2019 - Round 3 Results - Better Than We'd Expected

/

Had you, on Thursday morning, offered me a best of six from nine in head-to-head tipping, a 20ish lowest MAE, and only mildly negative log probability scores for the head-to-head probability predictors, I’d almost certainly have accepted fairly rapidly.

MoSHPlay_Marg was one of three Head-to-Head Tipsters bagging that six from nine score (the others being Consult The Ladder and C_Marg), which was enough to leave it atop the Leaderboard on 15 from 27 (56%), one tip clear of Consult The Ladder.

The round’s best MAE of 20.3 points per game belonged to MoSSBODS_Marg, but that wasn’t quite good enough to allow it to haul in MoSHBODS_Marg, who now holds a 2.5 points lead. The irony that, in the first season I introduce a hybrid model with player and team components, we find the Leaderboard topped by the team-only model which provides the team component of that hybrid, is not lost on me.

MoSHPlay_Marg did only slightly worse than MoSHBODS_Marg though, and is now only about 20 points adrift. (It did, by the way, miss the late change in the Melbourne v Essendon game, but that turned out to be a small benefit for it, worth about a point or so.) It now lies in 6th, up a couple of spots since last week.

Probability Scores for the Head-to-Head Probability Predictors were, as already mentioned, entirely negative this week, but only slightly so. Best, yet again was C_Prob, which is why is continues to lead the field. MoSHPlay_Prob is showing some early signs of miscalibration and overconfidence, which might need to be addressed in the offseason. It currently sits second-last on the Leaderboard.

WAGERS

Essentially, a break-even result this weekend for Investors, as profits from the Head-to-Head and Line Funds were almost exactly offset by the losses from a 2 and 4 performance by the Overs/Unders Fund.

At the end of the round, the Combined Portfolio is now up by 3.4% on a 11.8% ROI and a 0.3 turn.

(Note that last week’s Combined Result was slightly misstated, as it used last season’s Fund weightings rather than this season’s. The difference was about 0.2c.)