What two months of testing actually taught us
Written today, looking back. Less than we hoped about racing, considerably more than we expected about how to ask a question properly.
Start with the deaths
Three of the first five beliefs we tested did not survive, and it is worth being specific about what that means, because died is doing precise work in that sentence.
The draw decides it. Taken on its own as a way of beating the market, tested on held-out races the method had never seen, it returned a clean null. Not a weak result. Nothing.
Follow the money. Prices genuinely move and favourites move further than the rest of the field late on. We recorded our own timestamped snapshots rather than reconstructing them afterwards, then compared horses that had shortened against horses that had drifted at the same final price. The advantage disappeared. What looked like a signal about steamers turned out to be a fact about which price band they finished in.
Good form finds good horses. This one is true, which is what made it painful. Strong recent form genuinely predicts. Hold the final price constant and most of what it was telling us had already been absorbed by the time the stalls opened. The form was right and the price was right, which is an awkward combination if you were hoping one would beat the other.
And the survivors, such as they are
Two things held, and neither of them is an edge. Both are calibration.
Across 60,671 races with one clear favourite, the favourite won 35% and did not win 65%. Set against what the prices implied, that is close to what the market said would happen. And when we split favourites by price band, the win rate climbed as prices shortened, roughly in line with what each band was charging. We could not find a band where the horses were reliably better than the number attached to them.
Big fields make favourites more vulnerable, which is true, and the market prices it, which is the part people leave out.
So the surviving findings all say the same thing in different accents. The market is very good at the questions everybody is asking. That is not a fun conclusion to reach in month one of a research programme. It is a reliable one, and it saved us from several months of work we would otherwise have wasted.
The pattern underneath all of it
Every idea that died, died the same way, and once you have seen it three times you stop being surprised.
The observation was true. The effect was real. And it was already in the price, because it is visible to everyone, discussed in every preview, printed in every form guide. A real effect and an available opportunity are different claims, and racing conversation slides between them constantly.
That distinction is now the first thing we test, not the last. It has changed which ideas we bother to pick up. Anything that can be read straight off a racecard by a competent person is almost certainly priced, and building on it is a way of paying to agree with the market.
What changed about the question
This is the actual result of the two months, more than any individual finding.
We started with a version of the question everybody starts with. Which horse is best, and can we work it out better than the market. The answer, repeatedly and from several directions, was no, and no reason to expect otherwise.
If the best horse in the race loses about two times in three, then which one is best is a solved problem and somebody else solved it first. What remains is the race around the horse. How this particular contest is going to be run, and which runners that shape actually suits. That is not a cleverer version of the same question. It is a different question, and it is the one Elite Pass is now built around.
Whether it has an answer we can extract is genuinely undecided. I would rather say that than dress a change of direction up as a discovery.
What we learned about method
Three habits, all of them boring, all of them the reason the results above can be trusted at all.
Decide what counts as evidence before you look. Written down, in advance, where you cannot quietly adjust it once the result arrives. Almost every seductive racing finding I have ever been shown was produced by a rule chosen after the fact.
Settle everything against starting prices. Not against a feeling about the race, not against whether it looked right. Against the number that was actually available.
Keep the failures. Publicly, with the reasoning intact. A dead idea tells you where not to look next, and it is the only honest basis on which to ask anyone to take the surviving ones seriously.
Where this goes next
Forward, slowly, and without a promise attached.
The gaps we know about stay on the record. We do not have sectional or tracking data in the pre-race picture, which means our reading of pace and shape is inferred rather than measured, and that limit sits on top of the exact thing we now say matters most. Closing it is the largest single upgrade available to us and it is not close to done.
Sample size is the other constraint and there is no way to hurry it. The races that would tell us whether the shape work adds anything have not been run yet. They will be settled the same way as everything above, in public, whichever direction the answer goes.
Elite Pass has no validated betting edge, does not claim one and recommends no selections. Two months in, the most valuable thing we own is a list of things that do not work, and I am more comfortable with that than I expected to be.
Related research
- Why we timestamp our own pricesA price you read after the race has already been contaminated by the result. So we write ours down before the off, and never touch them again.
- Why we publish the ideas that failedThree of the first five beliefs we tested died. Keeping the dead ones in public is the only honest way to ask anyone to trust the ones still standing.
- The race that changed the questionA friend's horse had everything in its favour at Newmarket. The race went somewhere else entirely, and the question I left with was not about the horse.
Elite Pass publishes research, not guaranteed outcomes. Findings are measured against the market and remain subject to replication and sample size. The evidence room and the public ledger carry the full record.