Sports is the deepest category on the venue, but reading it requires first reading the much larger market beside it: traditional bookmaking. That is where sports pricing anchors, and the way that market charges is called vig — buried in the odds rather than itemised.
The implied probabilities of a bookmaker's odds sum to more than 100%. The excess is their margin — vig (also juice, or overround).
| Side | Decimal odds | Implied probability |
|---|---|---|
| A wins | 1.91 | 52.4% |
| B wins | 1.91 | 52.4% |
| Total | — | 104.7% |
An evenly matched fixture: true probabilities should be 50% each. But betting at 1.91 returns only 1.91× when fair odds would be 2.00. The difference is the margin.
| Sportsbook | Prediction market | |
|---|---|---|
| Cost form | vig, inside the odds | fee plus spread, itemised |
| Both sides sum to | > 100% | = 100% (held there by arbitrage) |
| Your counterparty | the house | another trader |
| Winning a lot | limits, then a ban | nobody is incentivised to stop you |
| Exiting early | usually not | yes — if someone bids |
Row two is structural: on a prediction market both sides must sum to $1, because if they do not, someone arbitrages it (split and merge are always available). So there is no margin hidden inside the price — the costs are explicit.
Add up three things: the fee (computable), the spread (computable), and the counterparty knowing more (not computable).
A soft book's vig might be four or five percentage points, which looks expensive. But being repeatedly picked off by better-informed counterparties on a prediction market can cost far more. Explicit costs are easy to underestimate, because you can only see the two you can compute.
| English | 中文 | Meaning |
|---|---|---|
| moneyline | 独赢 | straight winner |
| spread / handicap | 让分 | winner with a head start |
| total / over-under | 大小球 | combined score above or below a line |
| 3-way | 三路盘 | win / draw / loss — typical in football |
On a prediction market these all become binary outcome tokens: "does A win" is one board, "is the total above X" is another. A three-way is a three-outcome complementary set — see complementary sets and NegRisk.
| Sharp | Soft | |
|---|---|---|
| Serves | professionals | retail |
| Vig | low | high |
| Limits | high | low |
| Pricing | fast and accurate, treated as consensus | slower, follows others |
| Ban winners? | relatively not | yes |
The distinction matters practically: sharp odds are treated as the best available estimate of consensus, and every serious pricing exercise starts there.
| Concept | Meaning |
|---|---|
| Closing line | the final odds before kickoff, widely accepted as closest to true probability |
| CLV (closing line value) | how much better your entry was than the closing line |
P&L is heavily polluted by luck — the variance across a few dozen fixtures is enormous. CLV measures how much better your price was than where the market ultimately settled, which does not depend on the result and therefore converges far faster.
It is the same idea as the calibration table used elsewhere here: find a measure of yourself that the variance of outcomes cannot contaminate. See price is probability.
| Pregame | In-play | |
|---|---|---|
| Pace | slow; there is time to compute | extremely fast |
| Speed requirement | low | very high |
| Where the winners' money is | almost all of it | — |
That last row is the most useful prior in this category: consistent winners' money is almost entirely pregame. In-play is a latency race, and a latency race is not a battlefield an individual wins. The measured profile is in the three winner archetypes.
A sportsbook hides its cost inside the odds; a prediction market puts it on the surface as fee plus spread. But explicit costs are the easy ones to underestimate, because the most expensive item — the counterparty knowing more — is uncomputable on both.