Learn EdgePro

01 Expected Value (EV) #

Expected Value is the average profit (or loss) you would expect per £1 staked if you placed the same bet thousands of times. A bet at +5% EV means that, according to our model, the bookmaker's odds are priced about 5% too generously relative to the true probability — so over a long sample of similar bets, the data suggests you would gain roughly 5p per £1 staked.

Important context. EV is a long-run statistical property. A single +5% EV bet can absolutely lose — variance is the rule, not the exception. EV only matters when applied consistently to a large sample of independent picks with disciplined staking. A single winning streak doesn't prove the edge exists; a single losing day doesn't prove it doesn't.

02 Kelly staking #

The Kelly Criterion is a formula for sizing each bet in proportion to your perceived edge and your bankroll. Full Kelly is mathematically optimal for long-run bankroll growth but is extremely aggressive — drawdowns of 50%+ are common. Most professionals use a fractional Kelly to smooth variance.

Quarter Kelly means staking 25% of the full Kelly recommendation — significantly less volatile while still capturing most of the growth. EdgePro defaults to quarter Kelly and caps any single stake at 2% of bankroll, regardless of what the formula suggests. The calculator on every signal does this maths for you using your current bankroll figure.

03 De-vigging odds #

Bookmaker odds always include a margin called the overround (or "vig"). If every horse's implied probability is added up, the total exceeds 100% — typically 102% to 115% depending on the field size and the firm. That excess is the bookmaker's built-in edge.

De-vigging removes that margin proportionally so we can compare the bookmaker's view of "true" probability with our model's. It's the foundation of EV calculation: without it, every bet would appear to have negative expected value. EdgePro de-vigs every market before scoring, using either proportional or power methods depending on race type.

04 Place probability (Harville) #

Most bookmakers don't display separate place odds — they're implied by the each-way terms. To work out the fair value of a place bet we need to know the probability that a horse finishes in the places, not just wins.

The Harville method computes this from win probabilities alone. It assumes the second-place horse is drawn from the remaining field in proportion to their relative win chances, then the third, and so on. It's not perfect — it slightly underweights outsiders — but it's the industry standard, and we apply small corrections for known field-size biases.

This is what lets the each-way engine surface place value when no place market is quoted.

05 Each-way betting #

An each-way (e/w) bet is two equal stakes: one on the horse to win, one on it to place. UK place terms depend on field size and race type:

2-4 runners: win only · 5-7: 2 places, 1/4 odds · 8-15: 3 places, 1/5 odds · 16+ handicaps: 4 places, 1/4 odds.

The data suggests 10/1 and longer is the sweet spot for e/w value, because the place portion's payout (odds × 1/4 or 1/5) becomes meaningful while bookmakers' place probability estimates are weakest on outsiders.

Qualifying loss math: If you bet £2 e/w (£4 total) on a 12.0 horse with 3 places at 1/5, and it finishes 2nd, you lose the £2 win stake but the £2 place stake returns at 12.0 × 1/5 = 2.4, giving £4.80 back — a £0.80 profit on £4 outlay.

06 Lay betting #

Laying a horse means betting against it — you act as the bookmaker. On an exchange like Betfair, you accept someone else's back bet. If the horse loses, you keep their stake. If it wins, you pay their winnings (your liability).

Laying a 4.0 shot to £10 means you pocket £10 if it loses but owe £30 if it wins. That asymmetry is why bankroll discipline matters more when laying than when backing — a single losing lay can wipe out many winners.

Commission is also critical. Exchanges typically charge 2% to 5% on net winnings per market. A 6.0 lay that wins £100 returns roughly £95-£98 after commission. EdgePro factors commission into every lay EV calculation.

07 Pace map reading #

Race shape — who leads, who tracks, who closes — has a huge influence on outcomes. EdgePro categorises projected shapes into a few key tags:

LONE_LEADER means one horse is the only natural front-runner; uncontested fractions historically favour them, especially on stiff, galloping tracks. CONTESTED_PACE means two or more habitual leaders — they tend to burn each other off, which historically advantages closers. NO_PACE means a slow, tactical race where the held-up horses struggle to make ground.

Draw bias is overlaid on top. At Chester 5f, low draws have historically struck far more than high; at York 7f straight, the bias is reversed. The pace map combines both into a single Recommendation per runner.

08 Drawdown #

Drawdown is the percentage decline from your bankroll's recent peak. A reading of −3% intraday means you're down 3% from your highest balance today. Drawdowns are an unavoidable feature of any betting strategy with variance — even a +5% EV portfolio will see double-digit drawdowns periodically.

EdgePro includes an automatic brake that pauses new staking when intraday drawdown crosses a threshold you configure (default 4%). This isn't because the strategy stopped working — it's a behavioural circuit-breaker to prevent tilt-driven decisions.

Healthy variance vs systemic problem. A −10% week with normal bet volume and EV-appropriate stakes is uncomfortable but expected. A −10% week with model EV consistently below historical baseline, or with persistent CLV drift, is a signal to investigate.

09 Steam moves #

A "steam move" is a sharp, sudden contraction of a horse's price driven by heavy volume — historically associated with informed money entering the market. We distinguish two flavours:

Fundamental steam: the move is driven by new information — confirmed going change, jockey switch, withdrawal of a rival — and the new price reflects the new reality. Technical steam: the move is driven purely by money flow with no public news, which historically correlates more strongly with stable connections, sharp syndicates or insider activity.

Why the distinction matters: sharp money (consistent, large, early) has historically been more informative than recreational money (small, late, headline-driven). EdgePro's steam engine weights moves by volume profile and timing, not just price delta.