LOOKING AT THE 2006 RYDER CUP

Nothing fascinates me like the Ryder Cup. It is one of the only times in golf that people are actually required to evaluate talent and explore strategy. Not surprisingly, they are often wrong. Over the coming weeks and months, I’m going to take a look at some of the factors that play into selected the team, picking the match-ups. Let’s start with 2006:

Common wisdom suggested that the United States struggled in the Ryder Cup because they just couldn’t get the hottest players on the team. As usual, common wisdom seems like complete garbage, so I set about to take a look. Paul Azinger, who “cracked the code,” set about a new points system that allowed him to pick players that had played better in the given the year, or were on a roll coming into the Ryder Cup.

Did this matter?

Well, not really.

I took a look at players 2-year average, leading up to 2006 Ryder Cup, just 2006, Average from British Open to PGA in 2006(basically month leading up to when Captain makes picks) and average the month before the Ryder Cup was played. Then I compared that to average points per match of each player in the contest. Here’s what I got:

USA 2-year Average: -.462
European 2-year Average: -.431

Not surprisingly the US team was better, but not by much. I’d definitely consider that with some kind of home field advantage the Euro’s might have even been the favored sqaud. But that’s not where it gets interesting:

USA 1-year average: -.508
Euro 1-year average: -.437

USA 2-month: -.566
Euro 2-month: -.369

USA 1-month: -.549
Euro 1-month: -.340

As you can see, the US was relatively “hotter.” Their players had significantly better 2006’s and were much better coming into the event on the short term. The Euro’s were actually pretty far below their average in the months preceding up to the event. Remember, they promptly took their poor form smoked the United States team in 2006.

So, basically Azinger missed diagnosed a problem and adjusted accordingly. The real problem with the US team was that it contained players like J.J. Henry, Vaughn Taylor and Brett Wetterich, who were in the midst of having career years then fell apart. Obviously, that’s not what you’re looking for when your projecting(that’s what captains are doing) performance forward.

*I have to give credit to Tom Lehman. His picks of Verplank and Cink were probably very good ones.

Here’s a look at the correlation between each of the ranking and average points earned:



As you can see, a small correlation (remember negative is good. The better you are, the lower your actual score in my rankings and the higher the points you should be expected to earn) between being a good player and average points per match earned. That’s not surprising considering the small sample size of matches is incredibly random. It’s actually pretty normal compared to a stroke play event.

Again, not surprisingly, 1-year was less strongly correlated as that is a smaller sample more susceptible to variance. The even shorter sizes were negatively(actually positively, but that means low scores translated in more points) correlated. Again, not a huge surprise from what I’ve seen before. On a week-to-week basis recent form varies so widely and is only fractionally more important than overall skill, if any.

None of that surprises me, but it’s also not entirely fair, either. While average points per match does even out a different number of matches played by players, that still doesn’t bring into account a different strength of opponents as well as different strength of partners. So, I went back to even that out:

Basically, I compared each players opponent to their possible average among remaining teammates. For example, Jim Furyk played with Tiger Woods all four group matches. That’s an average of -1.08 when the rest of the American team Furyk could have played with would have been around -.42. So, Furyk played with a partner that was about -.66 standard deviations better than the US average.

The next step was to compare the player’s opponents to the average of European groups. The average American group would be around -.43. Of course, that’s not very comforting if you get Tiger and Furyk who are around -.93 together.  Then I used a match play simulator to adjust the number of points won per match based on both your opponent’s difficulty and the strenth of your partner.  Here’s what I got:

PPM=Points Per Match

Delta PAR/OPP= Difference from actual PAR/OPP compared with average PAR/OPP

USA Actual PPM delta PAR delta OPP ADJ PPM
Tiger Woods 0.60 0.36 0.02 0.53
Phil Mickelson 0.10 -0.04 -0.07 0.12
Jim Furyk 0.40 0.65 0.01 0.27
Chad Campbell 0.33 -0.15 -0.03 0.37
David Toms 0.13 -0.02 -0.06 0.14
Chris DiMarco 0.13 0.24 0.03 0.07
Vaughn Taylor 0.25 -0.17 0.00 0.28
J.J. Henry 0.50 -0.01 0.18 0.47
Zach Johnson 0.38 -0.06 0.02 0.38
Brett Wetterich 0.00 0.07 -0.05 0.00
Stewart Cink 0.50 -0.13 0.09 0.51
Scott Verplank 1.00 -0.11 -0.01 1.02
EUROPE Actual PPM delta PAR delta OPP ADJ PPM
Luke Donald 1.00 0.21 -0.17 0.99
Sergio Garcia 0.80 0.19 -0.18 0.80
Henrik Stenson 0.50 0.00 0.11 0.48
David Howell 1.00 -0.17 0.20 0.99
Colin Montgomerie 0.50 -0.11 -0.01 0.52
Robert Karlsson 0.38 -0.19 -0.18 0.45
Paul Casey 0.75 -0.16 -0.05 0.79
Padraig Harrington 0.10 -0.06 -0.07 0.13
Jose Maria Olazabal 1.00 0.19 -0.16 0.99
Paul McGinley 0.33 -0.03 -0.03 0.34
Darren Clarke 1.00 -0.15 -0.18 1.07
Lee Westwood 0.80 0.06 -0.07 0.80

Now a few graphs:



Once again, IN 2006 BEING A “HOTTER PLAYER” TRANSLATED INTO LESS POINTS AT THE RYDER CUP.

No matter how you slice it, the Ryder Cup is incredibly susceptible to a lot of variance.  Twenty-eight Matches in different formats over 3 days just leads to a lot of randomness.  It’s important to remember that when people take an incredibly-biased-results-based view of what happens this September.  It’s also important to remember that like most golf tournaments it’s not surprising that recent form didn’t mean too much in the 2006 Ryder Cup.  Sure, maybe it’s enough to tip the scales over equal players, but overall, in the long run, any team is better off just having the better players.

Advertisements

8 Comments

Filed under Uncategorized

8 responses to “LOOKING AT THE 2006 RYDER CUP

  1. am19psu

    I’d be very hesitant to assign any information to anything that only explains 4% of the variance. What were the p-values on the coefficients? I can’t imagine they are significant. It seems to me, given the small samples, it’s more analogous to a 5-game series where anything can happen due to large variance.

  2. am19psu

    That said, I think your overall conclusion is correct from a Bayesian perspective. I just don’t think these graphs showed anything relevant.

  3. “It seems to me, given the small samples, it’s more analogous to a 5-game series where anything can happen due to large variance.”

    Right.

    That’s the ultimate point I was trying to make. RC is an incredibly small sample size, with lots of variance.

    The other, would be to somewhat disprove the notion that recent form or a good season really matters. It’s too small a sample to say either way with just this, but it definitely didn’t in 2006 and my other “research” would tend to say that overall it doesn’t that much.

  4. am19psu

    Sure, but the point I was trying to make was that your two year form doesn’t really matter either, given the numbers you presented. Like I said, from a Bayesian perspective I agree with you, but your numbers don’t show that. By your numbers, you’d be just as well pulling a bunch of American PGAers out of a hat.

  5. N.

    Haha I liked this post 😉

    As you probably know, the smaller correlation between 1 year performance and adjusted PPM does not necessarily indicate that 1 year performance matters less than 2 year performance. What it does indicate is that 1 year performance was overvalued by the selection process.

    So your post shows the system is broken. In order to fix it, it might be good to see exactly how much 24/12/6/3 month performance matters for the next week or month. (For example, is 1 year average a better predictor than 2 year average?) In your Rankings FAQ, you say your “limited research [indicates] the relative importance of recent tournaments varies widely and is only fractionally more important than random tournaments.” I’m interested in that research, and may even do it on my own at some point.

  6. “Sure, but the point I was trying to make was that your two year form doesn’t really matter either, given the numbers you presented.”

    You’re right. I think where I messed up in doing this was focusing on Ryder Cup(incredibly random like a 3-game series or something) and trying to use that to disprove stuff that I had a good feeling is overrated. I’ve done this before without the Ryder Cup and come up with more meaningful data, so it was safe for me to make the assumptions, but based on this your right I basically said proved nothing mattered.

    As always, I think randomness is probably the biggest factor in determining who plays well in a tournament, in the Ryder Cup it might be even bigger.

  7. “So your post shows the system is broken. In order to fix it, it might be good to see exactly how much 24/12/6/3 month performance matters for the next week or month. (For example, is 1 year average a better predictor than 2 year average?) In your Rankings FAQ, you say your “limited research [indicates] the relative importance of recent tournaments varies widely and is only fractionally more important than random tournaments.” I’m interested in that research, and may even do it on my own at some point.”

    I did a short post on this a while back, I think it was called “Did last week matter?” Let me know what you’re thinking about and I’ll see what I can work up/send to you to do on your own.

  8. Okay, you got me curious. I’m going to spend the week adjusting my rankings to come up with some kind of Ryder Cup ranking.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s