How to Build a Correct Score Strategy Without False Guarantees

No correct score betting strategy guarantees profit. Your choice between a single bet, dutching selected scorelines, a lay bet and trading should depend on a defensible probability, the available odds and the total money at risk. Trading also adds execution risk: the opposing bet you plan to place may not be fully matched.

This guide is for adults. All prices and stakes below are educational examples, not offers on current matches. A model can help check the arithmetic; it cannot establish that its probability estimate is correct.

Which approach should you choose?

Choose by the possible payouts and maximum loss, not simply by how many scorelines appear to be covered. Broader coverage does not automatically mean better value.

Lay the draw covers every drawn scoreline, not just 0:0. It is a different position from laying 0:0.

A 17-way market is a format with 17 options, not an automatic purchase of 17 bets. Check the named scorelines and any catchall selections covering other results. There is no universal list for all such markets.

Confirm the settlement period before calculating anything. Under Betfair's Exchange General Rules, football markets default to 90 minutes plus stoppage time; extra time and penalties are excluded unless the market states otherwise. Half-time correct score is a separate market. For interruptions or abandoned matches, check the operator's specific market rules rather than guessing the settlement.

What data do you need before choosing a score?

Estimate the teams' future scoring rates before selecting a scoreline. One recent result or a handful of head-to-head meetings is not enough for a reliable estimate.

Gather comparable data:

• home and away attacking and conceding figures;
• league averages and the strength of opponents;
• team news and likely starting lineups;
• the source, date and sample size for each figure.

Give tiny recent samples and sparse head-to-head records less weight than more stable evidence. The expected goal means for a future match are model assumptions. Shot-based Expected Goals (xG) from the previous match are not automatically the scoring means for the next one.

The full estimation process belongs in the correct score prediction guide. Here, we test supplied inputs rather than present a statistically validated forecasting model.

During a match, red cards, elapsed time and the current score change the conditions. Rerunning an unchanged pre-match model does not make it a live prediction.

How do you check probability and value in the calculator?

Enter assumed goal means in the Correct Score Calculator, then compare the selected score's probability with decimal odds. This tests expected value under those assumptions, not whether the bet will win.

To reproduce the example:

1. Select decimal odds and goals input mode.
2. Enter 1.5 for the home team's mean goals and 1.0 for the away team's.
3. Set maximum goals to 6.
4. Leave the Dixon–Coles correction off: rho = 0.
5. In the comparison fields, enter 1 in Home and 0 in Away, selecting 1:0.
6. Enter odds of 9.00.

The Home and Away comparison fields represent home and away goal counts, not Match Result (1X2) selections.

Calculator inputs: home mean goals 1.5, away mean goals 1.0, maximum goals 6 and correction off

Local calculator demonstration: means of 1.5 and 1.0, a goal range from 0 to 6, and correction off. These images do not show a current match or available offer.

For 1:0, the screen shows probability 12.3%, fair odds 8.11, edge 1.2%, and EV 0.109 per unit staked.

Calculation for 1:0 at odds 9.00: probability 12.3%, fair odds 8.11 and EV 0.109

Result for the same inputs: scoreline 1:0 and odds 9.00. The displayed figures are rounded.

What do probability, edge and EV mean?

Probability describes how often the outcome occurs within the model; expected value describes average net profit per unit staked at that probability. They are not interchangeable.

For a back bet before fees:

EV = p × K − 1

Here, EV is expected net profit per unit staked, p is the outcome probability, and K is decimal odds. Model-derived fair odds are 1 / p.

The independently checked, unrounded probability is 0.1232518783, fair odds are 8.11346, and EV is 0.1092669, approximately 10.93% of the stake. The displayed 1.2% edge is the rounded percentage-point gap between the model probability and the reciprocal of the offered odds. It is not the bet's return on investment.

These comparisons expose two weaknesses: a worse price and an inaccurate probability. A positive calculator result does not protect against the latter.

On a stake of 10, expected net profit is approximately 1.09 as a model average. A single bet does not pay that average: a win returns 90, giving net profit of 80; a loss costs 10. Do not recompute an exact result using only the rounded 12.3% display.

What are the model's limitations?

The calculation uses a Poisson distribution and assumes independent team goal counts. That is a simplification, not a verified description of every match.

The truncated 0–6 score grid captures 0.998990844 of the original probability mass, then normalises it to 100%. The residual is not a ready-made probability for an “Any Other Score” selection: that selection's contents depend on the market.

When enabled, the Dixon–Coles correction uses a fixed rho = −0.1, not a parameter fitted to this match. It is off throughout these examples.

The calculator has no live match feed, does not place trades, and has no verified prediction win accuracy established in this guide.

To calculate market margin, you need a complete, non-overlapping set of outcomes covering every result, including catchalls. The sum of inverse odds for a few selected scores is not the full market's overround. See correct score odds explained. Do not subtract margin again from p × K − 1: the offered price is already in the formula.

How do you split a budget across selected scores?

Dutching approximately equalises the return across selected mutually exclusive scorelines. It does not protect you against a result outside that set.

Take a total budget of 30 and select:

• 1:0 at 9;
• 1:1 at 8;
• 2:0 at 12.

For equal returns before rounding:

Q = 1/9 + 1/8 + 1/12 = 0.31944444

S_i = B / (K_i × Q)

Here, Q is the sum of inverse odds for the selected outcomes, B is the total budget, K_i is each outcome's odds, and S_i is its stake.

The unrounded stakes are 10.43478, 11.73913 and 7.82609. The ideal equal return is 93.913, giving net profit of 63.913 after deducting all stakes.

Under the same model—means 1.5 and 1.0, maximum goals 6, correction off—the selected scores have combined coverage of 33.8943%, with a 66.1057% miss probability. Model-derived expected net profit for the set is 1.8311 on 30, before fees, using the ideal, unrounded stake allocation. The practical stakes below, rounded to two decimal places, can produce a slightly different expected net profit.

Rounding to two decimal places slightly changes the payouts:

The rounded stakes still total 30. A positive profit when a selected score wins is not guaranteed profit: an uncovered result loses the whole budget.

Adding 1:1 at odds 8, a negative-EV bet in this model, does not improve that individual leg's value. Broader coverage is not necessarily preferable to a single with a stronger model-derived edge.

The Dutching Calculator allocates stakes; it does not estimate coverage probability. Its “sum implied probability” concerns the selected outcomes, not necessarily the full market. Some interface verdicts assume an exhaustive outcome set, so do not treat them as guarantees for an incomplete selection of scorelines.

To reproduce this example, choose total-stake mode and enter a budget of 30. Label the three outcome rows 1:0, 1:1 and 2:0, with decimal odds of 9, 8 and 12, respectively. Set commission to 0 for this bookmaker example. The calculation gives an equal return of approximately 93.91 before stake rounding; the stakes rounded to two decimal places are 10.43, 11.74 and 7.83, totalling 30. Rounding makes the actual returns slightly different. The selection is incomplete: an uncovered scoreline loses all 30.

Do neighbouring scores or combination systems create an edge?

Adding scores one goal either side of a prediction is an arbitrary coverage rule, not a source of value. List the distinct scorelines, total every stake, and assess each price.

Likewise, a “2 from 10” system creates 45 doubles. At 1 per double, total exposure is 45. If exactly two selections win at odds of 7 each, the one winning double returns 49, producing net profit of 4. With zero or one winner, the loss is 45. Different odds change the profitability condition.

Mutually exclusive correct scores from the same match cannot both win as accumulator legs. Do not calculate them as compatible predictions.

How do you calculate the risk of laying 0:0?

Budget for the liability, not merely the lay stake. Laying 0:0 for 10 at odds of 9 risks 80.

As explained in Betfair's lay betting guide:

L = (K − 1) × S

Here, L is liability, K is lay odds, and S is lay stake. In this example, (9 − 1) × 10 = 80.

If the match finishes 0:0, you lose 80. Any other score produces profit of 10 before commission, or 9.50 under an assumed 5% commission.

Let p be the probability of 0:0:

EV = (1 − p) × 9.50 − p × 80

The break-even point is:

p = 9.50 / 89.50 ≈ 10.61%

At this price and commission, the probability of a non-0:0 result must exceed approximately 89.39%. Without commission, the threshold is 88.89%. There is no universal required hit rate: profitability depends on the price-based threshold.

Historical excerpt from Betfair's official PDF showing liability of £172 = £200 × (1.86 − 1)

Historical Betfair document from 2009, not a current interface or current pricing. Page 32 of the official eBook PDF gives £172 = £200 × (1.86 − 1). File metadata records creation on 22 June 2009; the excerpt's original currency and language are preserved.

A high hit rate alone proves little. Frequent profits of 9.50 must be weighed against occasional losses of 80.

When does trading equalise profit—and when does it add risk?

Profit can be equalised across outcomes only after both sides are matched at suitable prices. Planning to close a position is not the same as closing it.

Consider this hypothetical example in the same market on one betting exchange:

• back the score for 10 at 9;
• later lay the same score for 15 at 6;
• both bets are fully matched;
• assume 5% commission on net market profit.

Here, the matched bets genuinely produce a green up: equal profit across outcomes. The basis for charging commission on net market winnings is described in section 4 of Betfair's charges. Actual commission rates vary; 5% is not universal.

For this back-to-lay approach to produce profit, the odds must fall sufficiently for the chosen stakes and fees. Adverse price drift can create a loss. A goal does not guarantee a favourable price move for every scoreline.

If you have already laid 0:0 and a confirmed valid goal is scored, 0:0 is impossible at ordinary settlement—subject to VAR reversal, voiding or other special rules. That does not mean you must close at an extreme price or that an opposing bet is guaranteed to match.

Before trading, check:

• the spread between available back and lay prices;
• the amount available at your intended price;
• matched and unmatched portions of your bets;
• market suspensions, in-play delays and VAR rules;
• voiding and settlement conditions;
• net profit or loss across every outcome after fees.

Neither a timed entry, Cash Out nor a planned stop-loss guarantees an exit. Betfair's Cash Out explanation describes limitations, not assured execution. The example above is hypothetical settled arithmetic, not a report of actual trading profit.

How do you calculate a bookmaker back bet paired with an exchange lay?

Use a separate commission calculation when the back bet is at a bookmaker and the lay is on an exchange. The 4.75 result from the same-exchange example does not apply.

Reproduce this hypothetical scenario in the Lay Bet Calculator:

1. Enter 10 in Back stake—the bookmaker stake.
2. Enter 9 in Back odds.
3. Enter 6 in Lay odds (exchange).
4. Enter 5 in Exchange commission (%).
5. Select Qualifier, meaning an ordinary bet rather than a free bet.
6. Choose Equalise.

Inputs for a bookmaker back stake of 10 at odds 9, exchange lay odds 6 and commission 5%

Hypothetical inputs: back stake 10 at 9, lay odds 6 and exchange commission 5%. These are not currently available offers.

Here, exchange commission applies to the exchange winnings, while the bookmaker bet settles separately. The equalising calculation is:

S = 10 × 9 / (6 − 0.05) = 15.12605

S is the lay stake. The calculator displays rounded model results:

• lay stake: 15.13;
• liability: 75.63;
• profit in either outcome: 4.37.

Rounded model result: lay stake 15.13, liability 75.63 and profit of 4.37 in either outcome

Rounded internal calculation for a back stake of 10 at 9, lay odds 6 and commission 5%. Recheck the actual accepted stake separately.

The rounding distinction matters. If the accepted lay stake is 15.13, liability is 75.65, and net profit is 4.35 when the score occurs and approximately 4.37 when it does not. The displayed liability was rounded from the internal, unrounded stake.

Use the actual accepted stake and applicable commission to recheck both outcomes. The screenshot's qualifying-loss display of −0 is not a promise of a risk-free hedge.

This tool calculates amounts only. It does not place or match bets, or validate available prices. Its bookmaker-plus-exchange formula cannot replace the net-market-commission calculation for two bets on the same exchange market.

How do you limit exposure and allow for losing runs?

Set a money-at-risk limit before selecting a bet, and assess related positions together. Stake and maximum loss are not identical for every approach.

Suppose your bankroll is 1000, and you choose maximum exposure of 1%, or 10, per position. This is an illustration, not a universally safe recommendation.

• A single back bet can stake 10 for exposure of 10.
• A portfolio of selected scorelines must keep all stakes combined within 10.
• A standalone lay at odds 9 can risk 10 with a lay stake of 1.25, not 10.

Multiple positions on one match share underlying events. Do not assume independence merely because you placed them separately.

With a constant hit probability of 0.12 and independent trials, the chance of losing the next ten bets consecutively is:

0.88^10 = 27.85%

For the next twenty:

0.88^20 = 7.76%

These figures concern those specific next sequences. The chance of encountering such a run somewhere in a longer history is a different calculation.

Three independent correct score selections, each with probability 0.12, all win with probability 0.1728%. That calculation does not apply to mutually exclusive scores from one match.

Positive long-run EV does not ensure realised profit or create a safe bankroll. Chasing losses and staking progressions cannot repair negative EV. See the bankroll management guide for further limit-setting principles.

Skip the bet if the probability is too uncertain, the price is poor, total exposure exceeds your limit, or the trading plan relies on an exit you cannot execute.

How do you make the process repeatable?

Start with a paper record before wagering: write down the prediction before the match, then compare it with settlement afterwards. This tests the process without relying on selective memories of winners.

For each position, record:

• match, date and data sources;
• model version, goal means and settlement period;
• estimated outcome probability;
• offered and actually matched odds;
• stakes, liabilities and total possible loss;
• commission and other applicable fees;
• planned entry and exit for a trade;
• settlement and net profit or loss.

Use this checklist before deciding:

1. Confirm the market: match period, named scores, catchalls and interruption rules.
2. Check the inputs: sources, dates, sample sizes and the basis for goal means.
3. Compare probability with price: use unrounded calculations where available and include fees.
4. Map the outcomes: selected score winning, an uncovered score, and any partially matched hedge.
5. Total the exposure: include all related positions on the match.
6. Check execution: available amounts, prices, restrictions and whether you can decline entry.
7. Record the decision in advance: bet or skip, with a specific reason.

To check calibration, group predictions with similar probabilities and compare their expected frequency with the observed hit rate. Small samples produce unstable results.

Where clearly defined for your records, calculate ROI as net profit divided by total staked. For lays, record liability separately: a return measured only against lay stake can conceal substantial exposure. Hit rate without odds, stakes and net profit is insufficient.

Comparing your matched price with the closing price can provide additional diagnostic information, but is not proof of an edge. Your next practical step is to reproduce the model example, write out every monetary outcome, and test whether the decision still makes sense with a less favourable probability estimate.