Loading...
model_performance

Model Check: How the Predictions Performed This Week

Saturday, 27 June 2026

Seven days of racing, 687 races, and a first-pick win rate of 25.5% across the board. The any-top-3 win rate sits at 62%, meaning in nearly two thirds of all races this week one of the three model picks won. Place coverage from the top three reached 87.5% of all races. Those are the headline numbers for the week ending Friday.

Saturday the 21st was the standout day. From 120 races the model landed 39 first-pick winners, a single-day rate of 32.5% that is comfortably ahead of the weekly average. June 20th was the biggest card, 238 races, and the model held up well with 63 wins at 26.5%. Large cards are the real test of consistency because the field quality drops and upsets become more common, so holding above 25% on a 238-race day is a meaningful result.

The toughest day was Sunday the 22nd, where 30 wins from 152 races produced a first-pick rate of 19.7%. Sunday cards can be erratic, mixing low-grade trial races with weekend competitive cards, and the model finds those conditions harder to read. Thursday the 26th recovered strongly with 20 wins from 70 races at 28.6%.

The model does not aim to be right every time. A 25.5% win rate on a flat-stake basis against a random 16.7% baseline (six runners per race) represents genuine predictive value. The transparency on how these numbers are built is available in the historic results section, which logs every prediction made before the race goes off, not after.

This article was generated by RateThat.Dog's editorial engine, combining form analysis, pace profiles, trap bias data, trainer statistics, and deep reasoning models. Visit ratethat.dog for full racecards, speed ratings, and live results.