Here is the scenario every bettor knows. You back Man City at home before kickoff. Fair odds. Strong form. Solid numbers. Ten minutes in, Arsenal score. Twenty minutes in, City are a man down. Your 72% pick is now a 15% pick — and you are stuck watching it play out, unable to do anything about it.
That gap between "my bet is going badly" and "there is nothing I can do about it" is the most frustrating moment in betting. It is also the moment we built the mid-match hedge engine to fix.
What the Engine Actually Does
The hedge engine is a second model that runs alongside every active pre-match bet we place. It re-evaluates the win probability every 5 minutes during live play. When the probability of your original pick winning drops sharply — far enough that the math says "this ticket is almost certainly cooked" — it flags a counter-bet that you can place to claw back part of the loss.
It is not about magic, and it is not about exiting every bet that goes slightly wrong. It is about one specific situation: your bet has collapsed mid-match, the odds on the opposite outcome have moved in your favour, and the cheapest path forward is to hedge.
The Trigger Rules
The engine only fires when three conditions line up together. This is deliberate — a noisy hedge engine that fires on every 10% probability wobble would kill your bankroll with transaction costs. The calibrated thresholds are:
- Probability has dropped by 65% or more from the pre-match estimate. If your bet was priced at 70% pre-match and is now at 25%, that is roughly a 64% drop — right at the edge. A 70% pre-match bet at 20% current is clearly in the danger zone.
- Current probability is below 25%. This is the absolute floor. Without this, a pick that moves from 99% to 50% would trigger (a 49% drop of a 99% base is a 50% relative drop). We do not want to hedge picks that are still favourites.
- The match is between minute 60 and 85. Before minute 60 there is still too much match left — probabilities recover. After minute 85 the hedge odds have collapsed and there is nothing meaningful to buy.
All three must be true at the same time. Hit two out of three and we wait.
The Counter-Bet Math
When the engine fires, it does not tell you to exit — it tells you what to bet on the opposite side to minimize the expected loss. The formula is straightforward:
counter_stake = original_stake × original_odds ÷ counter_odds
Say you backed City at 2.50 for 3 units. At minute 70, City are 0-2 down and ten-man, counter-odds on Arsenal are now 1.40. The engine flags a counter-bet of:
3 × 2.50 ÷ 1.40 = 5.36 units on Arsenal
Place both bets and your payout is locked regardless of the final score. If City somehow win, you collect 3 × 1.50 = 4.5 units on the original and lose 5.36 on the counter, net −0.86u. If Arsenal close it out, you lose the original 3u but collect 5.36 × 0.40 = 2.14u on the counter, net −0.86u.
Without the hedge, your expected outcome was −3u × P(Arsenal win). With the hedge, it is a fixed −0.86u. On picks the engine fires, the expected loss reduction is typically 30-60% of the original stake.
Only for 1X2 and DNB — And Here Is Why
The engine currently hedges two markets: match result (1X2) and Draw No Bet (DNB). That is not an oversight. We tested it on every goals market we cover, and here is what the data says:
- Over/Under, BTTS: Goals markets do not swing hard enough during a match. An Over 2.5 bet at 2-0 in the 70th minute is not a hedge candidate — a third goal is still likely. Probabilities rarely cross the 65% drop threshold.
- Asian Handicap: The draw gap on AH bets at half-lines makes hedging structurally bad. You often end up losing on both sides.
- Clean Sheet, Win-to-Nil, Team Goals: These resolve binary on late-game events. The counter-bets rarely exist at reasonable prices, and even when they do, the engine cannot maintain the ledger cleanly.
So we hedge where the math works and stay out where it does not. If we find a new market with clean hedge mechanics, we will add it. We will not add one for the sake of feature parity.
What the Data Shows
Every hedge recommendation the engine has fired is logged to our database, along with whether the counter-bet won, what it returned, and what the original would have paid. You can see the rolling 30-day totals on our public track record page — hedges fired, counter-bet wins, realised savings in units, and average loss reduction percentage.
These numbers are not hand-picked. They are computed directly from settled hedges and refreshed every few minutes. When the engine has a rough week, the numbers go down. When it has a good week, they go up. Nothing is hidden.
What It Is Not
A few things to be clear about:
- It is not an auto-exit. The engine flags a counter-bet. It does not place it for you, and it does not close the original. You make the call.
- It does not fire on every losing bet. Most losing picks lose slowly — a 1-0 at the 80th minute on an Over 2.5 bet, a 0-0 on a BTTS Yes. Those resolve naturally. The engine only activates when a pick collapses fast.
- It will sometimes cost money. If the engine fires and the original somehow wins, you end up slightly worse off than if you had held. That is the price of insurance. On average — and this is what the track record measures — the savings on losing picks outweigh the cost on winners.
Why We Built It
A good bettor with a year of experience will tell you: the single most valuable discipline in this game is position sizing and loss management. The Kelly Criterion gives us the first half. The hedge engine gives us the second.
Pre-match models are good — ours predicts at 55-58% accuracy across the top European leagues. But the other 42-45% of the time, the match will swing against us, and for the worst subset of those, it will swing hard and fast. The engine is built for that subset and only that subset.
If you are a Pro subscriber, you already see hedge alerts live on the match detail page and in Telegram. If you are on the free tier, you can see the settled hedge track record on /performance — every fire, every save, every miss.