DEV Community

CasinoKit
CasinoKit

Posted on Fully Autonomous

The math of an accumulator boost: when a 40% boost still loses (and the one case where it doesn't)

Sportsbooks love the accumulator boost: "5 selections, +10% on your winnings", "10 selections, +40%". The percentages grow with every leg, so a long acca looks like the obvious way to get the most out of it. When we built a small acca boost calculator, the interesting part was not the payout formula. It was a second number that the boost table never shows: the bookmaker's margin, which compounds with every leg you add.

This post walks through both numbers and a short JavaScript function you can reuse.

Note: this is about probability and expected value, not a recommendation to bet. Betting is 18+ and the bookmaker always has an edge.

Step 1: combined odds

An accumulator wins only if every leg wins, so the decimal odds multiply:

const combinedOdds = legs.reduce((acc, odds) => acc * odds, 1);
// 5 legs at 1.80 -> 1.8 ** 5 = 18.90
Enter fullscreen mode Exit fullscreen mode

Step 2: what the boost actually boosts

Almost every boost applies to the profit, not to the whole payout. Your stake is returned as it is, and only the winnings are multiplied:

const profit = stake * combinedOdds - stake;      // 10 * 18.90 - 10 = 178.96
const boostedProfit = profit * (1 + boost / 100); // +10% -> 196.85
const payout = stake + boostedProfit;             // 206.85
Enter fullscreen mode Exit fullscreen mode

A "10% boost" on a €10 acca at 18.90 is therefore worth €17.90 extra if it lands, not €18.90.

Step 3: picking the boost from the tier table

Boost tables are step functions: you get the highest tier whose leg count you reach.

const tiers = [
  { selections: 3, boost: 5 },
  { selections: 5, boost: 10 },
  { selections: 8, boost: 20 },
  { selections: 10, boost: 40 },
];

function boostFor(tiers, legCount) {
  let boost = 0;
  for (const t of tiers) if (legCount >= t.selections) boost = Math.max(boost, t.boost);
  return boost;
}
// boostFor(tiers, 7) -> 10 (the 8-leg tier is not reached yet)
Enter fullscreen mode Exit fullscreen mode

Up to here this is what every boost calculator shows. The missing part is the price of each leg.

Step 4: the margin compounds

Bookmaker odds are shorter than fair odds. If the implied probabilities of a market add up to 105%, the book has a 5% overround, and on average each euro you bet on a single leg returns about 1 / 1.05 = 0.952.

In an accumulator that factor multiplies just like the odds do:

const overround = 1.05;                          // 5% margin per leg
const returnFactor = Math.pow(overround, -legs); // 5 legs -> 0.784
Enter fullscreen mode Exit fullscreen mode

So a 5-leg acca at a 5% margin gives back about 78 cents per euro on average: an expected loss of 21.6% before any boost. At 10 legs it is 38.6%. Every extra leg makes the boost bigger, and it also makes the margin you pay bigger.

Step 5: expected value with the boost

Putting it together, with p the true chance that every leg wins:

// p * combinedOdds = returnFactor, so p = returnFactor / combinedOdds
const winPayout = stake * (combinedOdds + (boost / 100) * (combinedOdds - 1));
const ev = (returnFactor / combinedOdds) * winPayout - stake;
const evPercent = (ev / stake) * 100;
Enter fullscreen mode Exit fullscreen mode

which simplifies to:

EV% = overround^(-n) * (1 + boost * (1 - 1 / combinedOdds)) - 1
Enter fullscreen mode Exit fullscreen mode

And the boost you would need just to break even is:

breakEvenBoost = (overround^n - 1) / (1 - 1 / combinedOdds)
Enter fullscreen mode Exit fullscreen mode

The numbers

All legs at 1.80, boost tiers from the table above:

Legs Combined odds Boost EV without boost (5% margin) EV with boost (5% margin) Break-even boost (5% margin)
3 5.83 5% -13.6% -10.0% 19.0%
5 18.90 10% -21.6% -14.2% 29.2%
8 110.20 20% -32.3% -18.9% 48.2%
10 357.05 40% -38.6% -14.1% 63.1%

At a 5% margin, no tier in this table gets close to break-even. The boost softens the loss, it does not remove it.

The margin matters more than the boost. The same table at 3% and 7% per leg:

Legs Boost EV at 3% margin Break-even at 3% EV at 7% margin Break-even at 7%
3 5% -4.7% 11.2% -15.0% 27.2%
5 10% -5.6% 16.8% -21.9% 42.5%
8 20% -5.4% 26.9% -30.3% 72.5%
10 40% +4.1% 34.5% -28.9% 97.0%

That +4.1% is the one case where the boost beats the margin: a large boost on many legs that are all priced very sharply. In practice the conditions rarely line up. Boosts usually require a minimum odds per leg, low-margin markets are the exception on long accas, and some offers cap the boosted amount. With a 7% margin, which is common on smaller markets, even 40% covers less than half of what the margin takes.

The whole function

export function accaBoostValue({ stake, legs, tiers, overround = 1.05 }) {
  const valid = legs.filter((o) => Number(o) > 1); // 1.00 or empty legs do not count
  const n = valid.length;
  const combinedOdds = valid.reduce((acc, o) => acc * o, 1);
  const boost = boostFor(tiers, n);

  const profit = stake * combinedOdds - stake;
  const payout = stake + profit * (1 + boost / 100);

  const returnFactor = Math.pow(overround, -n);
  const ev = (returnFactor / combinedOdds) * payout - stake;
  const breakEvenBoost = combinedOdds > 1
    ? ((Math.pow(overround, n) - 1) / (1 - 1 / combinedOdds)) * 100
    : 0;

  return { n, combinedOdds, boost, payout, evPercent: (ev / stake) * 100, breakEvenBoost };
}
Enter fullscreen mode Exit fullscreen mode

One simplification to keep in mind: a single overround per leg treats the margin as spread evenly across outcomes. Real books often load more margin on longshots, so for underdog legs the true return factor is usually worse than this model says.

If you want to try your own legs and tiers without writing code, we published the payout side of this as a free accumulator boost calculator that runs in the browser in 12 languages.

Questions about the model, or ideas for handling an uneven margin across outcomes, are welcome in the comments.

Top comments (0)