Since Ryan changed the proportions of points going to first place, second, third, etc. at the beginning of February, to increase incentives for going for the win, the zero datum has to be recalculated to fit the TAZD (Test Against Zero Datum) to the new incentives. I have done so, with the following observations.

1. I took the top 1,013 zero-scoring players on the leaderboard. Of these, only 641 had played one or more regular game.

2. Of these 641, I took the player average of percentage profiles with the following result:

03.1 03.3 05.6 13.9 22.1 24.0 26.0

These are likely of not much use since most of the 641 players had played only one or a few games and their zero scores likely reflect only an incidental situation and not their true skill.

3. I then took a game-weighted average of the percentages and came up with

09.4 09.2 10.6 12.7 16.7 19.1 22.3

These do not differ much from the numbers under the old incentives and has the curious result that second place gets a little more emphasis than first. But they do differ in the predicted direction.

4. One factor thay may affect the reliability of the zero datum is the buy-in effect of tournament play. Some of the current occupants of the zero-score part of the leaderboard may be there because of net negative draw in tournaments. Players with negative skill in regular play are more likely to have positive draw. So I decided to do a draw-weighted set of percentages (which is equivalent to point-weighted number of finishes) with this result:

05.7 06.2 07.8 10.3 17.3 22.6 30.1

Which of these three sets of numbers is the best fit for a typical zero-score player in regular play?

I invite your comments.