PaddocksEdge FAQ – Your Horse Racing Predictions Questions Answered

PaddocksEdge is a calibrated machine learning model that analyses every UK and Irish race card. It scores each runner across 669 individual data points and publishes only the selections where the model's calibrated win probability exceeds its release threshold.

PaddocksEdge produces calibrated probabilities rather than subjective tips. When the model assigns a 35% win probability, approximately 35% of similar runners win - verified against 196,000 historical runners. Every selection is published with its probability, tier, and 10:30 UK released price, and every outcome is logged on the Performance page. Hits and misses both.

The model evaluates 669 data points per runner and was trained on 18 months of UK and Irish racing: 196,000 runners across 19,600 races. It outputs a calibrated win probability for every runner, and those crossing the release threshold are published as selections. The threshold is set so that published probabilities match observed strike rates. Input categories are listed openly on the About page.

Every UK and Irish race card, every day. Flat, All-Weather and National Hunt. The model runs across the full card and publishes only the runners where the data passes the release threshold.

No. Horse racing is inherently unpredictable and no model guarantees outcomes. PaddocksEdge identifies the runners where 669 data points align most strongly, and publishes every selection alongside every outcome on the Performance page. Performance is measurable, transparent, and updated daily.

£19.99 per month after a 7-day free trial. A card is required to start the trial but no charge is taken during the trial. No contracts, no minimum term, cancel anytime.

Yes. No contracts, no minimum term. Cancel from your Account page at any time. Cancel within the 7-day trial and you pay nothing.

Selections are published from 10:30 UK each morning and refresh automatically as declarations finalise. Each selection displays the calibrated win probability, selection tier, race info, and the released price captured at 10:30 UK.

PaddocksEdge V4 was built over 18 months of data engineering and machine learning work. The team assembled a complete historical record of UK and Irish racing going back to November 2024 - 196,000 runners across 19,600 races. The model was trained on the bulk of that data and validated against held-out months to confirm performance is genuine and not the result of overfitting. The methodology is detailed on the About page.

Yes. The model is retrained every six months on the most recent data, and predicted performance is compared to actual performance. The methodology and results are published so subscribers can see how the model is evolving.

Yes. The model is continuously refined against fresh outcomes, and new analysis tools are on the roadmap. Major additions are announced to subscribers first.

Yes. The interpretation is done for you - no knowledge of machine learning, calibration, or probability theory is required. Open the app, read the selections and summaries, and use what's useful. The full 669-data-point breakdown for each selection is one tap away if you want to dig deeper.