You know the drill by now: convert every price in a complete market, add the implied probabilities, and the sum lands above 100%, the excess being the margin. One day the sum lands below 100%. That is an underround, and it never means what the excited part of your brain hopes it means. Almost always it means your inputs are lying to you.
What the number is
The no-vig calculator states the case directly: a sum below 100% is an underround, and it is flagged without changing your inputs, checked 2026-10-28. Every price is still one divided by it; the tool simply refuses to pretend the row is normal.
An underround says the prices, taken together, imply the outcomes are collectively less than certain. One side at 2.20 and the other at 2.10 gives 45.5% plus 47.6%, a sum of 93.1%. If those prices were real, simultaneous and complete, someone could take both and profit from either result. Which is the first clue that they are probably not real, simultaneous and complete.
Cause one: your timestamps disagree
The commonest underround is homemade. You priced one side from this minute's snapshot and the other from a stale cache, or you mixed a live price with a prematch close. Prices move; two honest prices from different minutes can easily sum below 100%, and the combination is a measurement error, not a market state.
The fix is procedural: compute the sum only from prices sharing a timestamp, as the normalization discipline of matching keys and timestamps requires. An underround that disappears when you align the inputs was never there.
Cause two: the market is incomplete
The second cause is quieter. A two-way sum reads fine for tennis and wrong for soccer, because soccer has a draw. If your row drops an outcome, the sum of what remains can sit below 100% and look like a gift. It is a missing column.
Check completeness before anything clever: count the outcomes the market should have and the outcomes your row actually carries.
Cause three: a genuine cross-book window
The rare case is real. Two books can disagree enough that the best price on each side, taken from different books, sums below 100%. It happens, it lasts seconds to minutes, and it closes precisely because it is visible to everyone watching.
Treat a candidate like that as a claim to verify, not a find to celebrate. Re-read both prices at the same second, confirm both are actually offered at stakes you care about, and only then believe the arithmetic. The odds converter turns each price into its implied probability for exactly this check, checked 2026-10-28.
The reading rule
From one book's feed, an underround is never an opportunity, because one book does not offer you both sides of its own mistake. It is a data alarm, and the correct response is to distrust the row.
Across books, it is a claim with a short shelf life. Verify first, and let the sum above 100% return before you relax.
Margin is the normal tax on prices. An underround is the market telling you to check your work, and nine times in ten the work is where the error lives.