MethodologyAdvanced1 May 2026· 5 min read

What is Race Confidence in Greyhound Betting? When the Model Tells You to Walk Away

Race Confidence is a 0-100 score on every ratethat.dog race that tells you how predictable the race actually is. High = bet with conviction. Low = walk away. Here's how it's built.

What is Race Confidence?

Race Confidence is a 0-100 score that sits on every ratethat.dog racecard. It tells you how predictable the model thinks the race is — not which dog will win, but whether the race is the kind of race a top pick is likely to win in. High Race Confidence = a race the model expects to behave; low Race Confidence = a race that's likely to throw up a surprise.

Used well, Race Confidence is the on/off switch on a system. If your filters surface a top pick in a race with a confidence of 80, you back it with normal stakes. If the same filters surface a top pick in a race with a confidence of 25, you skip — the underlying race is too messy to trust the rating.

How is Race Confidence calculated?

Four inputs, weighted: race grade (35% weight), track-and-distance profile (30%), the gap between the model's top dog and the rest of the field (25%), and a handicap penalty (10%) that knocks confidence down for handicap races where the start positions distort form.

The intuition: a higher-grade race has more consistent dogs, a track-distance with a strong bias signature is more predictable, a wide composite-score gap means one dog stands out clearly, and handicap races inject extra noise. Together they produce a single number you can read in two seconds.

What counts as high Race Confidence?

Above 70 is genuinely high. The race is likely to play out in line with form — favourites and top picks have an above-average chance, and an upset would be unusual.

50-70 is normal — bet your filters as you would. Below 50 is a noisy race; results in the bottom band hit at meaningfully lower rates than top-band races, so you should either skip or downsize. Below 30 is the model openly telling you it doesn't know — those races are best left alone unless you have a very specific human-readable angle.

How do I use Race Confidence in a system?

Two practical moves. One: add a minimum Race Confidence filter to your system — "composite top pick AND race confidence ≥ 60" weeds out the worst races your filter would otherwise back into. Two: tier your stakes — full unit on race confidence ≥ 70, half-unit on 50-70, skip below 50. Most users find the second approach lifts ROI without much loss in volume.

Race Confidence pairs especially well with Hot Dogs. A composite-60+ Hot Dog in a race confidence ≥ 70 race is one of the strongest signals on the platform.

Should I never bet a low-confidence race?

There's a case for it. Low-confidence races have wider odds — the market reflects the same noise the model sees. A genuinely strong reason to back a longshot in a chaotic race can pay better than a favourite in a tight one. But that's a niche, expert play; for most bettors most of the time, treating low-confidence races as skip races is the right default.

If in doubt: confidence high, bet what you would. Confidence low, walk away unless you have a specific edge.