Myth-busting: 2000 Guineas is the best Derby trial

"This year’s Guineas winner Dawn Approach has some illustrious hoofprints to fill, but this “myth” provides encouragement that he can do so, for it is not really a myth at all..."

Simon Rowlands analyses the recognised Derby Trials in an attempt to discover which provides the best pointers for Epsom...

This saying is not as popular as it once was - indeed, it took no small amount of Googling to relocate it - but the idea that the best trial for the second colts' classic is the first colts' classic has a certain ring of truth to it. 

Against that, the Guineas is run at a mile at Newmarket, while the Derby is run at just over a mile and a half at Epsom. It is asking something of a young horse to excel at both in the space of less than a month. 

As with many a "myth", its veracity depends in no small part on how you look at the evidence. Judged on winners alone, the Guineas is equal on four with the Dante for three-year-old "trials" this century, though in New Approach's case he ran in the Irish Guineas between defeat at Newmarket and success at Epsom. 

If we take a horse's most recent run prior to the Derby as its definitive trial for that race, the Dante is alone at the top, but from 28 representatives this century against 18 for the Guineas. The strike-rate for horses going to Epsom straight from the Guineas is higher than the same for the Dante, in other words. 

Regular readers of this blog will know that wins, and wins alone, can be a crude measure, and that strike-rates are often little better: the former disregards the wealth of nuanced information from the majority of horses that do not win, while the latter fails to allow for opportunity, such as reflected in field size. 

One way to look at this issue is to consider the percentage of rivals beaten by horses in a given trial and then the percentage of rivals beaten by that cohort in the Derby itself. Percentage of rivals beaten reflects, for instance, that it is better to finish second of 15 than second of five. 

There is one slight drawback with this approach, and that is our old friend Regression To The Mean, which dictates that winners of trials will be punished compared to horses out the back. If you have beaten 100% of your rivals in a trial, you can at best match that in the Derby itself; conversely, if you have finished plum last in a trial you can do no worse (and may do better) in the big race itself. There is a tendency for extremes to head towards the middle. 

An adjustment for this phenomenon alters the figures somewhat, but not the ranking, and has the Guineas as top on a factor of 0.78 (a horse that had beaten 50% of its rivals in the Guineas could be expected to beat 39% of its rivals in the Derby), followed by Leopardstown's Derrinstown Stud Derby Trial on 0.76, the Dante on 0.74 and the Chester Vase on 0.68. 

One interesting consequence of this is that it implies finishing second in this year's Guineas (beating 91.7% of rivals, with an expectation of beating 71.5% at Epsom) would be better than winning the Chester Vase (expectation of beating 68% of rivals at Epsom). 

The figures for all trials (latest run only) which produced five or more runners in the Derby this century are shown here: 

By this measure the Guineas is indeed the best trial for the Derby, but only just. By other measures, it could be argued that it is not, but only just. 

This year's Guineas winner Dawn Approach has some illustrious hoofprints to fill, but this "myth" provides encouragement that he can do so, for it is not really a myth at all. 

While the figures quoted are primarily descriptive of past events, I will, for interest, calculate expected % rivals beaten when all the named trials have been run and post those expectations in the comments section below.

Get open access to Timeform data for just £2.50 a day with Timeform Race Passes. It's like a Form Book, Black Book & Race Card all in one! Find Out More!

Product reviews and advice for best reference

Myth-busting: 'It's not value that matters- it's winners'

The exercise of betting is in effect an extended experiment in your ability to judge value itself...

Simon Rowlands returns to challenge the misconceptions surrounding 'value' in gambling... 

Here's a quick myth-buster in a busy week, involving no spreadsheets or tables. To bust this myth, you do not need evidence, just common sense. As you have found your way to this blog, I trust the latter can be taken for granted. 

"Value", in betting terms, is the concept that a horse - or event, or whatever - is offered at odds greater than its true probability of occurring (or the opposite if you are laying it). The theory is that all you need to do is identify "value", and wait for enough events to take place to nullify chance, and you should be quids in. 

The contra-argument is that you can identify as many 10/1-shots that should be 7/1 as you like, but if none of them wins then your "value" has been worthless.

This betrays a basic misunderstanding of the concept of value itself. If, as in the above example, the true odds of an event occurring are 7/1, then an event will occur roughly one time in eight if repeated often enough. If what you thought was a 7/1 chance happens once in a blue moon then it was almost certainly not a 7/1 chance in the first place. 

The tricky thing with betting on sporting events is that you seldom know what the true odds are of an event occurring. This is unlike, say, a fair coin toss, in which you know that there is a 50-50 chance of each of the two possible outcomes.

A further mistake, and one that seems to be inextricably linked with the "all that matters is winners" school of thought, is to imagine that there is only one possible outcome of an uncertain event, and all you need to do is to identify it. This - ahem, long-word alert - teleological argument (the idea that an outcome is predetermined and therefore explains the phenomena that gave rise to it) has no place in the nuanced modern world of uncertainty. 

We should think in terms of the probability of something uncertain happening and not assume that because it happened it was somehow inevitable. 

The uncertainty of sporting events does not negate the concept of value - you should still, in the long run, back a horse that is favourably priced - but it does mean that the exercise of betting is in effect an extended experiment in your ability to judge value itself.
The "it's only winners that matter" brigade miss another crucial point: namely, that, unless you back only winners or only losers, the odds at which you back the winners matters. 

With the exception of arbing and similar forms of trading, none of us backs only winners, or only losers, though it may sometimes seem as if the latter is the case! For those who back a mixture of winners and losers - you and me, in other words - we need the rewards from the winners to offset the losses of the losers.

If you are not achieving that over a significant length of time then you are almost certainly not getting value. The problem is yours, dear reader: it is not that value does not exist.
As something of a postscript, it is as well to acknowledge that "value", like many great concepts, can get a bad reputation in the wrong hands.

In particular, those - you know the ones I mean - who constantly harp on about value just because a horse is at long odds are doing the term a disservice. 

Value can exist at very short odds as well as at very long odds, and at everywhere in between. That is why describing every 16/1+ shot as "value" is misjudged and why the blanket statement "don't back at odds on" is also folly. 

But the latter is possibly a myth-busting blog for another day! 

Get open access to Timeform data for just £2.50 a day with Timeform Race Passes. It's like a Form Book, Black Book & Race Card all in one! Find Out More!

Product reviews and advice for best reference