Matter of Stats

View Original

2014 - Round 1b: A Few Matters of Housekeeping

In last week's post I revealed MatterOfStats' wagers and tips for the entirety of the 1st round, none of which I'll be altering this week in light of the results from the first half of that round. Investors are reminded that they have only SuperMargin wagers to look forward to this weekend, which have as perhaps their only virtue the fact that they generally render all but the last few minutes of a contest largely moot.

While reviewing that earlier post today, I noticed that I'd omitted from it the table of margin predictions - though not the commentary about them - an omission that I have now remedied. 

Seeing the length of the list of Margin Predictors and recognising that the number of Head-to-Head Tipsters was almost equally long, and reflecting that I've added two more Head-to-Head Probability Predictors this year got me to thinking: just what is MatterOfStats' official opinion about any particular game?

For me, the various beauty parades conducted on MatterOfStats to determine supremacy in the different forms of prediction are each of interest in themselves as empirical contests amongst algorithms and techniques of varying complexity. That is, I'll grant you, a fairly esoteric interest. So, I think it's time that MatterOfStats elected spokestipsters.

Accordingly, I have decided to appoint the following Tipsters and Predictors as the MatterOfStats' preferred choices:

  • For Head-to-Head Tipping: C-Marg (using a simple above/below 0 decision rule)
  • For Margin Prediction: C-Marg
  • For Head-to-Head Probability Prediction: C-Prob
  • For Line Probability Prediction: The Line Fund Algorithm (unopposed)

I don't in all honesty think that C-Prob or C-Marg will lead in their respective contests, but it seems too easy to choose Tipsters and Predictors that have as their priors and major influence, TAB Bookmaker opinion. There's no such influence - for better or worse - on C-Prob and C-Marg.

There's one other area in which MatterOfStats presents differing viewpoints and that's in relation to team ratings. MARS has served as MatterOfStats' main team rating system now since its development in 2008, the only significant change to it in that period being a decision to treat as the home team in every Final the team with the higher MARS Rating going into it. MARS has proven to be an accomplished tipster and predictor, and forms an important (alone almost equally as predictive as the TAB Bookmaker and statistically of non-zero importance even after accounting for the TAB Bookmaker's opinion) component of MAFL Fund algorithms.

Earlier this year I developed the VSRS, a very much simpler system for deriving team ratings with the added appeal that the difference in team ratings is designed such that its unit is points scored. As appealing as this system is, I'm sticking with MARS as the MatterOfStats preferred team rating system for this year, though I will be making regular comparisons between MARS and VSRS ratings and implied predictions.

So, that's it. We're all set for the second half of the 1st round, all the tipping, prediction and wagering information is now on display and, should C-Marg and C-Prob prove to be outstanding Predictors and Tipsters this year, at least I'll have anointed them in foresight rather than in hindsight.