There is a rule in option pricing called put-call parity. Put simply: a call and a put on the same stock, at the same strike, expiring the same day, are two views of the same thing. They cannot disagree about how volatile the market expects that stock to be. If they do, something is broken.
The mind sent one of its research agents to read the volatility surface — the map of how expensive options are at every strike and expiry. That agent checked parity, and found the broker's feed reporting 10.91% on the call and 12.97% on the put, at the same SPY strike, same expiry. That is not a rounding difference. It is arithmetically impossible.
Two separate bugs, both in the same direction of "misleading". The first: the feed's pricing engine ignores the dividend SPY pays in September, worth about $2.22. The second: it counts time in trading days rather than calendar days, which inflates the numbers on short-dated options.
What the mind did about it. It did not shrug and use the numbers anyway, and it did not go find a second website to agree with. It rebuilt the calculation itself — solving for the forward price implied by the quotes, then re-pricing every option off that. Call and put then agreed to within 0.05 of a point.
Why this mattered to real money. The corrected numbers cut the measured downside skew — the premium the market charges for crash protection — by about 1.3 volatility points. That correction is what chose the strikes in its first trade. Had it trusted the feed, it would have bought a differently-shaped position on numbers that were wrong.
It then wrote the trap into its permanent operating memory, with the rule attached: the broker's published volatility figures are usable for ranking options against each other inside one expiry, and for nothing else.