How Mandarin finds an edge

Sportsbooks aren't oracles — they're businesses with a fee baked into every price. Mandarin does one thing: work out what the real odds are, then find books paying more than they should.

The one idea behind all of it

Every price you see has the book's fee — the vig — baked in. Strip it out of the sharpest book's line and you get the true odds. Then it's just comparison.

1.Read the sharpest line
Pinnacle is the book professionals bet into. It takes sharp money and moves fast, which makes its prices the closest thing to the real odds.
2.Cut out the vig
We remove the fee from that line to get the fair probability — what the outcome is actually worth, with no house cut.
3.Find who's paying too much
We compare every book you can bet at against that fair probability. Anyone paying more than the true odds is handing you an edge.

A real example

This is exactly the math the app runs, on every game, every minute.

1. Pinnacle prices a fight −110 / −110.

Those odds imply 52.4% + 52.4% = 104.8%. It can't be 104.8% — that extra 4.8% is the vig.

2. Strip the vig → the true odds are 50% / 50%.

A genuine coin flip. That's our fair probability.

3. DraftKings is paying +110 on one fighter.

+110 returns $2.10 per $1. But the true chance is 50%.

50% × $2.10 = $1.05 → +5% EV

For every $100 you put on that bet you'd expect ~$105 back on average, over many bets. This single fight is still a coin flip — you can absolutely lose it. The edge only shows up over volume.

The two things Mandarin shows you

They are not the same, and the difference matters a lot.

Arbitrage

ARB

Two books disagree so badly you can back both sides and profit no matter who wins. Genuinely guaranteed — and genuinely rare. Most days the board shows zero, and that's the tool being honest, not broken.

Positive EV

+EV

One book is paying more than the true odds. You bet one side and can lose it. But make these repeatedly and the math works in your favour. This is the engine — it's what you'll see most days.

	Arbitrage	Positive EV
How often	Rare — often none	Several most days
You bet	Both sides	One side
This bet	Profit guaranteed	Can lose
Long run	Profit locked now	Profits over volume
Main risk	A line moves before you place leg 2	Variance — you need many bets

Every word on the board

The scanner is dense on purpose. Here's what each term means.

Vig (juice): The book's built-in fee. It's why a coin-flip game is priced -110/-110 instead of +100/+100 — the odds add up to more than 100%, and the extra is the house's cut.
De-vig: Stripping that fee back out of a book's prices to reveal what it really thinks the odds are. -110/-110 de-vigs to a clean 50%/50%.
Fair probability: The de-vigged, true chance of an outcome — our best estimate of reality. Everything else is measured against this number.
Fair value (anchor): Whose line we trust to define fair probability. We prefer Pinnacle — the book sharp bettors use, which prices most accurately. If it isn't available we blend at least 3 books into a consensus.
EV% (expected value): How much you'd expect to make per dollar, on average, over many identical bets. +5% EV means $100 staked is worth ~$105 in the long run — not on any single bet.
Edge / gap to arbitrage: For arbitrage rows: how much guaranteed profit is locked in. A negative number is the gap — how far the market still is from a true arb.
Confidence (0–100): A data-quality score, NOT the chance your bet wins. It rises with more books quoting, tighter agreement between them, a true sharp anchor, and sensible timing.
¼ Kelly: A staking guide. Full Kelly is the mathematically growth-optimal bet size; it's also brutally swingy, so we show a quarter of it as a percentage of your bankroll.
Move 1h: How the edge has shifted in the last hour. Green means it's moving toward a live arbitrage; the sparkline shows the recent path.
Verify flag: An edge too large to be believable is almost always a stale or unbettable line. We still show it, flagged red, so you confirm it on the book before trusting it.

What this can't do

+EV is a long-run average. You will lose plenty of individual bets — that's the math working normally, not the tool failing.
Odds move. A price can change between our refresh and your click, which kills the edge. Always confirm on the book before you stake.
Confidence is a data-quality score, not the chance your bet wins. A 90/100 bet can lose.
Books limit or close accounts that consistently beat them. That's their business, and it's a real cost of this strategy.
Mandarin never places a bet for you, and nothing here is financial advice. Betting may be restricted where you live — that's on you to check.

Open the scanner