Comparing players from the 1940s/50s

Hi all,

A project I've been interested in for some time is creating a ranking system that can be applied equally to the middle of the 20th century as today, and I wondered if anyone had any thoughts on some of the hazards one encounters.

A pretty decent version of the modern ranking system (or at the very least, the one used in the early days, in 1986, where it was total points on all tours weighted over 3 years) can be constructed for all years going back to about 1965, as results are available (published in the McCormack annuals) and the system works pretty well. As you go further back, though, two or three problems arise.

1) The length of the USPGA tour compared to the British tour, or others. Even by the early 40s, PGA tour pros were competing in a 45-tour event season, much like today, while in 1948 the British tour consisted of just 15 events. On any "total point" system, therefore, even if one conceded that the strength of the British events (including as they did players like Cotton, Daly, Ward, Von Nida and de Vicenzo) was comparable with the US tour, the US players would outweigh the others 3-to-1. And events in Europe, Australia, South Africa etc were limited in most cases to national Opens, PGAs and limited invitational events.

2) The difficulty of comparing strength of tours. Far fewer players than today competed on both sides of the Atlantic, or globally. So while it is possible to draw some conclusions from Ryder Cup results (the GB&I; team lost narrowly at home in 1949, and almost won in 1953, for example), it's hardly an exact science. The victorious US 1949 Ryder Cup side stayed on and were all invited into the 1949 British PGA Matchplay championship, and only one (Lloyd Mangrum) reached the semi-finals, but when US pros used to travel to play the Open in those years they invariably won (Snead, Hogan). And British and Commonwealth pros hardly ever played in US tour events or majors - Bobby Locke was probably the only one who seriously tried, and had a great deal of success between 1947 and 1950, when the Tour effectively excluded him.

3) Some players not playing full schedules. This applies not only to some of the British and Commonwealth pros (see above) like Peter Thomson, who might only play 10 events a year, but the elephant in the room is of course Ben Hogan, who between 1950 and 1955 played a total of about 25 events and won half of them, most of them majors. How to compare Hogan winning 2 or 3 majors a year out of 4 events entered, with Sam Snead winning 10 tour events out of 20?

I've been trying to find a way of objectively distilling results in a way that accommodates all of the above, and then as a test, applying that system to, say, 2014, and seeing if it still produces an outcome that makes sense.

Has anyone - long shot - ever tried anything similar, or does anyone have any thoughts on the above?

It is a laudable project, but if your goal is objectivity you are defeated from the start.  Any attempt to impose objectivity on an intrinsically subjective topic is doomed to failure.  You can bury the subjectivity deep so no one can see it if they don't know what they are looking for, but it will be there.  You can no more make this a matter of objective knowledge than you can objectively answer the question" "which blue is bluer".

As to the project itself, it is a big bite you have bitten off.  I was a pension actuary for over 20 years so I am pretty familiar with large scale mathematical modeling.  You have so many interrelated factors involved in what you are trying to do that your model is going to get huge.  And as it gets bigger and bigger there are going to be more and more value judgments and weightings to make (there goes that objectivity).

But good luck, it will be interesting to see your results.