Modeling · 6 min read
Five ways a position modeler flatters your trade, from frozen volatility to endpoint thinking and ignored costs, plus how to ask it harder questions.
A position modeler is the most honest tool in options trading and the easiest to fool, because you supply its assumptions. Feed it optimistic inputs and it will render a beautiful, precise, wrong picture of your trade. The errors below are not exotic. They are the default behaviors of most traders on most days, and each one biases the model in the same direction: your favor.
All five mistakes share a signature: they move modeled results in your favor, never against. That asymmetry is the diagnostic.
Every simulated leg needs an implied volatility, and where that number comes from matters enormously for far-from-the-money strikes. If your modeler cannot find a live quote for a deep wing and quietly falls back to a default, say a flat 20 percent, a cheap tail option can be priced at several times its real value. A put that actually trades at $0.15 modeled at $0.85 injects $70 of fantasy value per contract into your structure, and a six-leg position can accumulate hundreds of dollars of phantom edge this way while looking perfectly plausible on the chart.
The tell is a plateau or peak noticeably higher than what the same structure shows when priced leg by leg against the live chain. Insist on knowing each leg's volatility source: real quote, interpolated from the smile, or defaulted. A modeler that flags defaulted legs is protecting you from the single largest silent error in retail position modeling. Also model the skew honestly: pricing a put wing at the at-the-money volatility when the market actually charges 4 points more for downside strikes misstates both cost and Vega.
If you plan to close at 21 days to expiration, the expiration diagram describes a trade you will never hold. The gap is not small. An iron condor showing $310 of maximum profit at expiration might offer only $170 of realistically capturable profit at your planned exit date, because the final $140 accrues in the last, most dangerous weeks, the ones you are deliberately avoiding. Modeling the payoff at your actual exit horizon regularly cuts the attractive-looking reward by 30 to 50 percent, which changes whether the trade clears your bar at all.
Figure: Mistake 2, drawn: the expiration payoff is a trade you may never hold. At your real exit date you live on the earlier-dated dashed curve, and the modeled reward shrinks with it.
The same error hides inside multi-expiration structures. When held positions and simulated legs share a chart, an "at expiry" line is anchored to some specific expiration, usually the nearest one in the book. If your simulated spread expires in 30 days but a held position expires Friday, the expiry curve may be drawn at Friday's date, showing your new structure at a horizon where it has barely matured. Always confirm which date the curve is drawn at, and evaluate the structure at its own horizon.
The model fills every leg at the midpoint of the bid-ask spread. The market does not. On a four-leg structure where each leg carries a $0.10 spread, paying just 40 percent of the spread past mid on each leg costs $0.16 total, or $16 per one-lot and $32 round trip if you exit the same way. On a trade modeled to make $85, execution alone consumes 38 percent of the theoretical edge before commissions, and you have not been wrong about anything yet.
The modeler prices a smooth curve; the contract's fine print prices reality. American-style equity options can be assigned early, and a short in-the-money call over a dividend is a classic surprise, converting a tidy modeled spread into a short stock position plus a dividend liability overnight. Cash-settled index options cannot be assigned early, but some settle to a morning-auction value rather than a closing price, and a position "safely" out of the money at Thursday's close can settle painfully through the strike on Friday's open. Physical delivery on stock options means a pin near the strike at expiration leaves you guessing whether you own shares Monday morning.
None of this appears on the profit curve. A short 450 call modeled to expire worthless at $449.80 shows +$100 of clean profit; the version where you get assigned, carry short stock over a weekend gap, and buy back 2 percent higher shows −$800. The model is not wrong, it was never asked. Before entering, answer three questions off-chart: American or European exercise, cash or physical settlement, and what happens if spot pins your short strike at the close.
Zoom the price axis to ±3 percent and almost any short-premium structure looks like a broad, forgiving plateau. The wings, where the actual risk lives, sit just off-screen. If the underlying has moved 5 percent in a week twice this year, your modeled range must include ±5 percent, and ideally worse. A condor showing a maximum modeled loss of $120 on a ±3 percent window may carry a real maximum loss of $880 that simply was not drawn.
The same cherry-picking happens on the volatility axis. Modeling a short-Vega position only at today's implied volatility hides the scenario that actually hurts it. Shock volatility up 25 percent of its level and re-draw: if the trade only works on the unshocked chart, you are long a calm market, not short premium with edge. Set your default view wide enough to be uncomfortable, because the point of the model is to see the losses before the market shows them to you.
All five mistakes share a signature: they move modeled results in your favor, never against. That asymmetry is the diagnostic. An honest modeling habit deliberately leans the other way: flagged volatility sources, exit-date horizons, slippage haircuts, settlement rules checked, ranges wide enough to include the bad year. If a trade still looks good after all five corrections, you have found something. If it only looks good before them, the modeler did not find you a trade; it drew you a flattering portrait of one.
Educational content, not investment advice. Options involve substantial risk.