Tuesday, March 04, 2008

Golfweek/Sagarin Performance Index

So it’s pretty obvious that I have a negative opinion of the Rolex Rankings, but what about Jeff Sagarin’s Performance Index that appears in Golfweek magazine? I took the time recently to scope it out – the explanation of how the Index is compiled is here - and here’s my take.

Because I generally like the system, I’ll start by pointing out the biggest negative. Leaving out the KLPGA is a huge mistake, and omitting the Asian and Australian tours makes the PI even less reliable. I reiterate my opinion that lumping the multiple tours together is a difficult task to perfect because of the small amount of interlocking data (better to rate all the players on each Tour in separate lists, which I am considering doing here at HDLPGA) but Sagarin takes a hass-alfed approach of only adding the LET, JLPGA and Futures Tours to the mix. Just adding the Korean tour would be a big improvement.

I like the method of viewing a player’s posted score as a win, loss or tie versus all the other players in the event. For example, Paula Creamer’s record for the final round at Fields was 67-2-3. She shot a 66 that day, which was beaten by two players and tied by three others. It could be argued that the records should only be compiled at the end of the event rather than for each round, but that’s a minor detail. PI also factors in stroke differential in an undescribed fashion, collects the data for the last 52 weeks and comes up with a rating that professes to represent that player’s typical score relative to all the other players. All of that sounds great and the resulting rankings look somewhat similar to mine, so I’m obviously not going to complain too loudly.

The explainer doesn’t say if recent performances are weighted more, but it appears to do that when comparing week-to-week rankings. The comparisons would be a lot easier if they linked to previous rankings or had a “last week’s rank” column. Looking at one example, Sarah Lee was ranked #8 last week, #10 this week. Dropping her two spots after her T25 performance at HSBC seems a little harsh, if you don’t question her being slotted Top 10 in the first place. Considering ties as a half-win/half-loss, her overall winning percentage is 74.4, versus the Top 10 it’s 36.3, versus the Top 50 it’s 53.6. #11 Angela Stanford beats Sarah in all of these numbers by a substantial margin. The only advantage Lee has over Stanford is in the Strength-of-Schedule column - Sarah has the fifth-toughest where Angela’s is 15th. Perhaps the unknown “stroke differential” factor is in play here – Sarah posted four rounds of 66 or better last year while Angela only posted two. The SOS advantage and the “ability to go low”(?) hardly seem to outweigh Stanford’s edge in winning percentages. In fact, there are a few other players behind Stanford who also have better percentages than Lee along with comparable SOS ratings.

Aside from Sarah, the players in the Top 30 who seem grossly out-of-place to me are Stacy Prammanasudh, Morgan Pressel, Se Ri Pak (all too low), Stanford (slightly high), Juli Inkster, Karrie Webb (both too high) and Yuri Fudoh (a dead horse which I am no longer going to beat). I’m sure Webb gets a boost from her Australian Open win (an LET event) plus her good showing at the HSBC. Pak’s relatively low rating despite her #1 SOS leads me to believe that the stroke-differential piece is what boosts Sarah Lee above those other players. Some detail on how that is factored in would be very interesting (to geeks like me), especially if it included the reasoning behind its extra weight.

All things considered, I like PI as a measuring stick. If it would add the second-best women’s tour in the world and clarify its stroke-differential factor, its value would certainly exceed Rolex’s and could then be used to fill out the limited-field events like Evian and HSBC. Even if those things don’t happen (Rolex’s sponsorship will probably override any other reason to change the status quo), I’ll still regard PI as a useful alternative to my own system.

11 comments:

sag said...

If you only compiled records at the end of an event, it would be impossible to compare stroke differential between a player who played four rounds versus another who missed the cut. That's why it must be done round-by-round.

chemengr said...

I am new to reading your blog. What is your problem with Yuri Fudoh? She won the JLPGA money list for 6 straight years from 2000 thru 2005. She has won 40 JLPGA tournaments with 2 in 2006 and 2 in2007. Her best was 10 in 2003.

Hound Dog said...

Good point Sag but it wouldn't be impossible, just a bit more difficult. My point there was, I don't necessarily care so much what a player scores day-to-day, I care more about how they finish in the event.

Welcome to the site, Chemengr. My problem isn't with Fudoh. I know she has a great record on the JLPGA, thanks for giving me some of the details (since I can't read Japanese, the JLPGA site is really tough for me to find stuff).

The problem is the JLPGA, KLPGA and LET players don't play many events against the LPGA (and vice versa), so the data we have to compare them is relatively small. But that small amount of data is used in the Rolex Rankings and PI to influence the rating of EVERY player on each Tour.

It's not that I think that Fudoh can't possibly be a Top 10 player. I think it's impossible to prove or disprove it without more data. She's played 14 LPGA events in the last three years with two Top 10s and two missed cuts. That doesn't sound like a Top 10 player to me but like I said, I can't disprove it.

chemengr said...

When I analyse the Rolex ratings I think that are much better than you give them credit for being. Rolex ratings show the LPGA to be the dominate tour with the JPLGA to be second and the LET and KLPGA to be weak with others tours also weak. Most of the Japanese stay home on their tour. Whereas the best players from Europe and Korea come play on the LPGA. The ratings say that Korea is producing more better golfers than any other country and that most of them are on the LPGA tour leaving their home tour weak.
The Rolex ratings are for a two year period and therfore give a different picture than your ratings or the Sagarin rating. Karrie Webb rates much higher on the Rolex ratings because of her 5 wins in 2006 on the LPGA and her win earlier this year in Australia which your rating does not include. The Sagarin does include the Autralian win and therefore ranks her higher than yours.
When I look at the Rolex ratings for this week I think it says that there is no statistical difference between the golfers rated from 17 thru 24. The second decimal and third in some cases allow Rolex ranging to put the players in order, but statistically is not significant. I think that if the Rolex rating was done on a yearly basis then their ranking would be much closer to yours and the Sagarin.

Hound Dog said...

Agreed, that Rolex is taking a different picture than me or Sagarin, and that the JLPGA is more "self-contained" than the other tours.

That "self-containment" could be considered part of the problem. Until Ueda entered the LPGA, we had only Miyazato to give us regular info on the relative strength of the JLPGA while we have nearly four dozen KLPGA alums supplying data for that tour. You might be right that the JLPGA is the second-best tour simply because so many Koreans have "defected".

I probably shouldn't give Rolex so much grief. My original purpose behind discussing its system was to get people thinking about it rather than blindly accepting its results. During its first year of use, they made a couple of tweaks to it that showed me they were making a sincere effort to make the system work better. Like I did about a year ago with the TV coverage, I probably need to make a resolution to quit harping on Rolex. Thanks for the comments.

chemengr said...

You can call me cynical or a realist, but the reason the european and Korean golfers come to the LPGA is money. The money is much greater on the LPGA. The top three players on the JLPGA all made very close or over a million dollars (depending on exchange rate fluctuations), so it is not required for Japanese golfers to leave Japan to make significant money. The top player on the Korean tour made less than half a million and second place less than 200,000 dollars.

lijialefw said...

Welcome to our wow Gold and store. We are wow power leveling specilized, professional and reliable website for selling and service. By the wow powerleveling same token,we offer the best WoW service for our long-term and loyal customers You will find the benefits and wow gold value we created different from other sites. As to most people

wow gold said...

If you wow gold were in wow gold any doubt buy wow gold that the buy wow gold God Wars cheap wow gold Dungeon was cheap wow gold challenge enough wow power leveling for you, wow power leveling look no further power leveling than the giant power leveling demon, K’ril wow gold Tsutsaroth. Once buy wow gold thought of as cheap wow gold nothing more world of warcraft gold confirm his existence.

xyue said...

http://www.faysale.com
http://www.mmorpgvip.com

Game player said...

I can getSword of the New World Vischeaply,
Yesterday i boughtSword of the New World Gold for my brother.
i hope him like it. i will give Sword of the New World money to him
as birthday present. i like the cheap snw vis very much.
I usually buy vis and keep it in my store.
i can get silkroad gold cheaply,
Yesterday i bought sro gold for my brother.
i hope him like it. i will give silkroad online gold to him
as birthday present. i like the cheap silkroad gold very much.
I usually buy the silk road gold and keep it in my store.

xuemei said...

I like play online game, I also Buy metin2 gold and Metin2 gold, the Metin2 yang is very cheap, and use the Cheap metin2 yang can buy many things, I like Cheap metin2 gold, thanks, it is very good.

I like play online game, I also Buy Perfect World Gold and Perfect World Gold, the Perfect World Silver is very cheap, and use the Perfect World money can buy many things, I like cheap Perfect World Gold, thanks, it is very good.