Betfair Official Blog

The Observer Effect - is it useless to try to predict how a race will be run?

"The Assistant" uses Wilkipedia, financial markets and common sense to try and understand whether the Observer Effect is worth the paper it's written on

I was planning to be clever this week. I'd read an article about the observer effect and thought it offered a high-brow explanation of why, when betting, it's a waste of time considering how a race will be run.

I wanted to give the impression I knew about such things. That as well as being a damn good assistant trainer, and the most eligible bachelorette in British racing (oh bugger it, make that world racing), I could also hold my own at a dinner party with Curie, Greer and Vorderman.

Unfortunately, on reflection, it seems the observer effect has nothing to do with how a race will be run.

You see, I was sitting in the doctor's surgery, awaiting that quinquennial destruction of dignity that all girls love. Having searched in vain for Country Living or Heat, I could only find some godforsaken geek-journal to entertain me. In it was an article on the observer effect, and it was then, when probably not in the best state-of-mind for cogent thought, that I made the link I now know to be rubbish.

Let me explain. The effect describes how the act of observing something can change its behaviour. For example, a pride of lions may behave differently if being shot by a cameraman. When I say shot, I mean in the film making, not ballistic, sense. Although not a wildlife expert, I would say that gunfire always affects something's behaviour. Often by killing it.

I've seen the effect in action. One of our mucker-outers, let's call him Phil (because that rhymes with his real name), takes slap-dash-ness to Olympic level. Usually, it's hard to distinguish which of his assigned boxes are mucked-out and which not. But stand and watch, and his work takes on a newfound artistic precision.

I was going to say the same occurred with races. How on numerous occasions I had analysed a race, found there were no horses with a history of leading and so had suggested to the boss we instruct the jockey to go for a soft lead. And how, on every occasion, we found our horse four-wide, going through the first furlong of 12 at an all out sprint, as every other trainer-jockey combination thought the same thing.

Then I realised it wasn't an example of the observer effect. Because it's not about passive interference. Rather, the reactions of trainers and jockeys to their own predictions of how a race would be run was something altogether more complicated.

Then I thought it was a bit like financial markets, where predictions about boom and bust effect how market participators act, and so the booms and busts become self-fulfilling prophecies.

But after an hour on Wikipedia reading about technical analysis, efficient market hypothesis and game theory, I realised I was stuffed, as I didn't have a clue what any of it meant. And how I probably wouldn't be invited to the dinner party of great female minds after all.

I even had an uncomfortable half hour (even more uncomfortable than the smear test that started this whole thing off) where I was trying to understand how Schrödinger's cat might be relevant and asking whether a race existed before it was run? And did it exist in one form, or in multiple forms, depending on the as yet undetermined actions of the agents involved?

Then I saw that Phil had left a stable door open and a £300,000 colt was wandering unsupervised across to the mare's barn. I wondered if, like Schrödinger's cat, Phil already existed in both an alive and dead form, but then decided it wasn't the time for quantum physics, it was time to go catch the cat. Sorry, colt.

So after this rather frustrating meander through the world of Wikipedia, I was no wiser. Still there was the question: as a punter, is there any value in trying to decipher how a race will be run?

Successful punters typically approach races by assigning numerical values to each of a participant's previous performances to gauge ability. Then they assess how likely each participant is to return similar numerical values in the race in question (based on factors such as ground and weight carried). From this they create a probability of each participant winning, usually expressed as a tissue price. They might then back those participants whose odds are larger than the tissue price, and lay those whose price is shorter.

Every step of this process requires the punter to assess inexact variables, and in doing this, the accuracy of the final tissue price is continually reduced. Successful punters are those able to limit the inaccuracy most effectively.

On this basis, isn't it better to ignore how a race will be run? Because aren't the variables (the completely unpredictable machinations of trainers and jockeys) too vast to assign any meaningful number to?