Tuesday, October 23, 2007

Player Rating Theory

Bill Roberts commented under my latest Top 30 post about the low ranking for Natalie Gulbis (#25), so I thought this would be a good time to re-address my ranking process and the theory behind it.

Bill is correct when he points out that Natalie has played pretty well over her last five events and that she ranks higher on the money list than some players I have rated ahead of her. My rankings (and the point system that I use to get a framework of how to order the players) are based on a couple of principles: 1) That a player’s production over the entire season be considered equal and not weighted toward recent weeks, and 2) That a player’s production be judged over several categories which are weighted as I see fit. These principles are different from the Rolex and Sagarin methods in that they gather a player’s finishes in each event over the last two years, weigh the recent ones more heavily, and assign a strength-of-field factor to each event. I don’t do those things because frankly, I don’t have the time. I bend Point #1 during the first 20 weeks of the season, since I allow the previous season’s rating to figure in the equation during that time. That is only done to keep some sense of stability in the process – I don’t think anybody believes that Cristie Kerr isn’t one of the Top 20 players in the LPGA but because of her slow start she wouldn’t have been near my April Top 30 without that adjustment.

Natalie’s positives for 2007 are the victory at Evian and her 17th place on the money list. She’s missed three cuts, which is about average for players who rank 15-25. Where she gets hurt is her scoring average isn’t very good (30th) and her Top 10 Percentage (15%, only three out of twenty) is terrible. Those two sub-par stats drag her down a few spots below where Bill feels Natalie ought to be. A good comp for Natalie is Angela Stanford, who I rank at #16. Angela is 19th on the money list and has also missed three cuts. She has not won this year, but where she jumps ahead is scoring average (ninth) and Top 10s (52%, second only to Ochoa). Since Angela has scored consistently better and finishes in contention much more often, I have no problem ranking her higher despite the fact that she hasn’t won while Natalie has.

You might think I should give a player with a victory an extra edge over a player who does not. I do award a point bonus for each win while every other category I consider gets some boost (directly or not) by the victory. Wins give the player a significant boost in money, their scoring average that week was the best of everyone who played, they register a Top 10 finish and they don’t get penalized for missing a cut. When my point system rates two players evenly I do tend to give the edge to the player with a win. Keep in mind there’s a valid argument to do the opposite. One exercise I like to do sometimes is estimate how a player would score if their victory was converted to an also-ran finish, like 25th place, for example. You’d be surprised how hollow several of the Top 30 players’ seasons are if you remove their victory. Another question you have to answer while analyzing this stuff – do you automatically have to put Meaghan Francella, Young Kim and Silvia Cavalleri in your Top 30 purely because they won tournaments this year? Obviously, I answered “no”. I’ve tried to strike a balance between rewarding victories and rewarding good consistent play, and I believe I’ve accomplished that.

It ain’t perfect. A player who plays only half-a-dozen events, like Hee-Won Han this year, has an advantage in that if she plays well in those few events, her points generated by a good scoring average, Top 10 Percentage and few or no MCs are admittedly not as valuable as the same numbers for a player with 25-30 starts. The idle player’s money list standing falls and of course she has fewer chances to win and earn the victory bonus, but those don’t override the other three well enough. My stance in these cases is to place the player where I believe she belongs. Han was a Top 10 player when she went on maternity leave so I have no reason to believe she doesn’t belong at #29 now. Last year, Jin Joo Hong won the only LPGA event she played, but I chose to not rank her since I had no other data to support ranking her. Like Han, Michelle Wie ranked high last year and hasn’t played many events this year but her abysmal performances in those few 2007 events convinced me to drop her off the list.

Hopefully I haven’t rambled too much about this. I’ve given a lot of thought to this process over the last 12+ months, so the multi-threaded musings above reflect that. Feel free to comment on any of it.

1 comment:

Bill Roberts said...

In your latest post, I must say that I feel honored that you singled me out in your Player Rating Theory response. I accept your articulate response about your rating system, I forgot that you are a numbers guy, while I am a complete romantic and feel player as you may have suspected. Keep up the good work HD as I said previously, you are the Hemmingway of Ladies golf.