A dealer map needs to know not just how much open interest sits at a strike but which side the market maker is on. Open interest alone does not say. Someone is long and someone is short every contract.
The conventional assumption
The standard approach assumes customers buy calls and buy puts, so dealers are short both, and signs the gamma accordingly. It is a reasonable prior in equity index options, where the dominant customer behaviour is well established, and it is what makes the usual GEX calculation work.
It is an assumption nonetheless, and where customer behaviour differs from the prior, the sign can be wrong — which flips the entire interpretation of a strike from damping to amplifying.
What crypto venues expose
Crypto derivatives venues publish trade data that identifies the taker — the side that crossed the spread. The maker on the other side is typically a liquidity provider, so accumulating taker flow over time gives you dealer positioning as its mirror image, observed rather than assumed.
This is a genuinely unusual disclosure. It is why some published crypto gamma measures are built from cumulative taker flow instead of from open interest, and the difference is methodological, not cosmetic.
Where they disagree, and why that matters
Both are estimates and each has a characteristic weakness. Open-interest gamma is wrong when the customer-buys prior is wrong. Taker-flow gamma is wrong when the taker is not a customer — a dealer aggressively hedging is also a taker, and gets counted as customer flow.
Taker flow also needs a complete history from the point positions were opened; miss a period and the inferred inventory is permanently off, in the same way on-chain depth history cannot be backfilled.
The honest position
Neither is ground truth, because dealer inventory is private. Where the two methods agree, confidence is higher than either alone justifies. Where they disagree, the useful response is to widen the uncertainty rather than to pick the one that supports the trade — and to remember that a map is a model of obligation, not a report of it.