Why Veredus exists
Veredus started as a private research project: an automated pipeline ingesting UK & Irish results, hundreds of derived calculations, years of historical performance normalised into one consistent dataset — and a long series of experiments with machine-learning approaches on top of it.
Those experiments taught us the lesson the hard way: models come and go, but the durable asset is the data underneath — if it's engineered honestly. Point-in-time correct, leakage-free, every sentinel documented. So that became the product. We don't sell predictions, ours or anyone's. We sell the dataset we wished we could have bought.
Today that dataset covers 530,000+ UK & Irish races back to 1988, with 171 point-in-time features computed per runner and recomputed daily before racing. Explore it in the browser, read the field catalogue, or start with the free, ODbL-licensed measured course dataset.