28, and one more at an probably-won't-win price of $3.75. If you're looking for words to characterise the weekend's portfolio of wagers then, "diverse", "interesting" and "bold" would be among the more positive, while "motley", "perplexing" and "unfathomable" would be among the less favourable and, arguably, more accurate.

With level-stake betting, which is the Heuristic Fund's wagering strategy of record, return and risk correlate directly with market price, so Investors' fortunes are largely in the hands of the high-priced Dees and Blues this weekend. Here are the Ready Reckoner details:

Again this week with the bakers' half-dozen wagers, Investors might find a graphic a helpful way to understand what's required to return a profit.

Some conclusions that I draw from this graphic are that:

  • If Melbourne and Carlton win, a profit is assured
  • If neither Melbourne nor Carlton win, a profit is impossible
  • If Melbourne doesn't win on Saturday, a profit requires at least one win on Sunday and on Monday. On Sunday, a Collingwood or a Hawthorn win will be sufficient if Sydney's won on Saturday and the Dogs have lost on Friday, and a Fremantle win will be sufficient if both the Dogs and Sydney have won in their respective clashes.
  • If we've no winners at the end of Saturday, only Collingwood, Hawthorn and Carlton wins can save us.

Next a look at our tipsters, which shows that we've an unusually high level of agreement this round.

  • The Dogs are unanimous favourites in their contest with the Crows on Friday. Margin predictions range from 17 to 41.5 points.
  • Sydney are unanimous favourites taking on the Eagles, and in this game margin predictions range from 22 to 35 points, again foreshadowing a fairly substantial victory margin.
  • Melbourne are 8-7 favourites over the Lions in yet another game that sees the margin tipsters on one team, in this base Brisbane, and the non-margin tipsters on the other, the only exception being Consult The Ladder who's also tipping the Lions.
  • St Kilda are 14-1 favourites over Port Adelaide, despite Port's home ground advantage, a fact that has been deemed sufficient to warrant tipping Port only be Home Sweet Home. HAMP tips St Kilda to win by just 5 points and LAMP tip them to win by 'just' 23 points, which makes this game HAMP's and LAMP's Game of the Round (to the extent that it makes sense to label a game you're predicting as a four-goal win to the favourites as sufficiently close to justify the GoTR designation).
  • Collingwood are unanimous favourites over the Dons, with tipped margins for the Pies ranging from 18 to 28 points.
  • Hawthorn are 13-2 favourites over the Roos. Only Easily Impressed I and Short-Term Memory I are on the Roos. Predictions of Hawthorn's victory margin start at 15 points and run as high as 26 points, with BKB's prediction of a 21.5-point victory making this its Game of the Round - again only meaningful if you consider a 3-and-a-half-goal victory as "close".
  • Fremantle are unanimous favourites over Richmond. Predicted victory margins run from a low of 23 points to a high of 48.5 points. This game, which is being played at Subi, could get very ugly for the Tigers, and a sporting declaration at three-quarter or even half time by the Dockers, should not be discounted.
  • Geelong are 9-6 favourites over the Blues. Three of the five margin tipsters, all of which have selected the Cats to win, have done so by 7 points or fewer. That makes this game Chi's and ELO's Game of the Round. The remaining margin tipsters - LAMP and BKB - predict much more comfortable wins by the Cats of the order of 4.5 goals or so.

The high levels of agreement in this week's tips mean that there'll be little opportunity for significant changes to our tipster rankings by weekend's end. For example, the two leading tipsters - Consult The Ladder and Follow The Streak - differ only on one tip. It's for the Melbourne v Brisbane game, and Follow The Streak's on Melbourne, while Consult The Ladder's on the Lions.

Lastly, let's have a look at HELP's line betting tips for the week.

HELP has again been very conservative this week. No prediction has been made with an associated probability greater than 58%, which means that even the most disastrous round imaginable won't do too much damage to HELP's various probability scores. It also means that a perfect round won't repair much of previous weeks' damage either. Such is the price of timidity, the virtue of which can only be determined with certainty in retrospect.

MAFL 2010 : Round 4 Results

What a pleasant weekend's wagering that was.

On Saturday, Collingwood looked comfortable from the off and eventually annihilated the Hawks by 64 points, and then on Sunday St Kilda teased us long enough to let us feel that we earned our payoff having trailed by a point at quarter time, drawn level by half-time, led by a point at three-quarter time before running out 15-point winners. Two bets, two wins, the Heuristic Fund up another 4.3%, and the Recommended Portfolio now up 1.13% on the season. All of this with the prospect of the Hope Fund commencing trading in the coming round.

Here are the details of the weekend's wagering:

Favourites seemed to collectively recall their job descriptions this weekend, winning in all six games where one existed. In the two games with equal favourites the victors were the teams higher on the ladder, so BKB wound up tipping the card this week. Chi, Follow The Streak, and Consult The Ladder were the next-best tipsters, all bagging 7 for the round.

Those performances put Follow The Streak atop our tipster ladder on 25 from 32 (78%) and place Consult The Ladder in second on 24 from 32 (75%).

Though Investors have much to smile about courtesy of Shadow's tipping, they might have been happier still had the Heuristic Fund been originally entrusted to Follow The Streak, Consult The Ladder, Ride Your Luck or Easily Impressed I. Few investment strategies though are immune from post hoc regret.

On MARS ratings, St Kilda did enough in beating Freo to retain top spot, though the Cats' thumping of Port allowed them to narrow the gap between them and the Saints to just half a point. Collingwood's win and the Dogs' loss saw them swap places in third and fourth. The only other change in the top eight was Fremantle's elevation into it, due not to any improvement in Freo's rating but instead to the large rating declines suffered by Hawthorn and Adelaide, which simultaneously pushed them out of the top 8 and drove their rankings below the 1,000 mark.

Only seven teams remain with a rating above 1,000.

Margin prediction was a perilous occupation this week, as five games finished with victory margins of 40 points or more.

BKB emerged from the weekend with elevated mean and median APEs but still retaining the lowest mean APE and second-lowest median APE. ELO now has low median APE on its own and LAMP, finally being faithful to its name, now has the second-lowest mean APE.

Chi, despite being on the right team in seven of the eight games, had conservative margin predictions and, as a consequence, grievously insulted his mean and median APEs.

HELP's results were hard to distinguish from those of a notional naive tipster this week. It now has a 50% line tipping record for the season and overall sub-naive figures on all three probability scores.

This week I thought I'd finish with a look at what in previous seasons I've referred to as Alternative Premierships. In the table below I've shown the quarter-by-quarter performance of each team so far in season 2010.

St Kilda has performed well in quarters 1, 3 and 4, but has struggled a little in second quarters, winning two, losing two, and recording a sub-100 percentage, making it only the 10th-best team in this quarter.

Melbourne have struggled in the first quarter where they rank 11th, but have been positively diabolical in the second quarter, losing all four and scoring less than one point to their opponents' three. They've come to life in the second-half, however, producing performances that see them ranked 4th on their third-quarters and 1st on their final terms, where they're the only teams yet to lose a final term.

The Dogs have a different cadence to their performances: poor in the 1st and 3rd terms for which they're ranked 13th and 12th respectively, better in the 2nd and 4th terms where they're ranked 1st and 8th respectively. In 2nd terms they've outscored their opponents by almost two to one.

Fremantle's cadence matches the Dogs'. Freo ranks 10th and 13th for quarters 1 and 3, and 5th and 2nd for quarters 2 and 4.

Onto Round 5 then, where Hope begins ...

MAFL 2010 : Round 4

If last weekend was the SSO in the Opera House this weekend is an acapella duo down the local pub.

We've just two bets and they're both priced at under $1.60. One wager is on the Pies, who face Hawthorn on Saturday, and the other is on the Saints, who take on Fremantle on Sunday.

Here are the details:

The paucity of wagering action is reflected in the simplicity of the weekend's Ready Reckoner the simple message of which is that we need both teams to win if we're to record a net profit.

You might have noticed from the MAFL Wagers table above that two of the line markets are yet to be posted on TAB Sportsbet; at this point they'd be redundant anyway since the head-to-head markets for these games are already offering $1.90 for both sides with no start at all. Three other games have line markets with starts of fewer than 12 points, which might help explain why you're finding it hard to settle on your tips this week.

Our own tipsters suffered no such self-doubt or uncertainty in arriving at their selections, which are shown in the table below (now that'd be a fun bit of AI development - a tipping algorithm that's uncertain about its tips, a bit overwhelmed by the data at its disposal and struggling to make a final selection, maybe outputting 'Cats by 6 ... No, wait a minute, I forgot about that game 12 seasons ago ... make that Port by 2 ... oh, but the Cats are at home, so let me just calculate something else ... carry the 3 ... okay make that Cats by 2 ... oh, but hang on, it's played on a Sunday in April isn't it ... actually, can you get back to me? ... give me another hour ... or maybe two ...').

This week I've included for the first time a column showing each tipster's season-long performance (though 24 games is shy of 30, the sample size sometimes suggested as the minimum necessary to purge a sample from the pejorative label of 'small'). Adding this column gave me the first opportunity to review the tipping performances of LAMP and HAMP, which I'm pleased to report are both superior to BKB's 62.5%.

Here's a summary of the tips:

  • Essendon are 9-6 favourites over West Coast, though they're underdogs and not tipped by any of the margin-tipping tipsters. Chi tips the Eagles by just 2 points, making this his Game of the Round.
  • Sydney are 14-1 favourites over the Roos, with Home Sweet Home the only tipster coming down on the side of the Roos. Despite the lopsidedness of the tipping the game appears likely to be a close one since the largest predicted margin is the bookie's and it is only 11.5 points.
  • Carlton are 10-5 favourites over Adelaide, BKB being assigned Consult The Ladder's selection in the absence of a clear favourite in head-to-head prices. This game too looks destined to be a close one since the largest predicted margin is just 8 points. HAMP and LAMP are predicting margins of just 2 and 3 points respectively, making this one of their two Games of the Round.
  • Collingwood are 14-1 favourites over Hawthorn. Easily Impressed I is the shaglike tipster for this game, tipping Hawthorn solely on the basis that it deems their 16-point loss impressive - or, I suppose more correctly, less unimpressive - than the Pies' 28-point loss. If this heuristic was any more one-dimensional it'd be a point (ooh look, maths humour).
  • Brisbane are 8-7 favourites over the Bulldogs, with BKB again being allocated the selection of Consult The Ladder. This is yet another game thought likely to be won narrowly, Chi's tip of an 8 point victory by the Dogs being the largest predicted margin. The narrowness of victory predicted by HAMP and LAMP make this their other Game of the Round selection.
  • Melbourne are 13-2 favourites over Richmond. Whilst ELO now rates Melbourne a better team than Richmond, the numerical magnitude of the Dees' superiority is narrowly insufficient to overcome Richmond's home ground advantage. Thus ELO is one of the two Tiger tippers, but ELO predicts just a 1 point margin making this ELO's Game of the Round. The only other Richmond selection comes from Home Sweet Home, a tipster also swayed by Richmond's home team status.
  • Geelong are 10-5 favourites over Port Adelaide. Amongst the margin-tipsters, Chi's support is the most lukewarm as he's predicted a Cats win by just 8 points.
  • St Kilda are the round's only unanimous choice. They face Fremantle. Chi is, again, the margin-tipster with the lowest predicted margin of victory for the Saints, but even he has them as 10-point favourites.

We finish with the week's line tips from HELP, which this week are far more conservative than they were last week, all of them having attached probabilities in the 59% to 67% range. Probabilities in that range provide little opportunity to rack up large probability scores on any of the measures, but they also preclude the possibility of anything catastrophic.

HELP's selections are, however, all away teams, so they're interesting tips if not particularly brave ones.

A reminder that this is the last round in which there'll be only one Fund active. Next week Hope starts trading and from Round 6 everything's live.

MAFL 2010 : Round 3 Results

Well that was quite a Sunday.

Investors awoke on Sunday morning with the weekend's profitability hinging on an improbable victory by the Dees against the Crows, which then had to be followed by an almost equally unlikely win by the Dockers against the Cats. In the end, Investors got both results, with a victory by the Dogs providing the proverbial marzipan on the sweet, baked, traditionally tea-accompanying confection.

Sunday's three wins more than offset the financial damage inflicted by Friday's and Saturday's two wins and two losses performance, leaving the Heuristic Fund up 11.75% on the weekend and up 7% on the season. That, in turn, left Investors with the Recommended Portfolio up 1.175% on the weekend and up 0.7% for the season, which means that Shadow has done enough to earn at least three more weeks in control of the Heuristic Fund.

Here are the details of the weekend's wagering:

Only four favourites won during the round, which had the unprecedented effect of driving BKB to third-bottom in our tipping race and delivering Follow The Streak and Easily Impressed I into outright leadership on 18 from 24. The strong tipping performance of these two heuristics has been mirrored in the profitability of their home-team only wagering performances, so much so that had Freo not claimed its dramatic victory over the Cats on Sunday evening, Follow The Streak would have had control of the Heuristic Fund for the next three weeks. (Since Follow The Streak and Easily Impressed I would have been equal on profitability the tie-breaking rule would have been invoked. This rule hands control to the tipster with the superior performance across season 2009. That rule puts Follow The Streak ahead of Easily Impressed I.)

So far this season we've not looked at MARS Ratings; it's about time we did.

St Kilda are rated a clear first at this stage of the season having increased their lead over the Cats with their 28-point victory over Collingwood this weekend. Sydney recorded the weekend's largest ladder climb, jumping three spots into sixth, overtaking the Blues, Hawks and Crows in doing so. Essendon registered the week's next-biggest jump, rising two places to eleventh.

Four teams' rankings dropped on the strength of their weekend performances: Adelaide dropped furthest, falling two places to ninth, while Carlton, Port and the Eagles each fell by just one spot.

Overall, that leaves nine teams rated 1,000 plus and Fremantle narrowly rated sub-1,000. Richmond appear to be making an early-season race for the bottom.

All margin predictors recorded solid performances over this round, reducing both their mean average prediction errors and their median average prediction errors so that now all but Chi have sub-30 mean APEs and all have median APEs of 27 or lower.

HAMP and LAMP remain stubbornly misordered in this table, with HAMP narrowly heading LAMP on mean APE, and LAMP heading HAMP on median APE. Chi is in the odd position of having the worst mean APE but the (equal) best median APE, while BKB's performance continues to demonstrate why bookmakers book holidays more often than they do losses.

In contrast to BKB, HELP turned in another uninspiring set of numbers this weekend, correctly tipping just four of eight line winners - a performance that's no better than chance - and registering probability scores less than those that a naive tipster would have produced.

Collingwood's appalling 4.17 on Friday night got me to wondering about recent kicking performances that had been as bad as or worse. Looking at the 20 seasons before this one I could find only one performance as bad as the Pies' and just four that were worse.

Actually, the overall kicking performance this weekend, which saw 190 goals and 228 behinds kicked - a percentage of just 45.5 - seemed remarkably poor. Indeed, a review of the aggregate kicking performances of all teams in home-and-away rounds since 1995 shows that this weekend was the second-worst in that period, surpassed only by the 45.2% performance recorded in Round 14 of 1997.

Just for the heck of it I then had a look at the worst kicking performances, combined for both teams in a single game, all-time and in the last 30 completed seasons.

That Geelong v Melbourne game must have tested the patience of even the hardiest supporter - twenty scoring shots and only one of them good enough to thread the big sticks. The lone six-pointer came in the second quarter, leaving spectators to find whatever joy they could in seeing Geelong kick six behinds to Melbourne's three in the second half. I do hope they took a good book with them.

MAFL 2010 : Round 3

This week the Heuristic Fund will again be guided by the selections of the heuristic tipster named Shadow, a situation that will not persist into next week if Shadow's Round 3 tips are responsible for the Fund's third straight weekly loss.

Were Shadow hell-bent on retaining control of the Heuristic Fund - assuming, for the sake of the narrative, that it were sentient at all - it might have chosen to respond to this situation by wagering on a couple of short-priced favourites, hoping to eke out a small profit and in so doing earn the right to manage the Heuristic Fund for at least another 3 weeks.

As it happens, there's only one markedly short-priced favourite this weekend - Sydney at $1.10 facing Richmond - and Shadow has, indeed, thrown bread at the Swans ... but also at six other teams.

Yes, that's right, Investors have bets in every game but one this weekend, the Roos being the only home team felt unworthy of a flutter by Shadow. Here are the details:

Three weeks ago few could have imagined the Dees priced at just $2.80 taking on one of last year's semi-finalists, but that's the situation we have this week and it's one that's prompted Shadow to lob 5% on the Dees. This is the highest-priced team that Shadow is supporting this week and is one of only two wagers that Shadow has made on underdogs, the other being another 5% on Fremantle at $2.50.

Not surprisingly, this makes the Melbourne v Adelaide and Fremantle v Geelong contests those with the greatest differences between potential profit and potential loss. Here are the Ready Reckoner details for the week:

With so many wagers it's not easy to summarise in words the minimum set of favourable outcomes that will ensure Investors earn a profit this week, so I've opted for a graphic instead.

In the graphic, the ticks in each row are against those games where the outcome must be favourable if we're to make a profit and the blanks represent games where, in the context of the ticks in the same row, we're indifferent to the result. Considered as a whole, each row represents one set of outcomes that will produce a profit, however small, for the Heuristic Fund.

So, for example, since there's a tick in every row against at least one of Friday's or Saturday's games we know that it won't be possible to make a profit on the entire weekend if we're 0 for 4 come Saturday night. If, instead, we clean sweep (swept?) the Friday and Saturday bets, the fact that every row has at least one tick against a Sunday game tells us that we'd still need at least one of the Sunday bets to win for us to finish ahead.

The last row of the graphic tells us that if Port's our only success across Friday and Saturday, we'll need the two underdog bets on Sunday to get home if we're to finish in the black. The eleventh row - the last row in which there's a tick against the St Kilda game - tells us that if the Saints win on Friday but the three Saturday wagers lose, we'll need all three Sunday bets to win if we're to register a profit.

You can generate and analyse a range of other what-if scenarios using this graphic. I feel like there should be a simpler way to describe the sets of favourable outcomes necessary for profitability, but this is a task I've wrestled with for a few years now and still haven't found a way to satisfactorily resolve. Your suggestions are welcomed.

Anyway, time for a look at our tipsters.

In Friday's game there's unanimous support for the Saints to topple the Pies. Amongst the margin-tippers only ELO has the Saints covering the spread and even then only by half a point.

Next, the underdog Eagles have majority support in their clash with the Roos, though four of the five margin-tippers, including the bookies, are selecting a Roos win. HAMP is the lone Eagle-supporter and has them winning by just 2 points, making this one of its two Games of the Round. LAMP's on the Roos (which sounds a bit like the name for a depressing documentary on the Discovery Channel chronicling the night-time culling of one half of our coat of arms) but is predicting just a 4 point win, which also makes this one of its two Games of the Round. Assuming that the line market is eventually posted as Roos -6.5 it'll also be a Game of the Round for the bookies. The sentiment seems to be then that this game will be a close one.

In the Swans v Tigers game, Sydney are the unanimous selection, with comfortable wins predicted by all of the margin-tippers except Chi who's predicting a margin of only a point, making this his Game of the Round. Carlton, facing the Dons at the G on Saturday evening, are another unanimous tip, by margins ranging from one to four goals.

The Port v Lions game is the other game this weekend that seems likely to be a thriller according to our margin-tippers. ELO, HAMP and LAMP - and probably the bookies, again assuming the line market is posted with 6.5 points start to one team or the other - have this game as a Game of the Round, each predicting margins of 4 points or fewer. Majority tipping support is with the home team as it is in all but one game this week.

Melbourne take on Adelaide at the G on Sunday and enjoy unanimous support from all the non margin-tippers. Adelaide, on the other hand, has unanimous support from all margin-tippers and is predicted by them to win by between 1.5 and 3 goals.

In the Sunday game at Docklands where the Dogs face the Hawks, Chi and Consult the Ladder are the only tipsters opting for an underdog Hawks victory, and in the weekend's final game, which sees Fremantle pitted against the Cats, we again find unanimous non margin-tipper support for one team (Fremantle) and unanimous margin-tipper support for the other (Geelong).

So, in summary, this week's majority tips are: St Kilda (15-0), West Coast (10-5), Sydney (15-0), Carlton (15-0), Port Adelaide (10-5), Melbourne (10-5), Western Bulldogs (13-2) and Fremantle (10-5).

Lastly, let's have a look at HELP's line betting tips for the week.

There are a few high-risk probability forecasts in that lot, most obviously the 97% probability attaching to Richmond's winning on line betting, an outcome that will require the Tigers to lose by 39 points or fewer.

MAFL 2010 : Round 2 Results

If you didn't tip the round this weekend you've only yourself to blame. All eight winners were published in the online (and probably the print) version of Melbourne's Herald Sun late on Wednesday night as one of their forecaster's considered selections.

They were Jennifer Hawkins' tips.

It was that kind of round.

Investors are entitled to think themselves lucky that Collingwood snuck home against the Dees but equally aggrieved at the barnstorming finish from the Cats that prevented a Hawks payoff at $2.50, which would have been sufficient to drag the Heuristic Fund into profitability. In all, the weekend's two losses and one win meant that Investors with the Recommended Portfolio dropped 0.42% of Total Funds, leaving them down 0.48% on the season.

If Shadow records another loss in Round 3 it will surrender control of the Heuristic-Based Fund to some other algorithm (assuming that some other algorithm has a superior return at that point, which is not the case at present).

Here are the details of this round's wagers and results:

With only five favourites grabbing the points and six teams winning that were in a superior ladder position to their opponent, it was a good round for the simpler heuristic tipsters. These tipsters tended to follow Consult the Ladder this week and so bagged six from eight:

ELO went with the favourites in all eight games and so scored only five. Chi's only underdog tip was the Dees and, although vindicated, he wasn't rewarded for his bravery. As a result he scored only four for the week.

That leaves our tipping competition with nine equal leaders, all on 12 from 16, with ELO and BKB one tip back on 11 and then Chi back another two tips on nine. Home Sweet Home trails the pack on just 7 from 16.

The blowout wins of St Kilda and the Dogs, the surprise victories of Sydney and Port, and the narrower-than-expected loss of the Dees all combined to inflate the absolute prediction errors of the margin-tipsters this weekend.

HAMP did best, recording 34.0 points per game for the round, better even than BKB, which recorded 34.4. Next best were LAMP (37.5) and ELO (37.9), with Chi (39.6) responsible for the round's worst MAPE.

Cumulatively, that leaves the situation as shown here:

On mean APE then, BKB leads narrowly from HAMP, whereas on median APE, BKB is tied with Chi. At this point, HAMP leads LAMP on mean APE, while for median APE the situation is reversed. This is, as you might recognise, the opposite of what we'd ultimately expect since LAMP is meant to be low average and HAMP is meant to be low median. Still, it's only early days.

Let's finish again this week by reviewing the performance of HELP.

In short, it wasn't a weekend for the HELP family scrapbook. Just three correct line tips and considerably worse-than-naive probability scores on all three measures.

MAFL 2010 : Round 2

Before I launch into this week's wagering and tipping information, a final reminder: if you want to take part in this year's blog-readers' competition you need to have an entry to me before centre-bounce Thursday night. Details of the competition are in this blog entry.

Entry is free and there are no prizes, which I think has an appealling symmetry to it. Entries can be e-mailed to me at the usual address.

Anyway, onto the footy, which again starts on a Thursday this week but which has no Friday game, it now being traditional for the AFL to avoid games on Good Friday.

The Heuristic Fund is the only Fund active this week and is still under the stewardship of Shadow despite a small loss last weekend. Shadow's opted for just three bets this weekend, two of them on teams giving six or more goals start, making these bets akin to a small investment in the sovereign debt of a generally stable but occasionally startling regime. Even if both of these bets are landed the Heuristic Fund will rise by less than 0.1%, which for many Investors won't even be enough to round them up to the next highest 5 cents.

Our excitement then is all down to the weekend's third and final wager, which is on the underdog Hawks taking on the Cats on Monday and currently paying $2.50 the win.

Here are the weekend wagering details:

For completeness, I'll also provide the Ready Reckoner, dull though it is:

My first thought when I saw the opening markets this week was that it'd been a long time since we'd had two $8.50 shots or longer in the same round, but when I checked I discovered that we'd had Freo at $10 playing Geelong and Melbourne at $11 playing St Kilda in the final home-and-away round of last season. Oh how quickly the memory fades.

This weekend the teams offering the snowball-in-a-warm-place-like odds are Richmond and Melbourne, but the bookies' scorn hasn't been enough to discourage Chi from waving a paw in the direction of the Dees. Perhaps he's making an early and desperate bid to snatch control of the Heuristic Fund should Shadow turn in another two consecutive rounds of losses.

Chi's not alone amongst the heuristic tipsters in plumping for underdog victories this weekend, though he and Home Sweet Home are undoubtedly at the more distant extremities of that particular limb.

Here's a quick summary of the tipping highlights for each game:

  • Carlton (away to Brisbane), by virtue of its thumping win over Richmond in Round 1 and the proclivity of most non margin-tipping heuristics to revert to ladder position when they are otherwise undecided, is the first of the underdog teams with majority tipster support, favoured by nine tipsters but not by any with a memory longer than last week.
  • Collingwood (at home to Melbourne) is favoured by all but one tipster though clearly not by every man and his dog as I have the dissenting dog sitting next to me as I type this. Chi at least has the commonsense to make this his Game of the Round; he's tipping Melbourne by just 2 points.
  • St Kilda (at home to the Roos) is the weekend's only unanimous choice of our tipsters. Amongst the margin-tippers only LAMP has them winning by enough to cover the 37.5 point spread the bookies are insisting they give.
  • Port Adelaide (away to West Coast) is another of the tipster-favoured underdogs and is supported by the same amnesia addled set of tipsters that have opted for Carlton. HAMP, one of the tipsters siding with the bookies, has West Coast winning by 10 points, making this game one of its two Games of the Round. None of the other margin-tippers forecast this to be a blowout either - the largest forecast margin is Chi's 19 points.
  • Sydney (away to Adelaide) is another underdog supported by the same group of tipsters as Port and Carlton. HAMP rates this game as its other Game of the Round and tips Adelaide by 10 points.
  • Fremantle (away to Essendon) is the last of the quartet of teams with identical patterns of tipster support. The margin-tippers foresee this game too as being close; LAMP's predicted margin of 23 points is the largest among them.
  • The Dogs (away to Richmond) have only Home Sweet Home standing between them and a clean sweep of supporters and there's not a lot they could have done bar lobbying to have the game deemed a Dogs home game to have convinced Home Sweet Home to have tipped otherwise.
  • Hawthorn (at home to Geelong) is the fifth and final tipster-favoured underdog for the week. Once more, all the support is coming from the non margin-tipping tipsters. ELO (who tips Geelong by 5) and LAMP (who tips Geelong by 9) both have this as their Game of the Round, and the bookies are only offering Hawks' backers 13.5 points start.

Which leaves just one item for us to cover: the HELP forecasts.

This week HELP has adopted the Home Sweet Home approach to prediction in selecting all eight home teams as line betting winners, though in its case at least for six of the eight games it's enjoying the benefit of points start.

As you can see, HELP is quite confident about each of its selections, attaching probabilities ranging from 77% to 86% to each.

Good tipping, good punting and good fortune.

MAFL 2010 : Round 1 Results

It's hard to make money when you're betting solely on favourites.

The Heuristic-Based Fund landed 3 bets from 4 this weekend but the one it lost was the longest-priced and therefore most vital one from a profitability viewpoint, so it finished down 0.6% on the weekend. This leaves those with the Recommended Portfolio down 0.06% for the season - a mere flesh wound at most.

That means strike one for the Shadow Fund. Two more consecutive losses will see it hand over control of the Heuristic-Based Fund to whichever heuristic is best performed at the end of Round 3.

The fate of the Heuristic Fund notwithstanding, favourites generally fared well this weekend, winning six of eight games though covering the spread in only four.

Here are the details of the round:

(Three of this weekend's games were won by 56 points, which surely must be some sort of record, albeit a boring one.)

On the tipping front there's not much to report. The surfeit of victories by favourites resulted in most tipsters bagging 6 from 8 for the round, the only exceptions being Chi, who managed 5, and Home Sweet Home, who managed just 4.

Next we turn our attention to the performance of our margin-tipping heuristics where we find that BKB has started the season exceptionally well, producing the round's lowest mean and median absolute prediction errors.

Chi fared next best on mean APE, registering 27.63, and LAMP finished 2nd on median APE, recording a very respectable 19.5, its mean APE suffering significantly at the hands of the Dogs' and the Crows' losses.

Finally, let's take a look at the performance of the HELP algorithm this weekend.

It recorded 5 wins from 8 on line betting, which is barely better than chance.

As well as measuring HELP's win-loss record this season, I'm also going to assess how well it assigns probabilities to its predictions by using using what are called probability scoring metrics. There are three such metrics that are commonly used to measure accuracy of a forecaster's probability assessments in light of actual results:

  • The logarithmic score, which assigns a forecaster a score of 2+log(p) where p is the probability that he or she assigned to the winning team.
  • The quadratic score, which assigns a forecaster a score of 1-(1-p)^2 where p is again the probability that he or she assigned to the winning team.
  • The spherical score, which assigns a forecaster a score of p/sqrt(p^2 + (1-p)^2), where p is yet again the probability that he or she assigned to the winning team.

In a future blog I'll talk a bit more about the relative merits of each of these scoring approaches, but for now I'll just note that a naive forecaster, assigning a probability of 50% to every forecast, will return a score of 1 if the logarithmic measure is used, a score of 0.75 if the quadratic measure is used, and a score of about 0.71 if the spherical measure is used. These scores can be considered the minimum acceptable if we're to say that the HELP model is providing any guidance superior to coin-tossing.

Looking then at the scores for HELP in the last three columns of the table above, we can see that it performed across the weekend slightly worse than a naive forecaster would have, regardless of the scoring approach adopted. What hindered HELP this weekend was its assignment of an 87% probability to St Kilda winning on line betting and of a 90% probability to Adelaide doing the same. Its other incorrect forecast was that the Dogs would win on line betting, but it assigned a probability of just 59% to this outcome and so was not as severely punished for this error.

MAFL 2010 : Round 1

We're on.

From a wagering viewpoint, this season has started like most others, with the TAB Sportsbet bookie posting head-to-head odds more than a week before the first game and then leaving them unchanged for over a week. It's as if punters haven't yet realised the season's about to start. Indeed, as I type this, the odds are still what they were Monday a week ago.

MAFL, for its part, has also followed the early season script, wagering lightly, unwilling to risk much until the season's patterns have been established and calibrated. Only one Fund is active and only four bets have been made.

And so the cycle begins again for another season.

This year I've decided to do away with the need for readers of this blog to download PDFs to find out the details on wagers and results, so here instead are this week's wagering details, in-blog as it were:

As foreshadowed on the MAFL Online Twitter feed last week, we've four bets each of 5% of the Heuristic-Based Fund, tying most Investor's fates to the success of the Cats, the Lions, Port and the Dogs. The Cats are the shortest-priced at $1.18 and the Dogs are the longest-priced at $1.60, so there can be no suggestion that the Heuristic-Based Fund has subjected Investors to any substantial risk.

Amongst the remaining Funds, Hope will be released to wager commencing in Round 5 and all other Funds will be allowed off-leash one round later. Since all the Investors who've assembled their own portfolios this year have decided to weight the Heuristic-Based Fund zero, this week's Ready Reckoner is a simple one:

Clearly, no-one's going to be quaffing Moet on the strength of this weekend's results, but nor will anyone be stripping the lounge chairs to find loose change for food.

Investor profitability requires that at least 3 of the 4 wagered upon teams are successful and the Bulldogs must be among them.

Turning next to tips, you'll notice that HAMP and LAMP have been added to last year's cavalcade of heuristic tipsters.

As you can see, it's pretty much wall-to-wall consensus (the most common variety, of course), with almost every tip a nod to the favourite. The exceptions are Home Sweet Home, which has gone all contrarian in half of the games, and Chi, who's tipped the Roos to topple Port by a point, making it his Game of the Round.

HAMP and LAMP agree with Chi's game choice for Game of the Round but not with his tip for the winner. Bravely, ELO has plumped for the Hawks v Dees clash as its Game of the Round, oblivious to the fact that the TAB Sportsbet bookie has installed the Hawks as almost 5-goal favourites. That same bookie - and hence BKB - has opted for the Fremantle v Adelaide game as his Game of the Round (assuming that, when the line market posts, it's with one team or the other receiving 6.5 points start or fewer).

That leaves only one more set of predictions to reveal - those of the new HELP model.

I've decided to include the probability estimates alongside HELP's tips as I hope that tracking the calibration of HELP's tips will provide an interesting subject for discussion during the season ahead. To ensure that I can't be accused of blinding hindsight I'll put it on record now that I don't expect HELP to do well this year; on balance, I think it's probably been overfitted to previous data.

Well that's it for this blog. Remember that the first game of the season starts Thursday, not Friday, night. Investors can enjoy watching this game unfettered by financial considerations. Enjoy.