Why a Second Place Is Worth Almost as Much as a Win

The form signal hiding in the placing string

Ask anyone to summarise a horse's record and you will get wins: "two from eleven". The win is the sport's unit of achievement — it decides the trophy, the headline and most people's judgement of a form line.

As evidence about the future, that convention throws away most of the signal.

The test

Build two versions of a horse's career record. The first counts only wins — the classic strike rate. The second counts placings too, weighted almost as heavily: roughly 1.2 for a win, 1.1 for a second, 1 for a third, rather than the intuitive 3 : 2 : 1.

Then ask which version better predicts the horse's next win. We ran the comparison in six jurisdictions, on samples running from sixty thousand to over six hundred thousand runners.

The place-inclusive record predicts about twice as well, everywhere. And in no market did the intuitive steep weighting — the one that treats a win as three times a show — come out best. The optimal weights are flat or nearly flat in every jurisdiction we tested.

Why the win exaggerates

A win is a noisy event. It depends on who else turned up, how the race unfolded, a head-bob on the line. Two nearly identical performances can produce a win and a third, and the winner's record shows a 1 while the other shows nothing at all — a rounding of continuous quality into a binary that discards the information in between.

A horse that keeps finishing second and third against decent fields is demonstrating ability, fitness and consistency at nearly the same evidential value as the horse that wins — the placings just happen in slightly worse traffic. Meanwhile the "wins" column can flatter a horse that beat two weak fields and does nothing else.

This is also why the hardened form student's instinct — "he's a great place bet but never wins" — needs care. Consecutive placings are not a character flaw. Statistically, they are near-wins, and near-wins predict wins.

The professional echo

The near-flat weighting is not our invention; versions of it circulate in the quantitative betting literature, where win-place-show records have long been combined at ratios close to even. What we add is scale and breadth: the claim generalises across six very different racing cultures, from Hong Kong's compact handicap system to Scandinavian racing, without an exception. Findings that survive that many borders are rare — most form patterns we test are local. This one travels.

It also fits a broader theme in our research: continuous measures beat binary ones. The same logic that makes a performance rating more useful than a finishing position makes a place-inclusive record more useful than a win count. Whenever the sport rounds quality down to a yes/no, information dies in the rounding.

How to use it

When you scan a form line, read the placing string, not the win count. A record of 0-2-3 from six is a good recent record, whatever the strike rate says. A horse with one soft win and nothing else is weaker evidence than its "winner" label suggests. And if you keep your own figures on horses, weight seconds and thirds at nearly full value — the market grew up on win counts, and conventions that everyone inherits are exactly where mispricings hide.

On our racecards the career and season summaries count wins and placings side by side, and every start carries its own performance rating — so you can judge quality directly instead of through the win column. How those ratings are made is explained here.