Books vs. Exchanges: Why the Vig Is So Much Lower on an Exchange

Two venues can price the exact same game, and one quietly takes twenty times the cut of the other. The reason isn’t generosity or better math — it’s market structure. A sportsbook is your counterparty and pads both sides so it wins either way; an exchange just matches bettors and skims a small fee. This lesson puts the real vig numbers side by side, does the arithmetic on what that margin costs you per bet and over a season, and stays honest about where the exchange edge is thin.

Two ways to make a market

Everything below follows from one distinction, so start there. A sportsbook is the counterparty to every bet you place. It sets both prices, pads each one, and books the difference so it profits no matter which side wins. The margin is built into both sides at once — which is exactly why a two-way line’s implied probabilities sum to more than 100%. That excess is the overround, and it’s the vig.

An exchange — Kalshi, Betfair, Polymarket — is not the house. It runs an order book that matches one bettor’s YES against another’s NO and skims a small transaction fee. There is no two-sided margin to build in, because the exchange isn’t taking a position. Order-book competition drags the best bid and the best ask toward each other until little but the fee is left between them.

Keep that framing for the whole lesson: this is a comparison of two market structures and what each one costs, never a nudge to bet one place instead of another. The cheaper cut is a fact about the plumbing, not a recommendation.

The vig, in real numbers

The structural claim is easy to check against real prices. Below are the measured two-way vig figures from our own store — the overround left on a game after both sides are added up, averaged across markets. The exchange sits first; the books follow, cheapest to fattest.

You can reproduce any row from posted prices — convert both sides to implied probability, add them, subtract one:

The headline is the ratio. The exchange’s margin is roughly one-twentieth of a typical soft book’s — 0.21% against 4.77%. Even the sharpest offshore books here (BetUS and BetOnline, ~2.4%) are more than ten times the exchange’s cut. And the fat numbers aren’t cherry-picked outliers: a standard −110/−110 line carries an overround of exactly 4.76%, so DraftKings at 4.77% and BetMGM at 4.76% are simply the market norm written down.

Vig drag: what the margin actually costs you

The vig isn’t a number the book keeps in a drawer somewhere. It’s the distance between the price you’re offered and fair, paid on every single bet. Two clean facts pin down what that distance costs.

First, how far each price sits from the truth. Spread the margin proportionally and each posted side lands about in implied-probability points above fair. A 4.77% book prices each side roughly 2.4 points worse than the truth — on both sides at once, so you overpay whichever way you bet.

Second, what one bet into a fair market costs you — the hold:

That 4.55% is exactly the familiar “you have to hit 52.38% just to break even at −110.” The two facts are linked by a factor of two: near even odds the EV of a bet changes about twice as fast as the price does, so a price that’s worse in probability costs you about in expected value.

Now watch the vig do more than shrink an edge — watch it flip the sign. Take a game whose fair line is 55% / 45%, and suppose your own honest estimate for the favorite is 57%: you’re two points sharper than the market’s fair number, a genuine positive-CLV bettor. A posted price is the fair probability scaled up by the margin, , so:

One honest note before the season math: that +3.4% is before the exchange’s own trading fee, which we handle below — the fee trims it but doesn’t flip it back negative.

A real edge, three venues, one season

Per-bet cost is the whole story; a season just compounds it. Near even odds the vig subtracts straight from your edge, so your net expected value per bet is your gross edge (versus the fair, no-vig line) minus the margin:

Which means the sign of your season is decided by the vig alone. For any genuine edge between the exchange’s 0.21% and a typical book’s 4.77%, the same bettor is positive at the exchange and negative at the book:

Bankrolls are multiplicative, so the per-bet net edge compounds. Stake a fraction each time and the log-growth over bets drifts as:

Put a concrete bettor through it. A gross edge of +2.5% over the fair line is a strong but realistic top end — real edges run 1–3% — and the same bettor faces the same 1,000-bet season at three venues, staking the same small 3% of the current bankroll each time. Only the vig changes:

The exchange bettor keeps almost the whole edge and compounds up. At a sharp offshore book — BetOnline at 2.41%, a hair above BetUS’s category-low 2.37% — the margin eats nearly all of it, leaving a bankroll that treads water. At a typical 4.77% book the same skill drifts down toward ruin. Same edge, same seeded luck — the divergence below is 100% attributable to the vig.

Watch the vig drag two bankrolls apart

One season is an anecdote — so drive it yourself. Set your own true edge, pick a real soft book, choose a season length, and watch two bankrolls that start equal and face the same seeded run of luck — the identical sequence of random draws — pull apart, differing only in the vig each one pays. The draws are shared and the vig only shifts each venue’s win threshold, so anything you see is the vig and nothing else.

The step-by-step of why is below, then the live season. Watch the net-EV readout (your edge minus the venue’s vig), and try to find the break-even edge — the smallest edge at which the soft-book curve stops trending down.

Cheap isn’t free: the exchange’s own fee

The exchange’s thin two-way margin is before its own trading fee, and an honest comparison has to fold that in. Kalshi charges a per-contract taker fee — and its shape is the exact inverse of sportsbook vig:

The term peaks at a coin-flip price and shrinks toward the tails, so the fee is heaviest near 50/50 and lightest on longshots — the opposite of a sportsbook, which shades longshots hardest (the favorite–longshot bias):

To compare it against a book’s vig, translate the fee into the same currency — probability points — through the fee-adjusted break-even. A taker buying YES at price needs the true probability to clear the price plus the fee:

So the fee adds of break-even: about 1.75 points at , but only 0.63 points at (or 0.10).

Two more wrinkles worth knowing. The maker fee runs about a quarter of the taker fee, and settlement is free — hold a contract to resolution and you never pay an exit fee, whereas trading out is taxed on both ends. And some major events carry a flat 0.25% maker fee instead; on a sub-5-cent contract that flat rate is a large fraction of the price, so treat that case on its own rather than reaching for the standard formula.

Fold the fee back into the worked example above. At a working price near the taker fee is points, which trims the exchange’s +3.4% toward roughly +1.7% — still positive. The book, meanwhile, stays negative. The bottom line is honest and unglamorous: the exchange wins on cost everywhere, decisively at the tails and only narrowly near 50/50 — not by magic, just by structure.

The structural kicker: books limit winners, exchanges don’t

For anyone who actually wins, this matters more than the vig. A sportsbook profits by being your counterparty, so a consistently winning account is a liability — and books routinely limit winners to trivial maximum stakes ($5–$25) or ban them outright. An exchange only matches bettors and skims a fee, so it is indifferent to who wins and actively wants the volume. It does not limit winning accounts.

The arithmetic is blunt. Your realized dollar profit is the sum over bets of stake times edge:

Cap the stake near zero and the profit follows, no matter how large the percentage edge:

A beaten-but-throttled bettor and a plainly losing one end up in nearly the same place. That’s why a thin vig is necessary but not sufficient: you can grind past a soft book’s 4.77% margin and still be structurally capped at pocket change, while an exchange lets a real edge scale. It’s a description of how the two structures treat a winner — not a reason to bet anywhere.

Honest caveats: thin markets, stale quotes, and no promises

The exchange’s cheaper price is real, but three things keep it honest.

Staleness. A low vig on a quote nobody is updating isn’t value — it’s a number waiting to be wrong. Tie a tight spread to a fresh timestamp before you trust it. Every number on this site carries the time it was computed and the time the venue last moved; a thin margin on a stale line is the one case where the cheap price is the trap.

Depth. A tight two-way spread with no size behind it can’t absorb a real bet. Push size into a thin book and you move the price against yourself, so the price you actually get is worse than the one posted:

Exchange markets on obscure games can be far thinner than a soft book’s posted line, so the honest price to compare is the microprice you’d actually fill at, not the top of book.

Availability and law. The exchanges aren’t uniformly reachable. Kalshi is a CFTC-regulated event-contract exchange — the lowest-risk, already-integrated venue; Polymarket is US-restricted; Betfair is US-unavailable. That’s market structure, not legal advice.

And the line that matters most: this is a calculation, not advice. Lower vig lowers the cost of participating; it does not create an edge. If you have no edge, a thin-vig venue simply loses you money more slowly — even a 0.21% margin is a slow loss, not a win.

Nothing here says exchanges guarantee a profit, or that anyone should place any bet.

Every formula here lives on the Formula Sheet for quick reference.