Every sportsbook makes its money the same way: a small percentage skimmed from every dollar wagered, baked silently into the odds so you don't notice it the way you'd notice a 5% transaction fee. That percentage is called hold when you're describing what the book keeps, and vig (or "juice") when you're describing what the bettor pays. Same number, different framing. This guide explains both, shows how to calculate them, breaks down typical holds by market type, and explains why parlay vig is the worst hidden tax in sports betting.
The terms get used interchangeably, which is annoying because they describe slightly different units. Both are talking about the bookmaker's profit margin on a market.
If you have to pick one to think about, hold percentage is the cleaner metric because it's a percentage rather than a cents-per-dollar that varies with bet size. But the two are arithmetically equivalent: a market with 4.55% hold takes 4.55 cents from every dollar you put through it, on average, when action is balanced. The vig framing matters mostly for bettors who think about cost in dollar terms (closer to how a credit-card transaction fee feels).
The procedure is the same one you'd use to de-vig a line:
1. Convert each outcome's posted odds to implied probability.
2. Sum the implied probabilities. The total will exceed 100% — the excess is the vig.
3. Hold = 1 − (1 / sum of implied probabilities). That's the share of every dollar handled the book keeps.
Worked example: −110 / −110 (standard NFL spread).
The 4.55% hold means: across infinite balanced action on this market, the book keeps 4.55 cents of every dollar bet by anyone. The 4.76% "vig" figure you see in some places is the sum-above-100 measure rather than the dollars-kept measure — same idea, slightly different denominator. Most professional bettors use the 4.55% hold figure because it correctly reflects what the book actually keeps, not just the over-round.
Or skip the math entirely and use our Hold Calculator or Vig Calculator — both give you the same number from different angles.
Hold isn't constant across the sportsbook. It varies dramatically by market, with the highest holds on the markets the public bets least efficiently. A rough taxonomy across major US retail books:
| Market type | Typical Hold | Why |
|---|---|---|
| Major-sport ML (NFL, NHL, NBA) | 4−5% | High volume, sharp action keeps lines tight |
| Major-sport spreads & totals | 4−5% | Same dynamic; price moves trigger juice changes |
| NCAA & lower-volume sports | 5−7% | Less sharp action; books need bigger margin |
| Player props (single line) | 8−12% | Lower volume, harder to model, books charge more |
| Alternate spreads & totals | 6−10% | Wider hold built into "non-standard" prices |
| Same-game parlays (SGPs) | 15−30% | Correlated legs, books price defensively |
| Live in-game markets | 6−10% | Books need cushion for live volatility |
| Futures (season-long) | 15−30% | Long-tail outcomes, locked capital, high uncertainty |
The pattern: holds grow as markets become less liquid, harder to model, or more correlated. Most professional bettors concentrate their volume on the lowest-hold markets (major-sport moneylines, spreads, and totals) and stay out of the fat ones unless they have an unusually sharp edge on a specific niche.
Not all books charge the same hold. The biggest single decision a bettor can make to reduce their long-run cost is which book they bet at:
The math: a bettor with a real 3% edge against the fair line will earn meaningfully different long-run profit depending on which book they bet at. At 4.5% hold, their effective edge is roughly 3% − vig-share ≈ 0.7% after the book's take. At 2.5% hold, their effective edge is 3% − vig-share ≈ 1.7% — more than double the long-run growth rate.
Single-bet vig is bad. Parlay vig is worse because it compounds. Each leg of a parlay carries its own hold, and the effective vig on the whole parlay multiplies the per-leg vigs together. A 2-leg parlay of two −110 lines:
A 4-leg parlay at the same −110 per leg has effective vig around 16−17%. A 6-leg parlay around 22−24%. Same-game parlays carry even higher effective vig (often 25−35%) because the legs are correlated and books price the correlation conservatively to protect themselves.
This is why parlays make sharp sense only when each individual leg has clear positive expected value — the compounded vig requires substantial per-leg edge to overcome. Most public parlays don't clear that bar; they're entertainment, not strategy. See the Parlay Math guide for the full breakdown.
Three practical moves:
Sportsbook hold is the share of every dollar wagered that the book keeps as profit when action is balanced across all outcomes. On a typical −110/−110 two-way market, hold is about 4.55%. It's the book's framing of the same margin bettors call vig or juice.
Vig (vigorish) or juice is the cents-per-dollar premium you pay to place a bet — the same margin as hold, framed from the bettor's perspective. On a −110/−110 market, you risk $110 to win $100, paying about $4.55 per $100 wagered as vig.
Convert each outcome to implied probability, sum them, and apply: hold = 1 − (1 / sum of implied probabilities). For −110/−110, both sides imply 52.38%, sum 104.76%, hold = 1 − 1/1.0476 = 4.55%. Our Hold Calculator does this in one click.
It varies by market and book. Major-sport moneylines at retail US books: 4−5% hold. Same markets at sharp books: 2−3%. Player props at retail books: 8−15%. Same-game parlays: 15−30%. Live in-game markets: 6−10%. The fattest holds are on the markets the public bets least efficiently.
Pinnacle and Circa are the canonical low-hold US-accessible sharp books on major markets. Among US retail books, hold varies day to day. The most practical method is using a hold calculator on the specific game and shopping for the lowest number across the books you have accounts at.
Each leg of a parlay carries its own vig (typically 4−5%), and parlay vig compounds multiplicatively. A 4-leg parlay has effective vig around 16−20% — roughly 4× a single bet. Same-game parlays carry even higher effective vig because the legs are correlated and books price the correlation conservatively.