Most people who blow up their first arbitrage bet don't get the math wrong — they get the sequencing wrong.

Here's the mistake: you find a 4% arb across two books, split your stake 50/50 like every beginner guide tells you, and place Leg A first because it has better odds. By the time you go to place Leg B, the line moved 8 points and your "guaranteed profit" is now a guaranteed loss.

The fix is stake-sizing based on implied probability, not a flat split:

Stake_A = Total / (1 + OddsA/OddsB)

StakeB = Total - StakeA

Then — and this is the part everyone skips — you place the leg with the LOWER liquidity / faster-moving line FIRST, not the one with better odds. Book depth matters more than price. A -110 line on a sharp book can hold for minutes. A +250 line on a soft book can move in seconds once volume hits it.

Quick example on a $500 total stake at +120 / -140:

• Implied prob A (+120) = 100/220 = 45.45%

• Implied prob B (-140) = 140/240 = 58.33%

• Stake_A = $500 × (58.33 / (45.45+58.33)) ≈ $281

• Stake_B ≈ $219

Both legs return within ~1% of each other regardless of outcome — that's the actual "guaranteed" part. Skip the math and eyeball a 50/50 split, and you're just gambling with extra steps.

If you're scanning for arbs manually across books, you will miss this timing window constantly — it's why real-time alerting matters more than the arb-finding itself. Happy to answer stake-sizing questions in the comments.