Available in USContent
- Why operators love bet builder more than punters do
- How the correlation penalty actually works
- Leg combinations the market underpenalises
- Negatively-correlated legs where the book overcharges
- EIHL and IIHF bet builder coverage is uneven
- Stake sizing and the leg cap that protects you
- Frequently asked questions
Why operators love bet builder more than punters do
When bet365’s Global CMO Alex Sefton talks publicly about company strategy, he flags Bet Builder as central to the platform’s market position. That framing isn’t unique to one operator — every major UK sportsbook has invested heavily in same-game multi as a flagship product over the past three seasons. The reason is straightforward and not flattering to punters: bet builder generates an average overround of 12-18% per ticket, versus 4-6% on a standard moneyline. Operators love the product because the maths is in their favour by a wider margin than almost any other market.
I don’t avoid bet builder — used carefully, it has its place. But the average punter places bet builder tickets the way an amateur poker player chases inside straights. Multiple long-odds legs, no thought to correlation, expecting a 25-1 payout from a slip the book has priced at 18-1 after the correlation penalty bites. This piece is about reading bet builder properly: which leg combinations the book underprices, which ones it correctly penalises, and where the structural edge actually lives.
How the correlation penalty actually works
If you’ve ever wondered why your four-leg bet builder pays 9.50 when the four individual prices multiply to 14.20, the correlation penalty is the answer. UK books don’t simply multiply leg prices. They run a correlation model that estimates how related your selections are and discounts the combined price accordingly.
The math underneath: if two selections are positively correlated — both more likely to hit if one hits — the book treats them as overlapping rather than independent. Take “Edmonton to win” and “McDavid to score anytime.” Both more likely on nights Edmonton plays well. So pricing them as independent overstates the combined probability, and the book reduces the multiplied odds to reflect actual joint probability.
The penalty isn’t applied uniformly. Different operators run different correlation models, and the discount can range from 5% (low correlation, mostly independent legs) to 35% (heavy positive correlation, like four player props on the same team in a blowout scenario). The thing punters need to know: the penalty is always there. Even legs that look independent — moneyline on Match A plus moneyline on Match B — sometimes get marginal penalties because slate-wide outcomes share statistical structure.
The implication for ticket design: independent legs across different matches preserve more of the multiplied odds than same-match legs. But the trade-off is you’ve moved into accumulator territory and lost the appeal of building around a single match story.
Leg combinations the market underpenalises
Here’s where bet builder gets interesting for the disciplined punter. Some combinations of legs are positively correlated, but the book undercharges for the correlation. Those are the spots to attack.
The classic example: home favourite to win plus total Over the alternate line. If the home favourite wins, the match is more likely to be high-scoring — because their goals contribute to the total, and they typically score more than the opponent. The book applies a correlation penalty, but historically the penalty has been smaller than the actual joint probability would suggest. NHL match totals averaged 6.1 goals per game in 2024-25, and games where the home favourite won averaged higher.
The pattern I track: home favourite ML + Over alternate at 7 or 7.5. The standalone prices might multiply to 5.50. The bet builder price comes out at 4.50 after penalty. The fair price, given the joint probability, is around 4.80. That 0.30 gap, repeated across a season, is a structural edge.
Another positively-correlated combination: starting goaltender to record over 30 saves plus opposing team total Over 3.5. If the goaltender faces 35+ shots, his saves total goes up and the opponent’s shot volume points to a higher chance of breaking through for the team total. Books usually underpenalise this pairing.
Top forward to score plus team to win is the most-bet combination, and it’s also one of the most accurately priced. Don’t expect to find edge there. The edges live in less-obvious pairings that punters don’t naturally combine.
Negatively-correlated legs where the book overcharges
The opposite case is more useful. Some leg pairings are negatively correlated — one leg makes the other less likely — but books treat them as if independent or apply a penalty too small. The structural edge here is bigger than the positive-correlation case.
Example: home favourite to win plus Under on total. NHL home favourites went 64.1% SU in 2024-25, but the matches in which they won at a 64.1% rate had average totals slightly below the implied Over line. When a favourite wins by clamping down defensively — which a heavy ML favourite often does — totals lean Under. Books still treat these legs as roughly independent, sometimes with a small positive correlation penalty. They should treat them as negatively correlated with a small bonus.
Another: away underdog +1.5 puck line plus Over total. Underdogs covering +1.5 means they either won outright or lost by exactly one goal. Either scenario typically involves more goals than a blowout, so the combination is positively correlated — but in the right direction for an Over ticket. Books don’t always price the bonus correctly.
The third pattern: shot-suppression team to win plus total Under. Carolina, Boston, Florida and Vegas all run shot-suppression systems. When they win, totals stay low. The book usually pricesthese legs near-independently. The structural joint probability is higher than the multiplied odds would suggest after a fair correlation adjustment.
EIHL and IIHF bet builder coverage is uneven
EIHL bet builder markets exist on roughly four or five UK operators, with deeper coverage at the operators who lead on NHL props. Where available, the same correlation logic applies, but the underlying market is thinner and prices are softer.
The headline EIHL outright market in 2025-26 had Sheffield Steelers, Belfast Giants and Cardiff Devils priced at 3.50-5.00 for the regular-season title. Within a single match between two top-three clubs, the bet builder market typically lists 10-15 leg options compared to NHL’s 40-plus. Coverage tilts heavily toward team-level legs (winner, total goals, total shots) with limited player props beyond top scorers.
IIHF bet builder coverage during the May World Championship is broader — most UK operators ramp up listings for the tournament — but the same operator differential applies. For a comparison-heavy market context, the differences between accumulators and same-game bet builders piece walks through where one product fits over the other.
Stake sizing and the leg cap that protects you
Bet builder is a high-variance, high-margin product. Stake sizing needs to reflect both.
The cap I use: maximum five legs on any single bet builder ticket. Beyond five legs, two structural problems compound. First, the correlation penalty grows faster than linearly — each additional leg makes the model less reliable and the operator’s margin wider. Second, leg failure probability compounds. A five-leg bet with each leg at 65% probability has a 12% combined probability. A seven-leg version of the same ticket sits at 5%. The price multiplier doesn’t keep pace with the falling probability, so longer tickets are structurally worse value than shorter ones.
Stake sizing on bet builder: I use 50% of my standard ML unit stake. So if my unit is 1% of bankroll on a moneyline ticket, bet builder tickets are 0.5%. The smaller stake absorbs variance without forcing emotional adjustments mid-season.
Frequently asked questions
Why does my bet builder pay less than the multiplied odds suggest?
UK operators apply a correlation penalty to bet builder tickets. The model estimates how related your selections are and discounts the combined price to reflect actual joint probability rather than independent probability. Positively-correlated legs — for example, home team to win plus their top forward to score — get the biggest discount because both are more likely to hit on the same night. The penalty typically reduces the multiplied price by 5-35% depending on how related the legs are.
What’s the optimal number of legs in a hockey bet builder?
Three to five legs is the structural sweet spot. Below three legs the bet behaves more like a same-match parlay with limited correlation gains. Beyond five legs the correlation penalty grows faster than linearly and leg failure probability compounds, so the price multiplier rarely keeps pace with the falling probability. Beyond seven legs the joint probability typically falls below the implied price by a wide enough margin that even sharp leg selection can’t recover the structural cost.
