• DG Betting Blog
It is often the case that “course history” dominates much of the discussion leading into PGA Tour events. A lot of the chatter this week has surrounded fellow Canadian, and PGA West darling, Adam Hadwin. Hadwin has played the Desert Classic each of the last four years, collecting impressive finishes of T48, T6, 2nd, and T3 while averaging 2.24 strokes per round better than the average PGA TOUR player. To put these performances into context, since joining the PGA TOUR in 2015 Hadwin has averaged just 0.58 strokes better than the average tour player.
While his record at this event is interesting, does it actually contain any predictive value? We know from last week’s blog that golf scores are very random, and looking at small samples can often be misleading. Let’s try something similar this week, but focus our attention solely on course history.
Suppose we take all of Adam Hadwin’s scores since 2015 (391 rounds) and randomly distribute them across the courses he has played in that time span. What this does is eliminate any “course history” narratives as each course now has a random sample of scores assigned to it. The tables below summarize his top 10 strokes-gained performances by course in both the real and simulated data.
While his record at this event is interesting, does it actually contain any predictive value? We know from last week’s blog that golf scores are very random, and looking at small samples can often be misleading. Let’s try something similar this week, but focus our attention solely on course history.
Suppose we take all of Adam Hadwin’s scores since 2015 (391 rounds) and randomly distribute them across the courses he has played in that time span. What this does is eliminate any “course history” narratives as each course now has a random sample of scores assigned to it. The tables below summarize his top 10 strokes-gained performances by course in both the real and simulated data.
The numbers speak for themselves: even after removing the possibility of a course history effect, we still have many examples of Hadwin playing well above his ability at specific courses in the simulated data (even though the course label is meaningless). This suggests that the real course history numbers for Hadwin could be very misleading and just a product of the randomness inherent to golf scores.
We can get a sense of just how much betting markets are adjusting their view of Adam Hadwin by examining their offered odds over the last two weeks. Last week at the Sony Open, Hadwin’s odds for a top 20 were about 4.0 (implied odds of 25%). This week, against only a slightly weaker field, his offered top 20 odds are now 2.1 (implied odds of 47%).
Using our model of golf scores, we know roughly how to translate the predicted abilities of golfers (in units of strokes-gained per round) into finish probabilities. For example, a golfer who has the skill of an average PGA Tour professional would have a top 20 probability this week of approximately 13%. We can easily invert this mapping to get the ability levels implied by Bet365’s offered odds. First, we remove the “juice” from their odds (in a slightly complex way that assumes more juice is applied to longer odds), and second we map those odds to implied ability levels. The result is that at the Sony Open Bet365’s odds indicated they thought of Adam Hadwin as a +0.55 strokes-gained per round golfer (relative to the average PGA Tour pro); this week, they think of him as a +1.1 strokes-gained per round golfer. That is a massive shift, especially given that Hadwin played mediocre at the Sony: our predicted skill level for him was roughly +0.85 before and after that event.
Therefore, it looks like Bet365 is responding very strongly to Hadwin’s course history at PGA West. This week is an opportunity to take a stance against Hadwin (although we have not yet found a chance to do so).
Another golfer who has an even smaller sample of rounds at PGA West (8), but has performed very well in those rounds, is Jon Rahm. Again, Bet365 seems to be responding too strongly to this course history: the ability implied by their offered odds for Rahm went from 1.95 at the Sentry to 2.35 this week, while our estimates of Rahm’s ability went from 1.95 to 2.04 (Rahm played well at the Sentry). We have managed to find value betting against Rahm: we’ve taken Justin Rose in a tournament match-up against Rahm, where Rose is the underdog! (2.10 offered odds.)
We can get a sense of just how much betting markets are adjusting their view of Adam Hadwin by examining their offered odds over the last two weeks. Last week at the Sony Open, Hadwin’s odds for a top 20 were about 4.0 (implied odds of 25%). This week, against only a slightly weaker field, his offered top 20 odds are now 2.1 (implied odds of 47%).
Using our model of golf scores, we know roughly how to translate the predicted abilities of golfers (in units of strokes-gained per round) into finish probabilities. For example, a golfer who has the skill of an average PGA Tour professional would have a top 20 probability this week of approximately 13%. We can easily invert this mapping to get the ability levels implied by Bet365’s offered odds. First, we remove the “juice” from their odds (in a slightly complex way that assumes more juice is applied to longer odds), and second we map those odds to implied ability levels. The result is that at the Sony Open Bet365’s odds indicated they thought of Adam Hadwin as a +0.55 strokes-gained per round golfer (relative to the average PGA Tour pro); this week, they think of him as a +1.1 strokes-gained per round golfer. That is a massive shift, especially given that Hadwin played mediocre at the Sony: our predicted skill level for him was roughly +0.85 before and after that event.
Therefore, it looks like Bet365 is responding very strongly to Hadwin’s course history at PGA West. This week is an opportunity to take a stance against Hadwin (although we have not yet found a chance to do so).
Another golfer who has an even smaller sample of rounds at PGA West (8), but has performed very well in those rounds, is Jon Rahm. Again, Bet365 seems to be responding too strongly to this course history: the ability implied by their offered odds for Rahm went from 1.95 at the Sentry to 2.35 this week, while our estimates of Rahm’s ability went from 1.95 to 2.04 (Rahm played well at the Sentry). We have managed to find value betting against Rahm: we’ve taken Justin Rose in a tournament match-up against Rahm, where Rose is the underdog! (2.10 offered odds.)