By course form, I mean amateur or professional tournament performance(s) in the past by a player on a particular course – with the rider that the course should today fundamentally be the same (eg greens not changed from bermuda grass to bent grass).
Does Course History Matter?
The short answer, for those who can’t be bothered reading further, is: occasionally it does; mostly it doesn’t.
What are the elements in handicapping a field of golfers?
There are many, but I’ve summarised my primary five elements (plus the vexed ‘Miscellaneous’ category) below:
- Recent Form. Firstly, how has the golfer been playing recently? Define recently, you say. For me it’s six weeks or roughly his or her past four tournaments. Secondly, how strong have been the fields in which he has been playing? A guy coming in with three consecutive Top20’s on the PGA Tour may be rated 33/1 whereas the form of a guy coming in off three consecutive wins on the Australasian Tour will look good on paper but he may be rated 150/1;
- World Ranking. Yes, I know it’s imperfect but if I’m handicapping an asian field with a lesser-known competitor from South Africa’s Sunshine Tour & I’m trying to line him up against an equally obscure player from Japan’s JGTO Tour then their respective world rankings become relevant. (My software has an adjustment factor, by the way, for owgr points earned on all tours; it discounts Japanese points by 35%, for example));
- Course Form. How has a player has performed in previous tournaments on this week’s course? If he’s played it three times for 3xTop10’s that’s going to change my rating of his chances versus a player with identical recent form & world ranking who’s played the course four times and missed the cut every time;
- Course Fit. What are the characteristics of a player’s game versus those skills proven to be required in order to play well on this course? eg If the track is long (say 7,700yds) I will discount the chances of a guy who’s one of the shortest drivers on tour versus a long-hitter of otherwise similar ranking;
- Trends. By this, I mean is a player on the improve or decline? What are his form and career trajectories? eg Has he consistently been stepping up in performance quality over recent months / years? Or we often observe a decline in form after the birth of a player’s child. Let’s say the post-partum form line reads: 2 weeks off, mc, 2 weeks off, mc, 46, 15. We might conclude that his mind is now off nappies and back on his employment;
- Miscellaneous. This is where it gets interesting as this is mostly what we don’t know. There are myriad things that affect humans & a few will be known to us whilst most will not, but all these things affect sporting performance. Consider the following examples, among many more:
- An airline loses a golfer’s clubs, including his treasured lucky putter. We may know this and we may know from his twitter feed that he’s angry & upset about the situation and so we may downgrade his chances a bit;
- A golfer has been cheating on his wife. His wife has just found out and late on Wednesday night he’s received that dreaded confrontational and emotional phone call. As a consequence, he’s deeply unhappy & distracted when he tees off early on Thursday; we do not know this;
- A golfer may regularly wage an internal battle with his rising temper when paired with a slow, deliberate player such as Na or Cantlay. He may hide his anger but it may negatively affect his play. We may not know this;
- A golfer who typically voids his bowels before tee-off is constipated and tees-off Thursday morning with a ‘full load on board’ & is uncomfortable and plays poorly as a result. We do not know why.
My point here, if you have not yet grasped it, is that we’re assessing humans not machines and while we may have a bucketload of data on which to base our assessment of his winning chances, there is an equally large or larger shitload (no pun intended) of relevant data of which we’re blissfully unaware!
This introduces randomness & variance and, sadly, renders all golfer assessments invalid to some degree. Not only that, but in golf we’re trying to gauge the relative chances of 156 players – which is exponentially more complex than assessing the respective chances of a ‘field’ of: two football teams or eleven horses!
Golf betting, fantasy line-up construction & handicapping are very: DIFFICULT, FRUSTRATING & IMPERFECT!
Narrowing the Focus
As handicappers / assessors / gamblers we accept there is a lot we don’t know but we can only address what we do know, so we gather as much information as we can in the (limited) time available and process it to arrive at our conclusions – which we summarise as odds or prices or salaries.
I wrestled for ages with this, but I calculated that once you’ve firstly arranged a field of golfers in descending order of their chances (using rankings, knowledge and other peripheral data) then you can apply form in order to arrive at semi-intelligent odds / prices. In my estimation, that initial (pre-form) ranking process is 50% of the overall price assessment weighting.
In then applying form (the other 50% of the process), I use a mixture of ‘Recent Form’ and ‘Course form’ in a ratio that ranges from 80:20 to 100:0. Importantly, this means that my data analysis shows that recent form is, at a minimum, four times more relevant than course form and that often course form is often non-existent or irrelevant.
This conclusion is derived from mapping golfer performance (across 31 tours; Ladies & Men; 5 years) on each course versus his lead-in form to that week and his historical course form. In other words, did he perform: better, worse or the same as his prior form suggested he would?
Perhaps surprisingly, some courses suggest they merit up to a 20% ‘weighting’ while others suggest ‘zero weighting’. The average & median ratios are both 94 (recent):6 (course), by the way. There is also the issue here of possessing a sufficiently representative sample size but, since most courses are only used once a year for relevant events, I go with that I have!
- Course form does matter, on some courses; I’ve identified 72 such courses (roughly 6%) across the 31 Tours and;
- Course form is most often irrelevant, or statistically insignificant, and it represents 0-10% of all the known factors that feed into my assessment; average 3%;
- Given that we don’t know half of the relevant factors that affect golfer performance (see above), course form thus represents an average of perhaps 1.5% of all the factors influencing the golfer’s chances.
However, it can represent up to 10%, on particular courses, and that’s a big number when the odd stroke gained here and there can represent the difference between winning and placing 7th!
Therein lies the key to using course form. Use it selectively!
Good luck with your golf handicapping, betting & fantasy play!
Copyright MJ Miller (Mike) 2 October 2018