Luke Donald Back to #1

[MEfree]

Your link provides more details, but this was discussed a bit here http://thesandtrap.com/t/56702/owgr-biased-against-pga-tour-players

Originally Posted by Clambake

Nice little article in the latest Golf World magazine about the OWGR and suggested changes to make it more mathematically relevant based on studies by a couple Dartmouth professors:    http://www.golfdigest.com/golf-tours-news/2012-04/gwar-stachura-world-golf-ranking    They've come up with a methodology that better ranks the golfers by comparing how they score when playing the same courses and against each other.

One of the interesting things is a table showing how different players would rank in their methodology compared to the current system (at the end of 2010).    The version below is a little hard to read, but if you look at the article link it is more clear.     I'd like to see this table based on today.....would be interesting.

[bplewis24]

Originally Posted by Zeph

Why is it hilarious that the number one player can keep his spot without playing for a couple of weeks?

You're misinterpreting my amusement as an indictment of how the rankings work, but it's not.  I don't have much problem with how they calculate it, mostly because I haven't taken the time to understand it.  I find it funny because the golfing media/analysts make such a big deal about who is #1 and what it means to their careers, and then it's gone the next week....and then back again.  It sort of dilutes the whole thing and makes people care a lot less about it.  When one person has a strangle-hold on it for months at a time, it seems more significant because then we know who everybody is chasing.

At least, that just seems like a natural way of perceiving it to the average person.

Brandon

[Wansteadimp]

Originally Posted by bplewis24

You're misinterpreting my amusement as an indictment of how the rankings work, but it's not.  I don't have much problem with how they calculate it, mostly because I haven't taken the time to understand it.  I find it funny because the golfing media/analysts make such a big deal about who is #1 and what it means to their careers, and then it's gone the next week....and then back again.  It sort of dilutes the whole thing and makes people care a lot less about it.  When one person has a strangle-hold on it for months at a time, it seems more significant because then we know who everybody is chasing.

At least, that just seems like a natural way of perceiving it to the average person.

Brandon

Not sure we need a ranking system to tell us that Tiger was number 1 and that when he isn't at his best its all very even.

This year Mahan is the only 2 time winner in the US, Oosterhuizen has 2 wins in "Europe" Brandon Grace (who he?) has 3 but in low(ish) quality field events.

Its the nature of golf, its why you get so many different winners, form comes and goes and most pros consider a year where they win a good year and 2 wins is a fantastic year. Donald having 4 wins and all those top 10s last year is the exception (Tiger excluded) and we shouldn't expect him or anyone else to be doing it year in year out.

Any ranking system based on such flaky results is going to give a flaky ranking.

[brocks]

I think you misunderstood how weighted average works because I showed 10 points for all the events.  Here is a better example involving Player A and Player B who both earned a total of 40  points (pre weighting) in 3 events but with A having better results as his 20 point performance came in the most recent event while B`s 20 point performance came in the oldest event.  Notice that A has a higher average in both computations but that the OWGR "Average" is actually less than the lowest event points earned by either player.

I think I understand weighted averages; I also think that they give too much weight to older results. Your latest example shows that both your system and the current system work about the same for results that vary widely in time (under both systems, Player B's average is 65.4% of Player A's), and I don't see anything important about the fact that the current system gives an average below the raw points earned, so there's no reason to switch on that basis. But as I said in my previous post, when the points earned is about the same each week, your system IMO gives way too much weight to older results. Here's an extended example, so our non-mathematical friends can follow along. If I understand your assumptions, you are sticking the the two-year window and the present depreciation scheme (13 weeks at 100%, followed by a reduction of 1/92 per week for the next 91 weeks), but you are changing the divisor to the sum of the weights, with a minimum divisor of 22.5, correct? Then let's say Moe graduated from Q school and started playing every week. He happens to be very very good, and he earns 10 points every week (that's about like finishing 7th at Bay Hill). After 24 weeks, the sum of his event weights is 23.28, which exceeds your minimum divisor for the first time, so he is now getting his full value in the rankings. Since he's earned 10 points each week, and it's a weighted average, he has an average of 10.0. Under the current system, he would be well under the divisor of 40, so he would have an average of 5.82 (232.8/40). But he would still be ranked 5th in the world, which is what happens when you rack up 24 consecutive top tens in strong events, I guess. He keeps it up, and after 40 weeks, he meets the current WGR minimum divisor, and gets full value under the current ranking system. By now his total event weight is 35.89, so his average under the current system is 8.97, and he is tied for third in the world. His average under your system remains at 10. Just by the way, this is the highest he will ever be under the current system, so anybody who thinks the WGR doesn't give a bonus for winning can take comfort in the fact that a guy who has 40 straight top tens in strong events can't get past third (based on current competition). 12 weeks later (still earning 10 points per week), he reaches the maximum divisor under the current system, 52. The total weight of his events is now 43.52, and look what happened, his average under the current system has fallen to 8.37 (435.2/52). This is the problem we both have with the current system --- good results penalize strong players as they age. I've always used winning a major as an extreme example, but this example (top tens in strong events) may be better, because it shows that the anomalies begin before the events are even a year old. So maybe, in addition to shortening the window and depreciating less, they should also do away with the maximum divisor, and just use the same divisor for everybody, regardless of events played. If you play too few, you are penalized, and if you play more than the minimum, only your last 40 count. I need to think about that some more. But back to the scenario -- from now on, as long as Moe keeps earning 10 points a week, nothing changes in either system. His average is always 8.37 under the current system, and it's always 10.0 under yours. OK so far. But after a couple years of this, Moe crashes his clown car into a fire hydrant, and a dozen Colombian prostitutes spill out of it. Moe's wife demands that he give up golf for ten months. Moe stops playing. And here's where the problems with your system crop up. Under the current system, as soon as Moe stops playing, his average starts to drop, because his events are depreciating. He loses around a tenth of a point a week -- a little less at first, a little more later. In six weeks, his average drops below 8, in 16 weeks, it's below 7, in 25 weeks, it's below 6, and in 34 weeks, it's below 5. After 42 weeks, it's right at 4. Well, that makes sense, right? If a guy hasn't played for 10 months, he should expect to lose some ground, right? That's what happened to Tiger after his surgery in 2008, when he took 8 months off, except that when he stopped playing, his average was over 21(!) rather than 8, so even though he lost 12 points in those 8 months, he remained #1. And almost everyone, including me, thought that was wrong. IMO, if I can give him three shots a side (as I could when he couldn't even walk), he shouldn't be #1. After, say, six months of inactivity, his ranking should take a severe hit. But under your system, Moe wouldn't have lost even a tenth of a point. If another player had matched Moe, and earned 10 points a week every week Moe did, and then went on earning top ten finishes every week after Moe stopped playing, he would not pull ahead of Moe in the rankings until 42 weeks had passed. For ten months, Moe could sit on his couch and never touch a club, and his average under your system would still be 10.0. It's not until week 42 that the weight of his events over the last two years finally falls below 22.5, and the minimum divisor finally starts lowering his average. Without that, his average would stay at 10.0 for two full years, even though he never played. So both systems have problems, but I think that golf fans would be less happy with your system than the current one. I realize my scenario of a guy getting top tens every week is unrealistic, but that was just done to make the math easier; it would work the same way, albeit less dramatically, for anyone who took an extended layoff. If one of the secondary goals of the WGR is to encourage players to play more often (and the minimum divisor shows that it is), your system gives reverse incentives --- any time a player has a hangnail, he can take the week off, and it won't lower his average. And every time he plays, since he gets full weight for recent events, he makes up (minimum divisor-wise) for missing two events 14 months ago, or ten events 22 months ago. So I still think that the way to go is to keep the current system, but shorten the window to one year, and not speed up the depreciation to compensate, so that the oldest event in the window is still worth about 57% of its full weight. It would be more volatile, but also more fair.

[brocks]

Not sure we need a ranking system to tell us that Tiger was number 1 and that when he isn't at his best its all very even.

You're right, when Tiger was dominant, we didn't need the rankings to tell us who was #1. But when the Masters was taking the top 50 in the world, or the WGC Matchplay was taking the top 64, we needed the rankings to tell us who was in the top 50, or the top 64. Most people follow only a handful of golfers. Most Americans don't watch Euro Tour events unless their favorite American is playing. And I'm talking about golf enthusiasts. The average viewer, who watches only when Tiger is contention, wouldn't know Charl Schwarztel if he tripped over him. The rankings aren't perfect, and I've proposed improvements myself, but we need something like them to have order in the pro golf world.

[MEfree]

Very good point Brocks!  You are the first person to come up with an example where my proposal might not be equal or superior to the current system.  I do think that it is unrealistic for a pro to play 52 events in a year, and my system saw Tiger drop in the rankings very similar to how he did in the OWGR (because Tiger hand`t even played 52 events in 2 years and ran into my minimum divisor problem much sooner than in your example).

However, extreme examples are useful for showing potential problems, so I think my proposal should be tweaked.  I still believe that it is superior to use a proper weighted average, but think that instead of having a single minimum divisor, that there is (at least) a minimum divisor for the current ___ week period.  This could be done a number of different ways, but one possibility would be to take a players average points for each of the last eight 13 week periods with a 4 or 5 event minimum per period (and a 40 event overall if you used 4 instead of 5 per period).  Then you could take a weighted average of each of the periods.  The weightings could be done so that the overall value for each period is similar to the OWGR depreciation.  If a guy did not meet the minimum in the most recent period(s) his average would take a hit sooner rather than later.

Of course shorter periods (i.e. 6 weeks) could be used if you want to start penalizing a guy for not playing sooner and/or not have as drastic a drop as an event moved from period to period.

On a related note, I think that for year end rankings, they should only look at the last 12 months with no time weighting of events (to avoid something like the Masters from depreciating so that it is not worth less than the PGA (or only a bit more than Tigers 18 man event which is given too many points, but that is another subject) for year end ranking purposes.  For ongoing rankings, I actually think that it makes sense to go back (at least) two years, but give a much reduced weight to older events.  I personally think that it made sense that Tiger stayed in the top 50 for as long as he did- imagine a scenario where tiger had missed two years- I certainly would have taken him over the #200 in the world once he came back and was anywhere near fit.  I also think that it might make sense to delay depreciating the majors for 12 months, but the most obvious problem is their poorly computed average.

As you state, in my example, with identical schedule, my proposal has an proportional average that is the same for both players.  This is true only when they play the same schedule, but the ratio moves in favor of the guy who played fewer older events under the OWGR when they don`t play the same schedule.  THERE WOULD BE NO REASON TO CHANGE THE OWGR AVERAGING METHOD IF EVERYONE PLAYED THE SAME SCHEDULE.  However, as not everyone does, an averaging method that does not penalize a player for having played in older events should be used.  That is why my (tweaked) proposal is better than the current method.

Thanks for your input!

Originally Posted by brocks

I think I understand weighted averages; I also think that they give too much weight to older results. Your latest example shows that both your system and the current system work about the same for results that vary widely in time (under both systems, Player B's average is 65.4% of Player A's), and I don't see anything important about the fact that the current system gives an average below the raw points earned, so there's no reason to switch on that basis. But as I said in my previous post, when the points earned is about the same each week, your system IMO gives way too much weight to older results.

Here's an extended example, so our non-mathematical friends can follow along.

If I understand your assumptions, you are sticking the the two-year window and the present depreciation scheme (13 weeks at 100%, followed by a reduction of 1/92 per week for the next 91 weeks), but you are changing the divisor to the sum of the weights, with a minimum divisor of 22.5, correct?

Then let's say Moe graduated from Q school and started playing every week. He happens to be very very good, and he earns 10 points every week (that's about like finishing 7th at Bay Hill).

After 24 weeks, the sum of his event weights is 23.28, which exceeds your minimum divisor for the first time, so he is now getting his full value in the rankings. Since he's earned 10 points each week, and it's a weighted average, he has an average of 10.0.

Under the current system, he would be well under the divisor of 40, so he would have an average of 5.82 (232.8/40). But he would still be ranked 5th in the world, which is what happens when you rack up 24 consecutive top tens in strong events, I guess.

He keeps it up, and after 40 weeks, he meets the current WGR minimum divisor, and gets full value under the current ranking system. By now his total event weight is 35.89, so his average under the current system is 8.97, and he is tied for third in the world. His average under your system remains at 10.

Just by the way, this is the highest he will ever be under the current system, so anybody who thinks the WGR doesn't give a bonus for winning can take comfort in the fact that a guy who has 40 straight top tens in strong events can't get past third (based on current competition).

12 weeks later (still earning 10 points per week), he reaches the maximum divisor under the current system, 52. The total weight of his events is now 43.52, and look what happened, his average under the current system has fallen to 8.37 (435.2/52). This is the problem we both have with the current system --- good results penalize strong players as they age. I've always used winning a major as an extreme example, but this example (top tens in strong events) may be better, because it shows that the anomalies begin before the events are even a year old. So maybe, in addition to shortening the window and depreciating less, they should also do away with the maximum divisor, and just use the same divisor for everybody, regardless of events played. If you play too few, you are penalized, and if you play more than the minimum, only your last 40 count. I need to think about that some more.

But back to the scenario -- from now on, as long as Moe keeps earning 10 points a week, nothing changes in either system. His average is always 8.37 under the current system, and it's always 10.0 under yours. OK so far.

But after a couple years of this, Moe crashes his clown car into a fire hydrant, and a dozen Colombian prostitutes spill out of it. Moe's wife demands that he give up golf for ten months. Moe stops playing. And here's where the problems with your system crop up.

Under the current system, as soon as Moe stops playing, his average starts to drop, because his events are depreciating. He loses around a tenth of a point a week -- a little less at first, a little more later. In six weeks, his average drops below 8, in 16 weeks, it's below 7, in 25 weeks, it's below 6, and in 34 weeks, it's below 5. After 42 weeks, it's right at 4.

Well, that makes sense, right? If a guy hasn't played for 10 months, he should expect to lose some ground, right? That's what happened to Tiger after his surgery in 2008, when he took 8 months off, except that when he stopped playing, his average was over 21(!) rather than 8, so even though he lost 12 points in those 8 months, he remained #1. And almost everyone, including me, thought that was wrong. IMO, if I can give him three shots a side (as I could when he couldn't even walk), he shouldn't be #1. After, say, six months of inactivity, his ranking should take a severe hit.

But under your system, Moe wouldn't have lost even a tenth of a point. If another player had matched Moe, and earned 10 points a week every week Moe did, and then went on earning top ten finishes every week after Moe stopped playing, he would not pull ahead of Moe in the rankings until 42 weeks had passed. For ten months, Moe could sit on his couch and never touch a club, and his average under your system would still be 10.0. It's not until week 42 that the weight of his events over the last two years finally falls below 22.5, and the minimum divisor finally starts lowering his average. Without that, his average would stay at 10.0 for two full years, even though he never played.

So both systems have problems, but I think that golf fans would be less happy with your system than the current one. I realize my scenario of a guy getting top tens every week is unrealistic, but that was just done to make the math easier; it would work the same way, albeit less dramatically, for anyone who took an extended layoff.

If one of the secondary goals of the WGR is to encourage players to play more often (and the minimum divisor shows that it is), your system gives reverse incentives --- any time a player has a hangnail, he can take the week off, and it won't lower his average. And every time he plays, since he gets full weight for recent events, he makes up (minimum divisor-wise) for missing two events 14 months ago, or ten events 22 months ago.

So I still think that the way to go is to keep the current system, but shorten the window to one year, and not speed up the depreciation to compensate, so that the oldest event in the window is still worth about 57% of its full weight. It would be more volatile, but also more fair.

FWIW, according to the guys I talked to on the OWGR committee, the OWGR originally started as cumulative weighted points with no averaging at all- thus rewarding guys who played more often.  However, like Tennis, this often produced a #1 player who was a guy who played a lot but wasn`t considered the favorite to win when other top players, who played less often, were in the field.  Thus they went to their "averaging" method.  (It also was based on a 3 year period at one point)

[brocks]

Thanks for your input!

You're very welcome, and thank you for taking my post in the spirit intended.

[turtleback]

I think if we step back for a moment it is self-evident that if the #1 player is bouncing around from week to week, whether he happened to play that week or not, then the whole thing is a bit of an empty exercise for determining the best player in the world.  In that case there is NO best player in the world.  You can't be the best player one week and then not the best player the next week.  It isn't until someone seizes the #1 spot and then holds it for a period of time (measured in quarters, rather than in months) that they can be recognized as the best player in the world.  Luke is a very good player but is is joke to call him the best player in the world based on WGR.  Same with Rory.

I would argue that the only "best players in the world" that we have had in the OWGR era are Norman, Price, Faldo, Vijay, and Tiger.  Every one else who has held the #1 spot has a very weak claim to best in the world, IMO.  Maybe an honorable mention to Seve.

[bplewis24]

Originally Posted by turtleback

In that case there is NO best player in the world.

Word.

[brocks]

I think if we step back for a moment it is self-evident that if the #1 player is bouncing around from week to week, whether he happened to play that week or not, then the whole thing is a bit of an empty exercise for determining the best player in the world.  In that case there is NO best player in the world.  You can't be the best player one week and then not the best player the next week.  It isn't until someone seizes the #1 spot and then holds it for a period of time (measured in quarters, rather than in months) that they can be recognized as the best player in the world.  Luke is a very good player but is is joke to call him the best player in the world based on WGR.  Same with Rory.  I would argue that the only "best players in the world" that we have had in the OWGR era are Norman, Price, Faldo, Vijay, and Tiger.  Every one else who has held the #1 spot has a very weak claim to best in the world, IMO.  Maybe an honorable mention to Seve.

I agree, but I would go even further. I don't think the rankings have EVER determined the best player in the world. "Best player in the world" is a matter of consensus of the golf fans and press, and therefore it's never official. When Lee Westwood was #1, most people thought that Graeme McDowell or Martin Kaymer was the "true" best player, and lately most people have thought Rory is the best, even when Luke Donald was the official #1. In both Lee's and Luke's cases, they were not widely accepted because they hadn't won a major. But it's more than that, and it's more than holding #1 for several quarters, because Tiger had 14 majors and had been #1 for five straight years when he started stinking it up in 2010, and yet nobody, not even me, thought he was the best player in the world by August, even though he didn't lose his #1 ranking until Halloween. The rankings determine who's #1 (and below) in the rankings, no more and no less. It's an important function, and it shouldn't be minimized, but if a player isn't already the "best in the world" by consensus, then becoming #1 won't change that in the minds of many people. The rankings are a lagging indicator, so in a sense, they are the last to know who the best player is. That doesn't mean the rankings have no credibility; it just means that people are misusing them if they try to force a correspondence between #1 and best player. The rankings reward winning, but not ONLY winning. They are also designed to reward consistency and frequent play, and to penalize infrequent play. If a modern Bobby Jones came along and only played the four majors, and won three of them each year and placed second in the fourth, his average at the end of the year would be right around 10.0, a little more or less depending on which major he didn't win, because the minimum divisor would kill him. Most years, a 10-point average wouldn't be enough to make him #1 in the rankings (although it would right now). At the end of 2000, Phil had an 11-point average, and was only fourth in the rankings. And yet, I think most people would agree that Young Bobby was the best player in the world. They did when the original was playing like that.

[Wansteadimp]

Originally Posted by brocks

I agree, but I would go even further. I don't think the rankings have EVER determined the best player in the world. "Best player in the world" is a matter of consensus of the golf fans and press, and therefore it's never official. When Lee Westwood was #1, most people thought that Graeme McDowell or Martin Kaymer was the "true" best player, and lately most people have thought Rory is the best, even when Luke Donald was the official #1. In both Lee's and Luke's cases, they were not widely accepted because they hadn't won a major.

But it's more than that, and it's more than holding #1 for several quarters, because Tiger had 14 majors and had been #1 for five straight years when he started stinking it up in 2010, and yet nobody, not even me, thought he was the best player in the world by August, even though he didn't lose his #1 ranking until Halloween.

The rankings determine who's #1 (and below) in the rankings, no more and no less. It's an important function, and it shouldn't be minimized, but if a player isn't already the "best in the world" by consensus, then becoming #1 won't change that in the minds of many people. The rankings are a lagging indicator, so in a sense, they are the last to know who the best player is.

That doesn't mean the rankings have no credibility; it just means that people are misusing them if they try to force a correspondence between #1 and best player. The rankings reward winning, but not ONLY winning. They are also designed to reward consistency and frequent play, and to penalize infrequent play.

If a modern Bobby Jones came along and only played the four majors, and won three of them each year and placed second in the fourth, his average at the end of the year would be right around 10.0, a little more or less depending on which major he didn't win, because the minimum divisor would kill him. Most years, a 10-point average wouldn't be enough to make him #1 in the rankings (although it would right now). At the end of 2000, Phil had an 11-point average, and was only fourth in the rankings.

And yet, I think most people would agree that Young Bobby was the best player in the world. They did when the original was playing like that.

People thought Graeme McDowell was the best player in the world? Even as a European fan I never really thought that, Kaymer yes - I thought he was going to dominate - but messing with his swing has ruined that.

I've always steered away from comparing the rankings year on year, I assumed that the various changes in calculation over the years rendered it meaningless. Unless there is a resource that has redone the calculations historically?

[brocks]

People thought Graeme McDowell was the best player in the world? Even as a European fan I never really thought that...

Really? A lot of Americans did, at least based on what was being posted on the old TGC board. I'm talking about the beginning of 2011. He was the reigning US Open champ, he had backed it up with a win at the Andalucia Masters, he played brilliantly in the Ryder Cup, and to put a cherry on it, he beat Tiger head to head in the last round of the Chevron World Challenge, making fantastic putts on the final hole of regulation, and the first playoff hole. Maybe that event didn't get as much coverage in Europe, but it was a very, very impressive win to many Americans, including me. And then he began 2011 with two top three finishes, while the reigning world #1, Lee Westwood, opened his 2011 season with a T64 and a MC. He had won the Nedbank Challenge at the end of 2010, but a lot of Americans didn't notice or care, because it was the same week as the Chevron (maybe it worked the opposite way for you?) At any rate, there was no doubt in my mind at the time that McDowell was the better player.

[Wansteadimp]

I guess it was McDowell's history of streakiness, we'd seen him Europe win and play well for a couple of months and then go missing for a stretch previously. It never surprise me he was in contention/winning but I don't think I ever thought he was favourite for a tournament.

Ryder Cup is Ryder Cup, its match play, funny things happen (Poulter doesn't annoy me for example), confidence is key and GMac had a lot at that time.

I still think Kaymer will be a player with multiple majors. Even last year when he was terrible he walked away with a WGC title and a win against a good field in the Gulf.

And finally Sky have finally showed some golf from Quail Hollow!