18+ • Bet Responsibly • Racing Alpha does not accept betsBeGambleAware.org
RacingAlpha
Next · 7h 46m19:08Worcester

The track record

Every claim on this page is checkable: every pick timestamped before the off, every result settled in public, every price archived. This is the whole record — not a highlight reel.

Avg closing line value
+18.2%

Across 243 measured picks. Positive CLV is the strongest predictor of long-run edge.

Beat the close
71%

Share of advised prices better than the closing price. Coin-flip would be ~50% — but note the best-book-vs-SP caveat below.

Picks logged before the off
100%

261 posted, 243 settled. Append-only and database-frozen — every historical edit sits on a public audit log (15 recorded).

Where the season test stands

v1.0 public card · 2026-05-20 → 19 May 2027pick 243 of ~500
Strike rate
15.6%
38W · 205L
P/L (1pt level)
37.6pt
luck band ±38pt
ROI per pick
-15.5%
at advised price
Virtual bankroll
£624.30
of £1,000 · £10/pt

At a 16% strike rate, ±38pt of pure luck is normal over this sample — the current 37.6pt is inside that band. The test runs the full season; we publish the autopsy whatever it says. Why we publish drawdowns →

Bankroll over time

v1.0luck band£1,000 start
£464£714£964£1214£1463£62420 May4 Jul19 Aug

The two ledgers · public card vs shadow challenger

v1.0 (the open heuristic model) fronts the public card. v1.1, a machine-learned value model, has run silently alongside it since day one — same rules, same public settlement, fewer picks (it only bets when it sees value against the market).

v1.0 · public card
Settled
243
15 voids
Strike rate
15.6%
38W · 205L
P&L (1pt level)
-37.57 pts
Luck band (±1σ)
±38pt
pure variance at these prices
v1.1 · shadow (ML)
Settled
194
11 voids
Strike rate
15.5%
30W · 164L
P&L (1pt level)
-5.94 pts
Luck band (±1σ)
±35pt
pure variance at these prices
Pre-registered promotion rule · published 8 Aug 2026

v1.1 takes over the public card if and only if both models reach 300+ settled picks and v1.1 leads by 25+ pointsat that moment. The rule is fixed now, in advance, so the decision can't be fudged after the fact — and if the gap flips, v1.0 simply stays.

Status: v1.0 243/300 · v1.1 194/300 · current gap +31.6pt to v1.1 — sample still building; nothing promotes until 300/300.

Pick freeze: every published pick is database-frozen — any identity change or deletion is recorded in a public audit table; settlement stamps (win/lose/void) are the only fields that change after publication. The freeze earned its keep in its first week: it caught the daily picker re-publishing same-day picks with refreshed prices. Fixed 11 Aug 2026 — the picker is now strictly append-only, affected same-day picks were restored to their first-published advice, and the whole episode is on the audit log — full disclosure with every timestamp here. Race-by-race detail for both models is on the head-to-head page.

What is closing line value?

01

We advise a price at 9am — say 9/2.

02

The market closes at 4/1. The money agreed with us — our price was value.

03

Do that consistently and profit follows the maths — win, lose or drawdown in any given month.

Caveat we volunteer: best-book-vs-SP comparisons flatter everyone who shops prices. That's why we publish the raw pairs — the full CLV ledger →

What you'd have made following the card — pick a stake and period

Level stakes on every v1.0 pick we've published.

Stake per pick
Period
v1.054 settled
£44.00
+8.1ROI
10/54 wins · 19% strike
Heuristic picker

£10 level stake on every settled pick over the last 30 days. Voids and non-runners excluded. Settled at the advised price (not SP). Past performance does not predict future results — these are tracked picks, not advice. 18+ only · Bet responsibly.

Watch the test unfold — one email a morning

The full card, yesterday's settled results, and the running scoreboard. Free, unsubscribe any time.

18+ · Tips are opinions, not advice · BeGambleAware.org