The market moves. But does it tell you anything new?
784 races, watched from a long way out to the off. Money moves. Everybody who has ever looked at a screen knows that. The interesting part is what it means, and the honest answer took a while to arrive.
A price is not a fact, it is a process
The tempting way to think about a betting market is that it is a thermometer. You look at it and it tells you the temperature. The favourite is a certain price, that price implies a probability, and there it is.
That is not what a market is. A market is thousands of people arriving at different times with different information and different amounts of money, all of them nudging a number that already exists. It does not open at the right answer and sit there. It starts somewhere plausible and gets pushed around until the race starts and everything stops.
We took 784 races and watched that process rather than the finished product. Not to find a system in it. To find out, as a matter of description, what the shape of the thing actually is before it settles.
It converges, it does not arrive
Three things came out of it, and none of them will shock anyone who has watched a screen for a living.
The first is that the market does not arrive fully formed. The opening picture and the closing picture are recognisably related, but they are not the same picture. There is a genuine process of the field being re-sorted between the two.
The second is that favourites move more than the rest of the field does. The top of the market is where the attention goes and where the money goes, and it behaves differently from the horses further down the card as a result.
The third is the one that surprises people. Most of the movement has already happened by the time you get to the final few minutes. The last stretch before the off feels like where everything gets decided, because it is loud and it is fast and the numbers flicker. Mostly it is confirmation. The heavy lifting was done earlier.
All three are descriptions of what the market does. I am deliberately not putting figures on the size of the drift, because those numbers depend on how you sample and what you count, and I would rather you took the shape than a decimal.
Which makes a move look like information
Now the reasoning starts, and it starts well.
If prices move, and if they are more accurate at the end than at the beginning, then movement carries information. That is not superstition. It follows. Someone knew something or believed something strongly enough to act on it, the price responded, and the response is visible to anyone watching.
From there it is one small step to the conclusion the whole sport has already taken. Find the ones that are shortening. They are shortening for a reason. The reason may not be printed on the racecard, but the money is a signal about it, and the money is more honest than anything anyone will tell you.
Every part of that is defensible. I believed it for a long time and I would not call anyone naive for believing it now.
Information about what, exactly
Here is where it comes apart, and it comes apart on a question of comparison rather than a question of fact.
A horse that has shortened does win more often than a horse that has drifted. That much is true and it is the observation the whole thesis rests on. But look at what you have just compared. A horse that shortened is now a shorter price than it was. A horse that drifted is now a longer price than it was. You have compared a group of shorter-priced horses with a group of longer-priced horses and discovered that the shorter-priced ones win more often.
Of course they do. That is what a shorter price means.
So you have to do the thing that hurts. Hold the final price constant. Take horses that are all the same price at the off, and split them by where they came from: the ones that shortened into that price, and the ones that drifted into it. If movement carries anything beyond the price it produced, the shorteners should still be doing better.
They do not, much. Whatever advantage looked so obvious in the raw comparison thins right out once the final price is held still.
Why the reasoning feels airtight and still fails
Because it skips a step, and the missing step is the only one that matters.
The argument goes: the money knows something, therefore I can use it. The gap is between knows and can use. By the time a move is visible on your screen, the move has already happened. The information that caused it is not sitting there waiting for you. It has been converted into the price you are now being offered.
You are not getting the information. You are getting the receipt for it.
Anything you do next has to beat that price, not the one that existed before the money arrived. And the entire reason you noticed the horse in the first place was that the price stopped being the old one. The signal and the cost of acting on it are the same event.
This is the most common shape of failure in racing research and it barely looks like failure at all. You find a real pattern. You test it against outcomes. It works. What you have not done is ask whether the pattern is telling you anything the price has not already absorbed, which is a different test and a much less flattering one.
Why we record our own prices
A study like this only works if you know exactly what price existed at exactly what moment, which is more awkward than it sounds.
Elite Pass records its own timestamped pre-race market snapshots rather than reconstructing prices afterwards. It is slower, it is more fiddly, and it means we have far fewer races with a proper market receipt than we have races in total. We do it anyway, because a price recovered after the race has been contaminated by the result. You cannot see it fresh any more. Whatever you build on it inherits a knowledge of the outcome that nobody had at the time.
Half the mistakes in this field are not reasoning mistakes at all. They are bookkeeping mistakes wearing the clothes of a discovery.
Real is not the same as useful
Nothing above says market movement is meaningless. It plainly is not. The convergence is real, the favourite behaving differently is real, and the fact that most of it is finished before the closing few minutes is genuinely useful to know if you are deciding when to take a measurement.
What it does not appear to be is a source of advantage available to somebody watching it happen. Those are two different claims and racing has spent decades treating them as one.
Elite Pass has no validated betting edge and recommends no selections. This one went in the same drawer as most of what we test, which is now a fairly full drawer, and I would rather have a well documented null than an exciting story that falls over the first time anyone holds the price constant.
784 races · movement described qualitatively · final price held constant in the comparison
Check today’s races
See how Elite Pass reads today’s eligible races.
Related research
More notes
- What price control did to our best looking ideasI used to think the draw was worth money. Then we held the final price constant and watched idea after idea thin out. Hundreds of systems analysed, none validated.
- Everyone remembers the beaten favouriteShort-priced favourites getting turned over feel like events. The wins feel like the default. That asymmetry is memory, not maths.
- The market is very good, so what is left?Everything easy to see is already in the price. What remains is the awkward part: pace, position, course shape, and what a race will actually ask.
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. All insights.