Re(2): Handicapping a Doubles Team Ability.....when little or no data is available Posted on February 13, 2020 at 09:46:46 AM by StraymarJA
I see a problem right there with your assumption of "projected win rates"
Success doesn't have to be based solely on win, I think there are flaws with that logic, but then everyone has their own methods of assessments (aka handicap), just some more archaic than others, from just looking at program and gut instinct handicapping to refined assessments used in a simulation (as I'm sure you are familiar with), and pari-mutuels being a contest of trying to out-do the others placing wagers and turn a profit, some have some closely held secrets, contributing to their success, that would surely be foolish to divulge on this site.
You may have found success with using only win rates, but adjustments to those would be needed if using a simulation, as some players would be rated so low they would be long odds on winning the point, thus compromising the integrity of results of simulation output, even some pretty easy tri combos would be much harder than reality as those players do win points and get a fair amount of places and shows. Good example would be the player who wins only 1 in 50 games, now I know you have said in past you have a min value for input and even a max, so not truly based on win rate, right? So maybe what I call a UA curve, a Use Percentage vs Actual Percentage, and it's not a straight line. A curve where the actual calculated assessment percentage is used to read the Use Percentage which is the adjusted assessment value used in the simulation (already starting to feel like I'm divulging too much, lol, but hey, have no kids to pass this on to).
Handicapping is different for everyone, I doubt any two experienced handicappers do it the same (please don't ask me what "experienced" means Craig). I'm a firm believer the win, place and show should get weighting in a player rating if using a simulation based on my experience (not much for show though), and handicapping (or player assessments as I like to call it), IMO is something that should be continually refined (improving the quality of assessments). Being an engineer, I'm very familiar with and had training on all those Japanese quality concepts that almost drove the US completely out of the auto industry many years back. Many companies took their concepts and tried to re-brand it into many different names (Q+, Continuous Improvement, Engineering for Excellence, and many more).
It's obvious doubles are harder to handicap, that's a given, simply as there are more factors to consider.
With my initial posting, I'm throwing out for discussion, when you have basically little or no team data for doubles (as it can take a long time for players to get paired together enough times to feel comfortable with being a large enough sample of team data), but one obviously has MORE doubles game data to assess abilities (probably at least 8x more data than team data), then some estimation method could be derived using the only available data to create a team win rate or whatever you call it. With that in mind, should FC be weighted more than BC and I'm not seeing discussion yet on that, but HOPING to see some posting on that at least.
BRNE, I bet you have some insight into this? Would also like to hear feedback from former players and/or amateurs playing doubles on FC vs BC weighting to estimate a teams ability.