Value notes
How small an edge is too small to bet
A +0.3% EV is not a signal, it is rounding noise. Where to set the minimum bar so real edges pass and phantom edges fail.
Not every positive number is an opportunity. The EV calculator shows an edge as a percentage, and somewhere between +0.2% and +2% the number stops describing a real advantage and starts describing the noise in your inputs. Setting a minimum bar before you start is how you keep phantom edges out of your record.
Why a tiny edge is mostly noise
Every input to the EV figure is an estimate. The fair chance comes from removing a margin with a modeling assumption, your book's price moves while you check it, and the whole computation inherits both errors. A computed edge of +0.3% is smaller than the uncertainty of its own inputs, which means it is as likely to be negative as real.
The worked example on the calculator, a +3.80% edge at prices of 2.10, fair 2.17 and 2.25 offered, is comfortably above that noise floor, checked 2026-10-08. The trouble starts far below it.
What the bar actually filters
A minimum EV bar is not about greed; it is about sample hygiene. Bets at +0.3% will occupy your ledger, your attention and your bankroll fractions while contributing almost nothing but variance. The Kelly guide sizes stakes on the edge, so a phantom edge also produces a phantom stake: money committed to a bet whose edge never existed.
A bar of about 1% to 2% is a sane starting point for prices read by hand from a screen; automated pipelines with tighter inputs can defend a lower one. What matters is that the bar exists before the bet, not after the loss.
The bar is per market, not per account
Different markets carry different error sizes. A main line on a major league has a tight fair price and a small noise floor; a side market on a minor league has a wider one. Set the bar where the error lives: raise it for thin markets and for prices you did not just refresh, lower it only where your fair-price inputs are proven.
Whatever bar you choose, write it down next to the threshold that triggers your alerts. Two numbers, one for the market's move and one for your own, keep the whole process honest.
Skipping is a position
A bet you do not place at +0.3% costs you nothing and teaches you nothing; a bet you place there costs both. The bar turns a hundred maybes into a dozen real candidates, and your record, measured against the close as the CLV guide sets out, will tell you over a few hundred bets whether the bar is set right.
A minimum edge is a filter on your own arithmetic. Set it, write it down, and let the small numbers go.