What we still cannot see
Every racing method has holes in it. The difference between them is whether the holes are written down somewhere the reader can find.
Why publish this at all
Because a stated gap is useful and a hidden one is dangerous, and the difference shows up at exactly the wrong moment.
If you know what we cannot see, you can judge our output properly. You can tell when we are working inside our competence and when we are working near the edge of it. You can disagree with us for the right reasons. None of that is possible if we present a finished-looking picture and quietly leave out the parts that are missing.
There is also a self-interested version of the argument, and I will admit to it. Writing the gaps down stops us drifting into using data we do not have. It is very easy, over months, for an inference to harden into an assumption and then get treated as a measurement. A public list is a check on our own memory.
The big one: no sectional or tracking data pre-race
This is the gap that matters most, because it sits underneath the thing we most want to read.
Most of our work is about the shape of a race. How it is likely to be run, where the pressure comes from, which runners are being asked a question they can answer and which are not. That reading currently comes from run style history and from reading races, not from measured data about where horses actually were and how fast they were travelling at each stage.
Those are not the same thing. A run style label tells you a horse has tended to race prominently. It does not tell you what it cost that horse to get there, how hard the early part of any of those races was run, or whether prominent meant leading by two lengths or sitting third in a crawl. Sectional and tracking data answers those questions directly. We are answering them by inference.
Inference is not nothing. It is also not measurement, and we should not be allowed to blur the two just because the output looks the same on a page.
What that gap actually costs us
Three specific things, as far as we can tell.
We cannot separate intention from ability. A horse that led was either fast enough to lead or was sent forward from a position where it had no choice. Without measured early speed those look identical afterwards.
We cannot grade how hard a race was run except roughly. Overall time and the shape of the finish tell you something. They do not tell you where in the race the damage was done, which is usually the part that decided it.
We cannot settle our own pace reads cleanly. When we say a race was likely to be run a certain way and it was not, we would like to know precisely how wrong we were, in numbers, race after race. At the moment that check is coarser than it should be, and a coarse check finds only the large errors.
The other kind of blind spot
Some of what we cannot see is not missing data. It is a limit on what any amount of data will tell us.
The clearest example is one we have hit repeatedly. When we find a real effect in racing, the next question is whether it is already in the price, and the honest answer is often that we cannot fully separate the two. Draw bias is real and it died as a standalone edge on held-out races. Money moving towards a horse is real and the advantage vanished once we compared shorteners and drifters at the same final price. Strong recent form genuinely predicts and most of it is absorbed by the off. In each case the effect existed. The opportunity did not.
Then there is sample size, which is the least glamorous limit and the one that binds hardest. Anything we want to say about a narrow subset of races needs a great many of those races before it means anything. There is no clever way around that. You wait, or you fool yourself.
What it would take to close them
For the sectional and tracking gap, more than buying a feed.
It would need position and speed data for every runner, not just the ones that finished in front. It would need to be consistent across courses, because a measure that behaves differently at two tracks is a measure of the tracks. It would need to be machine readable in a way that survives blank cells and missing runs without silently turning them into zeroes, which is a genuinely common way to break this kind of work. And it would need to be captured before the off, with a timestamp, so that nothing from after the race can leak backwards into what we claim we knew beforehand.
That last condition is the strictest one and it is not negotiable. A number that arrives after the result is an explanation, not evidence, and the two are very easy to confuse when you would prefer the flattering version.
Until then
We work with what we have and we say which is which. Where the picture is inferred, it gets called inferred. Where a race does not give us enough to read, it gets a No Read rather than a confident-looking output produced to fill the gap.
An honest gap is a better thing to hand a reader than an invented detail. It is also the only version of this that stays true a year from now.
Elite Pass has no validated betting edge, does not claim one and recommends no selections. This post is a list of things we cannot do, which is roughly the opposite of a sales pitch and is published deliberately.
Related research
- We looked at 57 races and 41 got no readOn one race day the read assessed 57 races, scored 16 and refused 41. Exactly one race went the whole way. Here is what refusing a race actually involves.
- Why No Read might be the most useful thing we publishMost racing products must have an opinion on every race. Ours is allowed to say there is nothing here worth reading, and that refusal is the point.
- The bookies aren't afraid of AI. They should be rubbing their hands together.AI is becoming astonishingly good at explaining why a horse might win. That may be exactly why bookmakers have very little to fear.
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.