Monday, July 26, 2010

The Race for #1 in the World of Women's Golf

This post is a follow-up to yesterday's rundown of where the LPGA's various races--for money-list title, for Player of the Year, and for the Vare Trophy (for lowest scoring average)--stand. And an update of my speculations from 3 months ago on who has the short- and medium-term advantage in the race to be the best in the wide world of women's professional golf. The new Rolex Rankings are out and for the 1st time I can rememember there are 4 players averaging more than 10 points per event:

1. Ji-Yai Shin 10.66
2. Ai Miyazato 10.25
3. Cristie Kerr 10.18
4. Suzann Pettersen 10.14
5. Ya Ni Tseng 8.35
6. Na Yeon Choi 8.22
7. Paula Creamer 7.53
8. Anna Nordqvist 7.23
9. Song-Hee Kim 7.19
10. Karrie Webb 6.74

See the LPGA.com's rundown for more on the rankings and on the new #1 (although they forget in adding up Shin's victories that 1 of them was the LPGA-JLPGA dual-sponsored Mizuno Classic, so while it counts once on each tour, it can't count twice in her worldwide total). Me, I want to make a few quick observations about the Rolex system here.

I'll start with the fact that it's less volatile than Mike Southern predicted it would be--at least at the top of the list, where he focused. He had Shin at #1, but also had Pettersen leapfrogging Miyazato and Kerr, and Choi passing Tseng. I'm not completely sure why his predictions were mostly wrong, even after looking at the FAQ Rolex helpfully provides, but I can speculate on a few factors he may have overlooked and develop some points of broader significant from there.

One key thing you need to understand about the Rolex system is that events from 14 weeks ago and more don't have a fixed point value--they get discounted a little bit each week until they drop out of the system when they become 105 weeks old. So any estimate or prediction has to take into account both the ever-increasing points reduction along with the points cliff that are built into the Rolex system. While it may appear odd that Miyazato, for instance, lost only .02 points to her average from last week, despite the fact that her win at Evian a year ago became of less value and she replaced it with only a T19, it's entirely possible that the value of her win there last season wasn't reduced all that much and that it was partially offset by a bad event from 2 years ago dropping out of the system. And lo and behold, Ai-sama finished T52 in the 2008 Evian Masters. So one lesson from this example is that you have to estimate points to be lost as well as gained when making any predictions.

Next I'd like to highlight an implication of this point, namely, that Ai-sama is in a good position for the rest of the season when it comes to losing points. Sure, she had a great T5 at the '08 Women's British Open, but it's the last of her mere 3 top 10s that she's losing from that season. In fact, she missed 2 cuts right after it and in her last 3 events only managed a peak performance of T40. So what's dropping out of the system over the 2nd half of this season are events that never gave her all that many points to begin with that have been reduced over time to a tiny fraction of their original value (except for the missed cuts, about which more in a second). If she just plays all right the rest of the season, she'll gain far more than she's losing--as opposed to someone like Ji-Yai Shin, for instance, who has 3 LPGA wins from the 2nd half of 2008 that'll be dropping out of the system starting with the 1st next week at the WBO. Even though they're of radically reduced value, she'll still be losing a lot more points than Ai-sama over the rest of the LPGA season. Good thing for her she's been finishing so high so consistently this season, even with her appendectomy. And even better for her that she seems to be fully recovered from it.

Which leads me to another point. The Rolex system values both consistently high finishes and wins, but it values the latter more--and especially so for wins that come in majors. Hound Dog has demonstrated that there are many LPGA non-majors whose fields are as strong or as stronger than majors' fields tend to be, but Rolex persists in granting significant bonuses for wins in majors. So Shin's losing the WBO is going to lose her a lot more points than any of her other wins she'll be losing this year.

As a side note, here's where Suzann Pettersen's lack of wins and majors in '08 helps her ranking in '10. All she has to do is keep doing what she's been doing and regularly put herself in contention, and she'll gain many more points in the 2nd half of this season than she's losing from the 2nd half of that season. Pretty much the same goes for Cristie Kerr, although she does have a win at the '08 Safeway Classic that'll be falling off the Rolex cliff soon. But given that she was limited to 1 win in '08 and 1 win in '09 and put up consistently good and often great finishes at about the same rate those seasons as she has been doing this one--with the difference being in her favor, namely that she's bunched her wins recently and gotten a major that'll stay in the system for a long time to come--she, too, is in a position where it shouldn't be difficult to gain more points than she loses.

But Shin does have an ace in the hole. Unfortunately for her, it's the same one that Ai Miyazato has: the ability to play as many events on the JLPGA as they wish. More on that in a minute. For now, I want to go back to how important wins are in the Rolex system. Check out where everyone stood 2 months ago. I'd be hard-pressed to say who has played better in that span, Song-Hee Kim or Na Yeon Choi. Kim has a pair of silver medals, 3 other top 10s, and a worst finish of T22. Choi has a gold, 3 silvers, and no other finishes in the top 30, including a missed cut (the 1st of her LPGA career). Well, Rolex can tell you: Choi's average has gone up by 2.33 points while Kim's has risen by 1.67 points. Now, you might think from my earlier discussion that that's partly because Choi has lost fewer points than Kim from earlier in the '08 season. But my guess is that Kim actually fared better than Choi when it's come to events dropping off the Rolex cliff lately. That's because when you miss a cut you're guaranteed 0 points. Sure, its value isn't reduced over time, and lots of bad finishes will approach 0 points in the long run, but it helps your average immensely to lose those 0-point finishes--and Kim will lose her 4th (and last) of the year at the end of this week. So you'd think with her greater consistency, she'd have gained more points than Choi, right? Nope. Choi has only 1 more runner-up than Kim in the last 2 months and Kim is getting way more points in her other events. That shows approximately how much wins are worth in the Rolex system--shall we say, a heck of a lot?

Another way to get how much the Rolex system values wins is to compare it to the Golfweek/Sagarin Performance Index. Check out their top 10:

1. Suzann Pettersen 68.65
2. Ji-Yai Shin 68.66
3. Song-Hee Kim 68.86
4. Cristie Kerr 68.88
5. Na Yeon Choi 68.95
6. Ya Ni Tseng 69.00
7. Lorena Ochoa 69.13
8. Ai Miyazato 69.23
9. Angela Stanford 69.43
10. Karrie Webb 69.50

The GSPI is based primarily on how you finish relative to everybody else who played in the same event on the same day as you (W-L-T) and the difference between your score and theirs that day. Unlike the Rolex system, it doesn't include the KLPGA, only covers the last 52 weeks, and doesn't let you look back at past rankings. Fortunately, I've been recording GSPI figures since I started doing my Best of the LPGA ranking, which ranks players into tiers based on where they stand in the RR, GSPI, Hound Dog Top 70, and LPGA money list, then allows me to use my own judgment when ranking players within those tiers. So anyone who's interested can go back and compare how the LPGA's top players today have been doing relative to their competition with how Lorena Ochoa was doing at her prime. My own system accepts that each of the other systems is measuring different things and in fact uses that fact to conclude that players who do well in most or all of the systems must be doing somethinbg better than those who do well in some or none of them. But I'm not here to argue which ranking system is the best. Since everyone outside a very select group of Mostly Harmless and Hound Dog LPGA readers focuses exclusively on the Rolex Rankings, all I want to do right now is underscore how much more wins matter there than in any of the other systems for ranking players.

And I want to use this to return to Shin's and Miyazato's RR advantage, for it's not just any results that matter in the JLPGA events they choose to enter in the 2nd half of 2010. It's mostly wins. Sure, it's nice to play consistently well on the JLPGA, as Sakura Yokomine has over the last several seasons and Inbee Park has done thus far in this one. It's jump-started Park's LPGA career and she's actually passed Yokomine as a result. But Morgan Pressel just playing once on the JLPGA and winning 1 of their majors, the Salonpas Cup--equalling Park's '10 JLPGA win total in 7 starts--when coupled with Pressel's own resurgence on the LPGA, has also leapfrogged her ahead of Yokomine in the Rolex Rankings. Now, for Pressel and Park to keep moving up the RR, they're going to have to outplay those ranked higher than them. But there's nobody ranked higher there right now than Shin and Miyazato. So winning in Japan pays them huge dividents. Sure, every time they don't win in Japan, they're giving more points to whoever does, just because their presence raises the total number of points up for grabs in each event. Still, because the JLPGA's top 30 pretty much plays every event and there are so many JLPGAers in the top 200 of the Rolex Rankings, there are more points at stake in JLPGA events than any other tour in the world, except the LPGA. That's why it's so hard to knock the 6 JLPGA regulars between #16 and #25 in the Rolex Rankings any lower--so long as they keep winning over there. And that's why beating them--as Shin and Miyazato are perfectly capable of doing any given week--is worth something to the #1 and #2 players in the Rolex system.

But how much? And at what cost? They need to pace themselves and save their energy for the LPGA's end-of-season stretch run, when they could play 7 straight events in October and November if they wanted to. My guess is that they're going to play both JLPGA majors in September, when the only conflicting event is the NW Arkansas Championship early in the month. They may not play any other JLPGA events. And they might not even play them. I wouldn't be surprised if we saw both of them in the LET's Ladies Irish Open the week after the WBO, either, as Ai-sama has LET membership via her '09 Evian Masters victory and Shin could get a sponsor exemption at the drop of a hat. But I would be surprised if they played more than 1 other JLPGA event than the majors and the LPGA-JLPGA dual-sponsored Mizuno Classic. Except, that is, the RICOH Cup, the final JLPGA event--and major--of the year, open only to winners and leading money-winners on the JLPGA in 2010. (Right now, Ai-sama still needs to qualify for it, while Shin is already in, due to her JLPGA win.)

In any events they both enter outside the LPGA, Shin and Miyazato will be trying at least to finish ahead of each other and at best to win. For that could well be the difference in the race to finish the 2010 season atop the Rolex Rankings. But the Rolex Rankings are not the be-all and end-all. I'm sure both Shin and Miyazato value the LPGA Hall of Fame more highly than the Rolex Rankings. To get there, they'll need to win majors, Player of the Year titles, and Vare Trophies, in addition to racking up lots of regular wins. So I'm very curious to see how they put together their schedules in the 2nd half of 2010. From their perspective, the point is to play as well as possible for as long as possible, to maximize their odds of a peak performance every day they tee it up in a professional competition. Everything else is a distraction to them. But it's the kind of distraction that from the perspective of golf writers and fans heightens interest in those competitions.

So I'll close this rambling post by asking you: What criteria would you recommend using to identify the best players in the world of women's professional golf? How would you prioritize and combine those criteria, in a quantifiable way? How would you solve the problem of ranking players who compete regularly on different tours?

[Update 1 (4:27 pm): Appreciate the linkage--and thoughts on the Evian Masters and Women's British Open--from Golf Girl!]

[Update 2 (4:31 pm): Just remembered that Shin is the defending champion at the NW Arkansas event. So she'll be there rather than at the JLPGA's 2nd major, the Konica Minolta Cup. My guess is that means Miyazato will skip the JLPGA major, too.]

[Update 3 (7/29/10, 9:46 am): Great overview of the new #1's season to date by Happy Fan.]

[Update 4 (8/6/10, 4:22 am): If you haven't read the comments thread for this post, please do so and then head over to Ruthless Golf for Mike's sequel to/expansion of his comments on the thread.]

4 comments:

Mike said...

This is a cool post, TC! I'm going to link to it in Tuesday's preview post about the WBO. I'm going to need to read it a few more times to see if I can improve my track record.

Just to help you figure out where I went wrong, here's how I currently make my predicti-- er, guesses about the Rolex:

1) I make a guess at how many extra points the winner might get to add on to their existing total. (For example, I estimated that Choi got 45 points for winning the Jamie Farr Classic. I didn't know how many points Rolex would give the Evian, but I doubted it would be as much as the U.S. Open's 95, so I used 45 to be "safe." As it turned out, it was somewhere between the Farr and the U.S. Open.) My method is an inexact science at best, since I know points will be lost in other places.

2) I use this estimate to guess how many points might be lost off the back end. This is mostly a matter of feel, since I have no way of knowing how much a given position is worth in that tournament, especially after two years. But as a rough guide, I would estimate a two-year old win to be worth between a new 5th- to 10th- finish.

3) While I know that tournaments lose some value after 13 weeks, I don't try to guess how much that value will be since I don't know how Rolex figures point values to begin with. I certainly didn't know that they decrease a little more each week; I figured they devalued a bit every few months, maybe quarterly. That would certainly affect my guesses.

Like HoundDog, I've noticed that Rolex seems to care less about strength of field than about the status of the tournament. I give majors way more weight in my guesses because of that. And I go back and compare previous week's rankings using the LPGA's PDF files to see how much movement there has been.

My two main misses were that Pettersen would leapfrog both Ai and Cristie, and that Choi would pass Tseng. The Pettersen miss is pretty easy to explain: I don't know how Rolex decides how many points each position gets, but I expected Ai and Cristie to get fewer points than they did. (As I'll point out in the Tuesday post, I also expected Cristie to lose slightly more than Ai, but Cristie lost more relative to Ai than I expected.

The Choi-Tseng miss is less clear to me. It wasn't that I overestimated Choi's points; she ended up about where I expected. But I expected Tseng to lose points and she actually gained a few fractions. Perhaps you know why, but I didn't think her T11 would be worth as much as it apparently was. I had assumed the gap between 1st and 2nd would be greater than the gap between 2nd and 3rd, which would be bigger than the gap between 3rd and 4th, and so on. Am I wrong on that?

There's absolutely no question that you're right when you say the Rolex is less important than the WGHoF to Ai and Jiyai -- when Jiyai was told after her win that she was #1 again, she had no idea that her win would do that.

I have some thoughts on how to rank players, but I think I've rambled enough for one comment!

BTW, Jiyai won't be hurt by losing any points this week -- she'll be chalking up her 2nd major. ;-)

The Constructivist said...

I'm picking Ai-sama out of loyalty and her WBO track record, but otherwise Shin would be my pick, too! You should play the PakPicker, btw--it would be fun!

I liked your comment and post--thanks for the insight into how you do this.

Here's why I don't want to get too wrapped up in trying to figure out the Rolex system. Most ranking systems have similar strengths and weaknesses. Once you decide how you're going to calculate strength of field, and quantify/weight different criteria, the initial rankings that result then go on to influence all future rankings. All rankings systems get a little circular over time, then, as they use their own current rankings to weight future performances. But at least they each become more and more internally consistent.

That's why I think they're mostly useful for watching trends over time and comparing and contrasting trends across different systems. Really fine differences, as exist between the top 4 in the RR and top 2 in the GSPI, are pretty much insignificant. Thus it becomes a judgment call as to who is actually a better player. The result is my system.

I'd love to hear your thoughts on ranking players when you have more time!

Mike said...

To me, the biggest problem with the ranking systems is lack of agreement. What exactly are we ranking?

Take the PGA rankings as an example. The only thing everybody agrees on is that Tiger is (or at least has been) the best player in the world. Now, with Tiger struggling to get his life back in order, should he still be #1 in the rankings? Do the rankings tell us who are the best players... or just the hottest players? The rankings cover two years, but give extra weight to the last three months... so when did these players really deserve their position? And if you have minimum or maximum divisors, aren't you tinkering with the actual results each player should have?

Without getting into a lot of detail, I'd like to see a few changes to the system, regardless of what tour is being ranked:

1) I'd like to see two different sets of ranking figures -- one that covers two years, and one that covers six months. Call them a consistency list and a hot list. Comparing them would give us a better picture of who is playing consistently vs who is on a run. You could also use the hot lists to determine if a player who looks consistent just had a hot run that skews the long-term figures.

2) Don't reduce point values over time. With two different rankings covering different time periods, we can use the same simple math to keep track of things. If you're hot and stay hot, your half-year and two-year stats match; if you cool off, your two-year stat is higher than your half-year; and if you're cold then heat up, your half-year stat is better than your two-year. Simple math now not only tells your stats, but your trend as well -- something the current complex system can't do with any clarity.

3) I think the whole idea of minimum divisors should be trashed. If you're meeting the minimum playing requirements for your tour, you shouldn't be penalized just because someone else wants to play every week. By the same token, maximum divisors skew the figures the other way. Use the actual number of tournaments a player plays to figure their ranking.

4) Point values for tournaments should be adjusted for strength of field... but don't get stupid about it. Except for majors, which should give more points because they're more important, give all the other tournaments the same number of points but use a simple strength of field adjustment.

Here's an example: Let's say we've settled on 100 points for the winner of non-major events. If any of the Top 5 players show up, give the full 100 points. If the top player is between 6 and 12, reduce points by 5% (so the winner now gets 95). Between 13 and 25, reduce by 10%; between 26 and 50, by 15%; below 51, by 20%. Now everybody knows exactly how many points are available at a given event. And if, say, Phil decides to show up at this week's Greenbrier Classic and everybody else is lower than 51 but one of them beats him, they get full points because they beat one of the big boys. Events then become "upper tier" or "lower tier" based on who shows up, not who sponsors the event.

I think this could also have a beneficial effect on sponsors. Once points can be affected so much by one player, it's in the other players' best interests to spread themselves around -- something the tours often have trouble getting the name players to do -- because just one player who wants to move up in the rankings can show up at a "lower tier" event and turn it into a "higher tier" event, grabbing some extra points if he can win it. Strategy -- how you choose which events you play in -- now becomes a factor in your ability to move up in the rankings. You want to reach the top of BOTH rankings at the same time! This benefits both the lower-ranked players and the sponsors, who won't complain.

That's what I'd like to see. It's easy to understand, easy to figure... and maybe a bit subversive, but I think it would give us a lot more useful info than we get now.

The Constructivist said...

I like these ideas--sound like they're worth a post or 3 at your blog!