Inside the TAB sports board, market by market
Every bookmaker takes a cut. That is not a scandal, it is the business model, the same way a currency exchange makes its money on the spread rather than on predicting currencies. What is odd is how rarely anyone says how big the cut is, or on which bets. On a $100 bet, the answer at TAB is about $7.20, and on some bets it is half as much again.
So we measured it. On 14 June 2026 we captured the entire TAB board in one pass: 146,956 individual prices. After filtering to the markets where the margin can be calculated honestly (more on that below), that leaves 15,131 markets across football, rugby league, basketball, tennis, baseball and ice hockey.
on a TAB market
on an even-money bet
in one snapshot
How the margin is calculated
Every price implies a probability: odds of 2.00 imply 50%, odds of 4.00 imply 25%. Add up the implied probabilities of every outcome in a market and a fair book would total exactly 100%. Real books total more. That excess is the cut, and it is the only thing standing between a coin-flip bettor and breaking even.
Take a real tennis match from the snapshot. Sonego against Basilashvili, priced at 1.55 and 2.30:
| Selection | TAB price | Implied chance |
|---|---|---|
| Lorenzo Sonego | 1.55 | 64.5% |
| Nikoloz Basilashvili | 2.30 | 43.5% |
| Total | 108.0% |
Those two implied chances add to 108%, not 100%. The extra 8 points are the house margin: 7.4% of everything staked.
The answer: 7.2%, and it is remarkably consistent
Across all 15,131 markets, the median margin is 7.18%. Nearly three quarters of the board, 73.1% of markets, sits between 6% and 8%.
That tight clustering is itself the finding. TAB is not pricing each market on its merits and landing wherever the numbers fall. It is applying a house cut of roughly 7% almost everywhere, then a visibly larger one to a specific group of bets sitting out at 10–12%.
What 7.2% actually costs you
Abstract percentages are easy to shrug off, so put it in the only terms that matter. A genuine 50/50, a coin flip, should pay 2.00. At the median TAB margin the book totals 107.2%, so each side is priced at 2 ÷ 1.072 = 1.866.
That means you need to win 53.6% of your bets to break even. Not to profit: to break even. Somebody genuinely, unusually good, right 53% of the time on even-money bets, which almost nobody sustains, still loses money slowly at that price.
This is the number that makes most betting systems pointless. They are not beaten by bad luck, they are beaten by arithmetic.
Where the cut is worst: the draw
Break it down by sport and the straight two-outcome bets all sit in a tight band from 6.5% to 7.1%. Then there is football, where the draw is a third thing you can back.
| Bet type | Number measured | Median cut |
|---|---|---|
| Basketball | 173 | 6.5% |
| Tennis | 304 | 6.7% |
| Baseball | 757 | 6.7% |
| Rugby league | 916 | 6.7% |
| Ice hockey | 214 | 6.9% |
| Football, two outcomes | 7,866 | 7.1% |
| Football, with the draw | 1,642 | 10.3% |
You can filter this by sport, down to the individual matches, on what TAB actually charges.
Adding the draw takes the cut from 7.1% to 10.3%, half as much again, on the single most popular bet in the world's most popular sport. A third result gives the bookmaker a third place to hide its cut, and it uses it.
If you bet football and you do not need the draw, the two-outcome version of the same match is materially cheaper.
What to do with this
Three things follow, and none of them is "stop betting".
- Look at the type of bet, not just the price. The same opinion backed as a two-outcome bet instead of a three-outcome one costs you about three percentage points before anything else happens.
- Treat the novelty bets as entertainment. The ones excluded from this study, correct score, six-packs, score-and-scorer combinations, carry cuts far above 7%. They are priced to be fun, not to be beaten.
- Judge any edge against 7.2%, not against zero. A tipster beating the market by 3% is not making you money at TAB. The cut is the bar, and it is higher than most people betting into it realise.
If the third one is new to you, what positive expected value betting is walks through it with a coin you can flip yourself.
And if you bet the races rather than the sport, this article covers the cheap half of the board. What TAB actually charges applies the same measurement to 2,537 races: the fixed odds on racing run at about 24%, roughly three times what you have just read about, and dearer than TAB's own pool on the very same race.
What this study deliberately excludes
The sum above only works when a bet's outcomes cannot both happen and cover every possibility between them. Plenty of TAB bets are not like that, and including them would have produced much bigger, much more dramatic numbers that would also have been wrong.
Double Chance is the clearest example. Its three selections, home-or-draw, home-or-away and draw-or-away, each cover two outcomes, so every outcome is counted twice. Naively summing them gives an apparent "margin" above 50%, which is an artefact of the double-counting and not a real cut. The same applies to six-packs and score-and-result combinations.
So those are excluded, along with anything where fewer than two prices were captured. What remains is every clean two-outcome bet, plus football with the draw, where the arithmetic is unambiguous.
Two honest limits on top of that. This is one snapshot from one day, and margins move with liquidity and event profile, so treat 7.2% as a typical figure rather than a constant. And a margin is what TAB charges, not what you lose: it is the headwind you bet into, not a prediction about your results.
The point of measuring it is simple. You cannot tell whether a price is good until you know what a normal price costs.
Happy betting, Seb
Why a New Zealander ended up measuring all this is on the why page.