This posts builds on an earlier post of ours that attempted to quantify pressure in golf. We construct a measure that is similar in spirit to the strokes-gained measures used by the PGA Tour to analyze putting, approaches, and driving. The purpose of this measure is to analyze which players perform the best when they get in contention to win a golf tournament relative to their normal standard of play, and also relative to how their peers perform when they get in contention.
First, I’ll outline how the measure is constructed:
- Calculate the difference between a player’s score and the field average for that day. Call this the relative score.
- Using only round 1 and 2 scores, calculate the average relative score for each player. This is meant to capture how a player plays relative to the field in low pressure situations.
- Using only round 4 data, calculate the average relative score for each player in different “pressure zones”. For example, our “Playing Near/With the Lead” (PWL) zone is being in a position of 5th or better heading into the final round, and our “Chasing the Leaders” (CL) zone is being in 6th-12th place heading into the final round.
- Let’s just focus on the PWL measure for now. For each player, take the difference between their average PWL relative score and their average relative score under no pressure.
- To clarify, at this point we have a value for each player which indicates how much their average relative score differs between no pressure situations and the PWL zone. The final step is to calculate a weighted average of all these values (where the weights are equal to the number of times a player gets in the PWL zone). Then the “strokes gained in the PWL zone” will be a player’s value minus the average of all the players’ values.
If this was a bit confusing, an example will make it clear. Let’s calculate Tiger Woods’ strokes gained in the PWL zone as an example. This is done using PGA Tour data from 1983-2016.
- In rounds 1 and 2, Tiger’s average relative score is -2.4186.
- When Tiger is in the PWL zone (top 5 heading into the final round) his average relative score is -2.7168.
- So this says that Tiger plays 0.2982 shots better than his usual standard of play when he is in the PWL zone.
- Additionally, we calculate (by averaging all of the measures constructed like Tiger’s above) that the average player plays 0.118 shots worse when they are in the PWL zone. Therefore, we say that Tiger is gaining 0.2982 – (-0.118) = 0.416 strokes on the average player when he is in the PWL zone.
We are not trying to make the claim that this statistic is solely quantifying pressure. We think of it as reflecting a player’s ability to close tournaments, a player’s ability to continue a hot streak of good play, and also a player’s ability to perform under pressure. As much as possible, we want this to just be a descriptive statistic; it tells us whether a player tends to play better or worse than usual when they are in contention.
The main issue with this statistic is sample size. There are many players who have only made it into the pressure zones once or twice. In these instances, the statistic is uninformative with regards to that player’s true ability when in contention. Fortunately, using data from 1983-2016 on the PGA Tour, there are many players with a substantial number of trips to the PWL zone. The following table shows the strokes gained in the PWL zone (top 5 entering final round) for all players for which we have at least 40 observations.
To conclude, this statistic indicates how a player performs, relative to his usual standard of play, when he is in a pressure zone compared to how the average player performs in the pressure zone. This is analogous to strokes gained putting, for example, which indicates how a player performed on a specific length putt compared to how the average player performed on it.
I misunderstood the approach first time through (read a little hastily). The approach to compare in contention performance relative to their average performance (and the average for all ‘in contention’ performances) seems sound.
I’m not sure that to some extent you aren’t capturing relative scoring when a player is on form vs off (which is still informative). There has to be more pressure when in contention, but to be in contention you have to get there first with a stronger than typical performance over three days (unless you’re Tiger).
Maybe don’t average relative score for rounds 1 & 2 where they didn’t make the cut (off form)? Also not sure why you’d exclude 3rd round when that matters equally to 1 & 2 for getting into contention?
Some players may also adopt a more conservative strategy in earlier rounds to not ‘play themselves out’ of a tournament and make the cut then play more aggressively on the weekend. Some may play aggressively from the start and tend to ‘boom or bust’ if their game is ‘on’ vs ‘off’.
We only use round 1 and 2 because we are trying to get a measure of how the players plays relative to the field in pressure-free situations. Round 3 there could be some pressure perhaps, and sample size is big enough that we don’t need to include it.
We call it Strokes gained in contention because we are not claiming it solely quantifies pressure. It could be, as you say, also capturing whether a player plays really well when they are on (i.e. in contention). But, if you are saying that the measure, in part, captures the fact that everyone has to be playing their best to be in contention, and this somehow confounds the results, that is wrong. Because we difference out the average for all “in contention” performances, this takes out any effect (such as playing better to get in a position to contend) that are common to all players in contention.
It may not be all that common, but I wonder about the relative pressure effect between a tight leader group and one with lots of strokes between the leader and chasers. Is someone behind the leader by 5+ strokes really PWL even if they’re in 4th place? Same if someone is CL if in 11th place, but within 2 strokes of the lead?