Re(1): El Barba stat line-weird
Posted on March 27, 2023 at 02:48:03 PM by Craig G
This illustrates a very conspicuous hole in jai-alai theory. By that I mean not addressed in 'Calculated Bets', 'Ballet with Bullets', 'Hall of Fame', QOW archives, or elsewhere.
Magic City makes it easy to see last season stats, so we see that the two most struggling guys win-wise were El Barba and Diaz.
El Barba 345: 20-32-36
Diaz 244: 14-28-38
Now when you think about the 'Standard Model', which would be something along the lines of what Prof Skiena used, and presumably also the computer-based guys hereabouts, the assumption is that the playing skill is somewhat uniform. From day-to-day, game-to-game, and point-to-point.
However, if you imagine players who consistently wins only 45% of their points, they DO NOT pick up sharply in the show department. Or 40%, or whatever. If you do a sim with them playing in each post an equal number of times, you find that they do below average in each of the W-P-S categories. In the case of a player winning 45% of all points, the projected finishing rates would be 11.6 - 9.6 - 8.7 per 100 games. Yes, P & S are are on an ascending pattern, but still below average, which is 1 out of 8 per category.
So we have a huge hcap problem which nobody wants to tackle. Winner types show less than expected and outclassed types tend to show more than expected.
Step 1 here would be to define the problem. Step 2 would be to gather data, maybe season-ending for as many frontons as possible.
If this pattern is real and you can work out the rationale for it, then it shows that the simple-minded computer programs are a little bit lame. Obviously far better than nothing, yet not all they could be since they can't predict reality accurately.