Model Risk: When the Map Is Wrong

Every map is a simplification. A road map leaves out potholes; the positioning map leaves out things nobody can see in public data. The gap between a model and the market it describes is called model risk. You can’t remove it, but you can learn where it hides and decide how much weight each level deserves.

Start with what StratosHedge actually computes. For each strike it estimates gamma exposure as gamma × open interest × 100 × spot, in dollars of stock per $1 move. Calls count as positive and puts as negative. The Ceiling is the strike with the largest positive total, the Floor is the most negative, and the Flip Line is the strike gap nearest the price where net GEX turns from negative below to positive above; on the nearest expiry it is held between readings so it doesn’t hop on small moves. (The textbook version, where the running total of GEX across strikes crosses zero, is a cruder proxy.) Every one of those steps carries an assumption.

Five places the map can be wrong

1. The sign is an assumption

   The model relies on the dealer-side sign assumption: customers mostly sell calls (covered calls, for example) and buy puts for protection, so dealers end up long calls and short puts. Real dealer inventory isn’t public. When customers sell puts in size for income, or pile into calls during a speculative rally, dealers may hold the opposite of what the model assumes at those strikes, and the real sign there can flip.

2. Open interest is a day old

   Open interest settles once a day, overnight. A position opened at 10 a.m. doesn’t appear in it until the next morning. During the session, changes in the map come from price and time reshaping the gamma of last night’s positions, not from new ones. On days with heavy new positioning, the map can lag well behind what dealers actually hold.

3. The window is narrow

   Levels use the nearest expiration (0DTE for SPY on most days) and about 30 strikes on each side of price. Big positions in next week’s or next month’s expiration, or far from price, sit outside the map even when they matter. The Magnet also comes from a volume-based estimate supplied by the data provider.

4. Not every contract has a dealer behind it

   Open interest counts every open contract, including trades between two customers, yet the model treats all of it as dealer inventory. Dealers also hedge their whole book with futures, other expirations and other options, not one strike at a time. Nor does the map know why a position exists: a fund’s protective puts may stay on for months, while a short-term directional bet can be closed within hours, and both look identical in open interest. The map shows potential hedging pressure, not actual orders.

5. The flip is a proxy

   The Flip Line is the strike gap nearest price where net GEX turns from negative below to positive above: an estimate of where behavior changes, not a measured price. Sometimes there’s no crossing near price, and the regime falls back to the sign of total net GEX. Near the line it reads “Near Flip Line · Mixed”, because that’s where readings are least reliable.

How to hold levels loosely

• Read each level as an area to watch, not a line that must hold. The Strength Zone’s thickness shows its estimated relative strength at each moment.
• Let price confirm. A level matters more when price slows, stalls or turns there; if price slices straight through, the map may be wrong about that strike.
• Be most skeptical when the assumptions are shaky: speculative call buying in a single stock, heavy put selling, or right after a big expiration rolls off.
• Check the next several expirations in the GEX Matrix before leaning on a near-term level. The Matrix loads only the nearest five, so a large monthly or quarterly position can be missing there too.
• Size every idea so that the map being wrong costs an amount you’ve already accepted. When the map and price disagree, price wins.