2014 - MARS and ChiPS Ratings After R2

Before performing the analysis tonight I thought that team ChiPS Ratings would have been much more affected by the results of Round 2 than team MARS Ratings but, as it turns out, Ratings were similarly affected for both Ratings systems.

View fullsize

For only four teams are ChiPS Rankings more than a single place different from MARS Rankings:

• Collingwood is ranked 6th by ChiPS and 8th by MARS
• Sydney is ranked 10th by ChiPS and 6th by MARS
• Port Adelaide is ranked 7th by ChiPS and 9th by MARS
• St Kilda is ranked 15th by ChiPS and 13th by MARS

Six more teams have ChiPS and MARS Rankings differing by a single spot - Richmond, the Kangaroos, Essendon, West Coast, the Brisbane Lions and the Gold Coast - while the remaining eight teams have identical ChiPS and MARS Rankings. Some of the gaps between teams are more or less assailable in a single round, but the rankings are remarkably consistent.

Lastly, the correlation between ChiPS and MARS Ratings increased a little this week, rising from +0.977 to +0.983. Team Rating Systems, it seems, have much in common.

2014 - Round 1a Results: Well, That Was Nice

Based on the traffic to the MatterOfStats website today, I'm guessing that many regular visitors already know that the first four-ninths of Round 1 of the 2014 season has proven to be profitable for anyone holding the Recommended Portfolio.

The Head-to-Head Fund recorded two successful wagers from two attempts, most notably recognising the now-obvious value that subsisted in the Giants' $13 pre-game price-tag. This Fund is now up by over 9c on the season and is comfortably buttressed against a relatively extended period of ineptitude. The Line Fund also benefited from the Giants' strong showing, its wager with over 8 goals start never seeming to be seriously in danger.

Only the Margin Fund finished the weekend unfulfilled, with both of its Blues'-related wagers coming up short despite promising signs at the end of the first three quarters. It shed 2.5c as a consequence of its poor judgement.

So, all told, the Recommended Portfolio is up by almost 3c after just four games and never have I felt more compelled to remind myself that extrapolation almost always ends in tears.

On the tipping front, Home Sweet Home is enjoying a very rare period of relative accuracy, its 2 from 4 as good as any of the MatterOfStats (MOS) Head-to-Head Tipsters. It too, I suspect, should take heed of the pitfalls in extrapolation.

In other MOS contests, H2H_Unadj_3 leads all-comers amongst the Margin Predictors, although its 42.23 mean absolute prediction error is a long way from elevating it into MOS folklore, and H2H_Adj and H2H_Unadj have, jointly, the best Head-to-Head Probability Predictor performances.

As well, the Line Fund algorithm is enjoying a very rare period of net positive log probability scoring.

Of course, I'd do well to remember that five-ninths of a round is a long time in football ...

MatterOfStats in 2014

Last year at around this same time I said 

Compared to most previous seasons, the number of changes to MAFL Funds and algorithms this year has been very small.

For 2014 I've treated last year as a worthy precedent and made only a handful of changes, none of them significant.

• Last year, the Recommended Portfolio comprised 60% Line Fund, 30% Head-to-Head Fund an 10% Margin Fund. This year the same three Funds will operate - and will employ the same wagering strategies as last year, including the imposition of wager caps that vary across the season - but the weightings will change to the following
  • Line Fund 50%
  • Head-to-Head Fund 30%
  • Margin Fund 20%

I'd love to be able to present an empirical case for these weightings but, alas, none exists. If pressed I'd simply relay my sense that, last year, a Fund carrying only 10% of the total portfolio seemed to be something of an irrelevance, regardless of how active it was.

• This year I wanted to do something in memory of MatterOfStats' (then MAFL's) inaugural mascot, who guilted me out of half a banana and half an apple for the last time late last year, so I've created ChiPS from which spring C-Marg and C-Prob, respectively a new Margin and a new Probability Predictor.

That's it. Everything else will be this year as it was last year including, I hope, our wagering success.

2013 : MARS Ratings After Round 27 (Finals Week 4)

Hawthorn did just enough in the Grand Final to snatch a little less than half a Rating Point (RP) from Fremantle to end the season Ranked 1st with a Rating of 1,052.3 having accumulated 27 RPs across the season, the most of any team.

Fremantle exited the GF Ranked 3rd having netted just under 24 RPs across the season, behind the Cats who finished Ranked 2nd after bolstering their Rating by 25 RPs. The Roos, who accumulated 17 RPs, the 4th-highest of any team, finished 5th on the MARS Ladder, while the Swans, who eventually squirreled only 10.6 RPs, finished Ranked 4th.

At the other end of the Ladder, Melbourne shed most RPs across the season, 50.3, to finish Ranked 17th, and GWS shed 2nd-most, 41 RPs, to finish Ranked 18th.

The final team MARS Rating worms for the season show the Lions, Pies, Dockers, Cats, Hawks, Roos and Tigers preparing to enter season 2014 with Ratings momentum behind them, and the Dons, Giants, Dees, Swans and Eagles with work to do to reverse their Ratings slides.