Understanding risk

Why positive-EV 2UP bets can lose.

A positive estimate of the average result does not change how an individual bet settles. Many ordinary losses can arrive before the uncommon outcome that drives the estimate.

The qualifying loss is a real loss

A balanced 2UP back/lay match often loses a little when no full turnaround occurs. If the team wins, the back profit may be slightly smaller than the lay liability. If it draws or loses without triggering early payout, the lay winnings after commission may be slightly smaller than the back stake.

In our £100 worked example at 0% commission, those ordinary net losses are about £1.58 and £1.59. A full turnaround would produce about £308.41 net. The large possible payout does not refund earlier losses unless it actually occurs and the promotion applies.

The public calculator shows those settlement amounts. It cannot establish how often the profitable outcome will happen.

A positive average can coexist with long losing runs

Consider a deliberately simplified teaching example: every ordinary result loses £2, a full turnaround wins £198, and the assumed chance of that turnaround is 2% on each bet. These are invented numbers, not observed frequencies or a forecast.

Estimated EV per bet = 0.98 × (−£2) + 0.02 × £198 = +£2

Under those assumptions, 50 bets have an expected combined profit of £100. But a run of 50 ordinary outcomes loses £100. If the outcomes are independent and the 2% chance is correct, the probability of no turnaround in 50 bets is 0.98 to the power of 50, about 36.4%. That is just the probability of zero turnarounds, not a general formula for the probability of finishing down under other stakes or outcomes.

One full turnaround in those 50 bets would produce £198 − (49 × £2) = £100 net. Two would produce £300. The average across all possible sequences is different from a promise about the sequence you experience. There is no match count after which you are owed a turnaround.

Variance and a wrong estimate are different problems

Variance means results differ from the expected average even if the assumptions are correct. Estimation error means the expected average itself may be wrong. You cannot reliably tell which explains a short run just by looking at the profit line.

In the teaching example, the extra gain from a turnaround is £200 compared with an ordinary result. The break-even probability is £2 ÷ £200 = 1%. At a true probability of 0.5%, the average becomes −£2 + 0.005 × £200 = −£1 per bet. Repeating that bet more often would not turn it into a profitable strategy.

Where an EV estimate can fail

  • Incomplete or unrepresentative data. Historical games may not describe the current teams, competition or situation well enough.
  • A different event from the one estimated. Reaching a two-goal lead alone does not deliver the double win; the team must also fail to win and the early payout must stand.
  • Different accepted prices or fees. Moving odds, a partial lay or an incorrect commission setting can turn the intended match into a different position.
  • Promotion and settlement exceptions. Account restrictions, an ineligible market, cash-out or a void may change one side’s payout.
  • Costs outside the displayed bet. Subscription fees and any other applicable charges affect the result of using a service, even if an individual pair has positive estimated EV.

These limitations do not prove every estimate is useless. They explain why an EV number should be treated as an estimate with assumptions rather than an entitlement to profit. A higher rating addresses matching cost; it cannot remove probability uncertainty. See rating versus EV.

More bets do not always mean independent bets

Backing the same team in the same match at several bookmakers does not create several independent chances of a turnaround. Those bets share the same match outcome. Exposure can also overlap across related markets. The independence assumption used for the 36.4% illustration should not be carried over automatically to a real portfolio.

A bankroll must cover bookmaker stakes, exchange liabilities and ordinary losses while bets remain unsettled. Expected future winnings are not available cash. Raising stakes after losses increases the money exposed; it does not make the next turnaround more likely.

Review the position you actually placed

Record accepted odds, matched lay stakes, commission and settlement, then compare actual results with the assumptions you used. Keep changes in stake size separate from changes in performance. A run of wins does not establish that a model is accurate any more than a short losing run establishes that it is wrong.

The 2UP Matcher provides comparison, calculations and a profit tracker, and our free 2UP variance simulator lets you inspect 400 simulated runs before staking. Those tools help you inspect a decision; they do not guarantee returns. If losses or time spent betting are becoming difficult to control, stop and seek support rather than increasing exposure to recover them.

Track your 2UP positions and running net total.

Compare football 2UP odds with calibrated EV, built-in calculators and profit tracking — 7 days free, then £10/mo.

Start 7-day free trial

Check the actual offer before betting. Eligibility, qualifying markets, competitions, stake limits, exclusions and settlement rules vary. A price on a comparison board does not establish your account’s eligibility. Read your bookmaker’s current terms, including restrictions on promotional use and cash-out, and the exchange’s market rules.

18+ only. Gambling involves risk. Expected value is a long-term estimate, not a guarantee of profit. Prices can move, bets can go unmatched, and promotion eligibility can change.