Re(1): Jai Alai Player Metrics Posted on January 12, 2018 at 02:29:03 AM by StraymarJA
Here is my approach for how I ranked abilities back in early 1990's with actual statistics to use in my Jai-Alai computer simulation. Involved post position history for the players plus their actual win percents.
I will assume a singles game such as Dania game 10 as simple example.
My simulation of 20 million games for 7-point singles with all teams equal and a small singles game server advantage (was ~2 or 3% built-in, can't remember which it used) gave following rounded-off win percents.
To keep example simple I'm just showing 3-games played for Player A, and he had been in Posts 4,5,7.
So, add the simulation %'s for those posts: 12.5+10.9+9.7=33.1
Divide by the 3-games played to get the average Player A simulation win %: 33.1/3=11.0%
You take Player A's actual singles game win % and say it's 13.0%.
Divide actual win % by simulation win %: 13.0/11.0=1.18 factor
Do this for all of the players for game 10 singles.
I also normalized all the factors (took each factor divided by the average of the 8-factors). Can't assume the average of all factors is 1.0 if not always the same group of 8-players. This works good for ranking players.
This is how I generated my inputs to load into my computer simulation to simulate two twin trifecta games for the Sunday mandatory twin-tri at Orlando in the early-mid 1990's. Replies: