To compute net GEX, you must assume which side of each option dealers are on. The standard convention assumes dealers are long calls and short puts. It's a reasonable baseline — and worth knowing where it comes from and when it fails.

Why the baseline usually holds

The assumption reflects typical customer behavior: retail and momentum traders tend to buy calls (so dealers sell them and are effectively long that call gamma on their books via hedging convention), while funds and hedgers buy puts for protection (so dealers are short those puts). On an average day, the aggregate leaves dealers positioned close to what the model assumes, so the sign is right.

When it breaks

Unusual flow can invert it. A wave of institutional call selling (overwriting), heavy put buying for a hedge, or a big structured trade can put dealers on the opposite side of a strike than assumed — and since the sign flips that strike's contribution, a wrong assumption can push the whole net-GEX read, and the flip, to the wrong place. The number looks precise; the assumption underneath just failed.

GEX is only as right as its guess about who dealers traded against — and on abnormal-flow days, that guess is exactly what's wrong.

How to stay honest about it

Treat GEX as a strong-but-fallible read, not gospel. When the tape flatly contradicts the map — price ripping in a supposedly “pinned” positive-gamma regime, or grinding calmly in a “negative-gamma” one — suspect the positioning assumption before you trust the number over your eyes. The tape is the ground truth; GEX is a model of it. When they disagree loudly, the model is the thing to doubt.