What a figure cannot say
This page is the one that keeps the rest honest. A figure is a compression of something that already happened, and there are three things no compression can do: it cannot price the next race, it cannot rank anything without its scale, and it cannot be right more often than the thing it summarises allows.
- Sample A field
- 8 runners
- Highest-rated finished
- 3rd
- Winner was rated
- 4th of 8
- Top-rated win rate in the sample
- 3 of 8
Three limits, each with a number attached. None of them is an argument against using a figure; they are what a figure can be asked.
First limit: it cannot price the next race
A price is a number somebody is willing to take the other side of, and it already contains every figure its author can see plus everything else: the going report, the draw, the stable's form, the field, the money and the margin. A figure compared with a price is a comparison between a summary of the past and a forecast of the future, and where they disagree the price is not necessarily wrong.
Thirty-seven and a half per cent is not a small number, and it is not a prediction. It is what the sample's own figure achieved across eight invented races, and eight races is not enough to say anything at all about the rate. It is printed here because leaving it out would be the more dishonest choice.
Second limit: it cannot rank across scales
Sample C's figures separate by 7.0 points on a scale of ten points a second. Sample H's single runner moves 16.9 points across six of its own runs. Put a figure from one scale beside a figure from another and the comparison is between two publishers' decisions rather than between two runners, and the ordering may well survive while every number in it changes.
Third limit: it cannot exceed its own evidence
Six runs is a form line, and six runs of one runner in sample H spread 16.9 points against a 7.0 point field margin in sample C. A comparison between two runners built on records that noisy is a comparison of two averages whose own variation is larger than the gap being compared. That is not a reason to stop reading figures; it is the reason a figure is a starting question rather than an answer.
- Ask what question a figure is being used to answer, and whether it was computed for that question.
- Never compare two numbers from two publishers without saying whose scale each is on.
- Treat any single race as unable to confirm or refute a rate.
- Prefer the published spread to the published number whenever both are available.
- Read a price as a forecast and a figure as a summary; they are different kinds of claim.