In the days of my youth I was a devotee of horse racing.
Most of my betting activity was theoretical with my biggest outlay a daily copy of the now defunct Sporting Chronicle and the weekly Handicap Book. With plenty of time to spare (I was unemployed for the best part of a year in the late sixties) I became reasonably adept at reading form and, though I never kept a proper record, I believe that my small bets at least covered the cost of the publications I devoured enthusiastically.
The back page of the Handicap Book included a selection of readers’ “infallible” systems which, inevitably, turned out to be less than infallible though many possessed a grain of sense. Several involved the number obtained by adding together the horse’s last three placings and one which I refined into a useful guide for eliminating losers involved betting on whichever horse in a handicap was closest to the mid-point of the weights provided it had been placed last time out.
The Handicap Book also offered an array of professional systems (ie, the advertisers were keen to part the punter from his money) one of which came to mind a few days ago and is the reason for this post.
The system required a diligent reading of the racing papers (preferably, though it was claimed that it could be done with any daily paper) since it listed about 30 “key” races which gave the pointer to the selection which was to be backed next it time it ran if it had been placed first (or in some cases second) in the “key” race.
The system provided “proof” that it worked by giving the results for the previous five seasons — and aficionados of tree-ring proxies might just be starting to see where this is going! If I recall correctly, the system had shown a profit of about 50 points to a one point level stake each year with a strike rate of about 75%. Impressive.
Needless to say the system failed miserably in each of the following five years and there was, of course, no way that the bookies could be persuaded to ignore the real-life results and pay out on the theory!
Further research through the pages of the Raceform annuals showed that prior to the five year “validation period” the system had been equally wayward and in one especially perverse incident no fewer than six qualifiers turned out for their next race in the same race (I have a vague recollection that it was actually the Ayr Gold Cup, but what the hell!) and far from running the six-fold dead heat that ought to have resulted, none of them actually passed the winning post first.
There are several conclusions that may be drawn from this, the obvious one being that the seller of this system was a crook and that anyone who paid good money for it was a mug. (I was never quite that much of a mug; the copy came second-hand from someone who was!). It’s also possible that he thought he was onto a good thing and was as surprised as anyone when it failed to work.
I’m sure that we should all exercise a little Christian charity and assume the latter but it’s hard to do, especially since it was the work of only a couple of days to go back in time and prove conclusively that his research did not stand up to scrutiny. A sceptical turn of mind would also ask why, in a field which is less chaotic than the climate (most of the time, at least), only 30 races out of the 2500+ that comprised the Flat season in those days should qualify as key races for this system. One barely needs any scientific training to draw the conclusion that to select 30 races which happen to give the required result over a limited period of time is cherry-picking of the first order and that it would equally be possible to select another 30 to give the desired result over a different time period or to select 2,470 that failed miserably.
The point of course is that as soon as it becomes clear that your hypothesis breaks down you are supposed to go back to basics and find out why and come up with a new hypothesis. The fact that there are comprehensive records that allowed me to prove this system was (and had always been) rubbish simply made life easier. Where no such reliable records exist there is a temptation to make unwarranted assumptions though why a reputable scientist would fall into that trap when it is inevitable that — sooner or later — someone will cotton on is difficult to understand. The assumption, in climate systems as in racing systems, has to be that once the facts start to diverge from the theory then the theory is wrong. And if the theory is wrong then it always was wrong and any time when it was “right” was purely coincidence.