Matter of Stats

View Original

2017 - Round 19 Results - Still Close

Five favourites won this week and one drew (after spotting their opponents a 50 point lead - welcome to the 2017 AFL season). With relatively low levels of disagreement amongst the Head-to-Head Predictors, that made for a narrow range of scores (5.5 to 6.5 from 9), which left MoSSBODS_Marg two tips clear at the top of the Leaderboard and now on 102 from 162 (63%).

The all-Tipster average for the round was a creditable 6.2 from 9.

You know the bookies are having a tough year when they're only level with C_Marg on head-to-head tips after 19 rounds.

And, speaking of C_Marg, it was a very rare week of accurate margin predicting from it, so much so that it recorded the best mean absolute error (MAE) for the round of 22.4 points per game, just slightly ahead of MoSSBODS_Marg.

MoSHBODS_Marg recorded a relatively high 23.9 points per game but retained top spot on the Leaderboard in what remains an amazingly close contest. Only 13 points separates 1st from 4th.

The round's worst MAE belonged to Bookie_3 (24.2 points per game) and the all-Predictor average came in at 23.4 points per game.

On Head-to-Head Probability Prediction, the round's best result belonged to C_Prob though it wasn't sufficient to drag it from the foot of the Leaderboard. MoSHBODS_Prob's performance, though slightly inferior to C_Prob's, was enough to leave it atop the Leaderboard and now with a small buffer over Bookie_OE.

WAGERS

A goal relatively late in the Blues v Cats game flipped our unders bet in this game onto the loss side of the ledger and, as it turned out, nudged the entire round's wagering mildly into the red.

That goal meant the Overs/Unders Fund finished with an 0 and 2 record, and its loss couldn't quite be overcome by profits to both the Head-to-Head and Line Funds.

As a result, the Overall Portfolio fell by 0.2c to end the week up by 29.4c on the season (a 17.3% ROI on a 1.70 turn).

Now it's easy to claim that Investors might have been unlucky this week to wind up with a loss since one of the over/under bets missed by less than a goal, but it's equally easy to overlook previous weeks in which the opposite might have been the case. 

So, we should ask: across the season, how 'lucky' have Investors been? Or, more precisely, have Investors tended to win or lose more money in bets decided by small margins?

The table below answers that question and shows the amounts bet and returned grouped according to the number of points by which the bet was decided.

For the head-to-head data, the number of points by which a bet was decided is, simply, the absolute size of the victory margin. We see then, for example, that Investors have been, in aggregate, 'unlucky' in close games, having recorded a -12.3% return in games decided by less than a goal. They've done much better in games decided by 1 to less than 4 goals, however.

In the Line market, the number of points by which a bet was decided is the absolute value of the actual margin plus the handicap. So, a game where Investors were giving 10.5 start that was won by the favourites by 17 points would be categorised as having been decided by 6.5 points. Here, Investors have been slightly luckier in close games, having registered a 1.4% return in line bets decided by less than a goal. They've been unlucky, however, in games decided by 1 to less than 2 goals, and done much better in games decided by 2 goals or more.

For the over/under market, the number of points by which a bet was decided is the absolute value of the actual total less the total set by the bookmaker. So, if the total offered was 172.5 and the final total score was 190, the bet will be assessed as having been decided by 17.5 points. In this market too, Investors have been unlucky in bets decided by less than a goal, but have made profits in every other category.

Overall then, the picture here is quite a good one as it reveals that, generally, returns have been greatest in games where the bets were ultimately decided by 2 goals or more. In that sense, this year's wagering success can't reasonably be attributed to having been lucky in games decided by relatively few points.