Analytics first, in the open

Fifteen months from a spreadsheet to this

It began in the middle of 2025 as one person trying to answer a narrow question: could you rate a field consistently enough that the ratings still meant something a month later? It ran on a laptop, one meeting at a time, and most of what it produced was wrong.

By late 2025 a handful of friends were using it before their own bets, and that changed what it had to be. A number you produce for yourself can be rough — you know what you meant by it. A number you hand to someone else has to survive them not knowing.

A few mates became a wider circle of punters, and the questions got harder. Not "what does it like today", but "how did last month go", and "why did it like that one". A spreadsheet cannot answer those.

Around December 2025 it stopped being a tool and became a site — the dark-themed version some of you will remember. That was the first version built to be used rather than operated.

Then the part that took longest. Making it public meant answering a question we had been able to avoid: how would anyone know if it was any good? Not whether we believed it — whether a stranger could check. That is what the last stretch has been about, and it is why the published record on this site starts more recently than the product does. We would rather show you a short record you can verify than a long one you have to take on faith.

Why we built it

Most racing products hand you a tip and ask you to trust it. We wanted the opposite: the numbers behind every call — speed figures, run-style, a fair price — and a market view that shows you why a runner is value, not just that it is. So we built our own multi-factor model and our own speed and pace engines, then wired them to live pricing from several independent markets. The result is a tool that finds the gap between what a runner is worth and what the market is paying, before the gates open.

What the platform does — factually

These describe how the product works — not a promise of returns. Live performance is tracked openly inside the app; predictions are analytical estimates, not guarantees.

Three pillars, no black box

Every runner is scored on ten factors, then reconciled against the real market. Here is exactly what goes in — and what each part does.

The model returns per-factor scores only — the ranking is computed by the app, deterministically, from those scores plus the market. Go deeper: how Core works · how Prime works. Australian thoroughbred racing only. Predictions are analytical estimates, not guarantees.

Our high-conviction tips, graded honestly

Every settled selection is published bet by bet — winners and losers, the price available when the tip went out, how it moved against the market's close, and where it finished. We don't restate the totals here: a number repeated on a marketing page is a number that can drift from the ledger it came from. Read the record itself, and download the CSV to recompute it yourself.

See every bet, checkable →

Where the settled sample is too small to support a strike rate or a return, the record says so rather than show a flattering number. 18+. Predictions are estimates, not guarantees.

A small, analytics-led team

VectorOdds is built and run by a small team of racing analysts and engineers who bet, model and care about doing it honestly. We're deliberately not putting stock-photo headshots or invented bios here — when we introduce the people behind it, they'll be real.

Gamble responsibly

VectorOdds is an analytics service — we don't accept wagers. We show the record honestly, we never imply guaranteed returns, and we put responsible-gambling support in front of you. 18+ only. If it stops being fun, step away: free and confidential support is available at 1800 858 858, Gambling Help Online and BetStop.