Myth-busting: Trends Analysis

Trends are a popular talking point as we approach the Festival

In preparation for a series of "myth-busting" articles, Timeform's Head of R & D, Simon Rowlands, warns about the pitfalls of simplistic trends analysis...

Ah, the countdown to the Cheltenham Festival: there is nothing quite like it! 

I write that not in fevered anticipation of the big event itself - though that fevered anticipation will undoubtedly transpire even for an old cynic like me - but in the certain knowledge that no time of year produces such an overload of half-truths and downright falsehoods, which are then recycled endlessly through the prism of a hundred Festival guides and preview nights. 

What better time, then, to be asked to write a "myth-busting" series? It should be like shooting fish in a barrel. But, before moving on to bust some specific myths (and to show that a few of the myths are not myths at all), let us return once more to the subject of "trends". 

"Trends" - in which apparently positive or negative generalities are identified from past editions of the same race - do sometimes have merit. But they are also the source of a great number of the aforementioned half-truths and falsehoods. 

The good news is that there is no better time of year to capitalise as a punter by being healthily sceptical about this approach.  

I dealt with this in a 2009 blog, and what I said then holds true now (with only a slight amount of rewriting):

Evolutionary biologist Richard Dawkins noted in The God Delusion that it is a human foible to see patterns and meaning where patterns and meaning almost certainly do not exist. As a punter, I have to say that sounds familiar.

A recent New Scientist feature expanded on the debate, including as follows: "The over-attribution of cause and effect probably evolved for survival. If there are predators around, it is no good spotting them nine times out of 10. Running away when you don't have to is a small price to pay for avoiding danger when the threat is real."

The article went on to describe the findings of an academic experiment, which asked people what patterns they could see in arrangements of dots or stock market information. Half of the participants were made to feel a lack of control before the experiment took place and half were not.
The results were striking. The subjects who sensed a loss of control were much more likely to see patterns where there were none. When we feel a lack of control we fall back on superstitious ways of thinking.

Replace arrangements of dots and stock market information with horseracing results and the dynamics of the Betfair market - a small step in my view - and you should be able to see where I am coming from with this one.

There are patterns and meaning in horseracing results - if there were not, there would be little point in trying to predict anything from them - but the regular over-attribution of cause and effect is widespread and not to be shrugged off as a price worth paying in the way that it was in primitive societies.

You are best taking precautions against predators, whether real or imagined, but a bad bet is a bad bet and should be avoided if possible.

This is the time of year when you are likely to have patterns and meaning in a horseracing context rammed down your throat in the lead up to the Cheltenham Festival. In particular, so-called trends analysis seems to be everywhere. 

Trends exist in horseracing and can be important, of that there is no doubt. For instance, younger and more lightly-raced horses fare better or worse in handicaps at certain stages of the season than might otherwise be expected. You would be a fool to ignore such things entirely.

But trends are frequently worthless, or worse, for a number of reasons, notably when they seek to replace specific knowledge of individual horses with generalities that are far less relevant, and when they are compiled in a crude and simplistic manner, which is far too often. Here are some headline warnings where trends analysis is concerned:

• If a trend exists, then it will affect not just winners but losers and the degree to which the winners win and the losers lose: trends analysis should always reflect this. 

• In addition, concentrating on winners alone often results in small samples and questionable findings.

• Trends, where they exist, usually change over time, so be wary of information from a long way back. 

• Question seriously any findings that do not seem to obey logic and common sense: they may well be down to chance. 

• Disregard trends analysis that assumes all factors are of the same significance and which advocate filtering (e.g. Horse A qualifies on four of five criteria when the missing criterion outweighs all the others combined).

• Understand that many writers on trends, like writers about astrology and things that go bump in the night, have space to fill for a gullible public and jobs to hold down: it is in their best interests to make trends seem more significant than they are.

• The easiest way to make trends seem more significant than they are is to analyse them crudely, such as by considering winners only from the past ten years, with the result that chance happenings appear significant to the credulous.

With Cheltenham in March now on the horizon, take time out to assume control of your fate and (as the soothsayer in Shakespeare's Julius Caesar might have said) beware the ideas of March.

Look out for Brand new features with Timeform Race Passes - In-Play Hints, Running Notes & Warning Horses. Find out more at timeform.com.