Brad Lasater — Research Log
Building a systematic volatility trading system in the open: what I tried, what I expected, what happened, and what I had wrong.
https://bradlasater.com/log/
2026-09-08T10:32:27-06:00
Brad Lasater
brad@bradlasater.com
https://bradlasater.com/
© 2026 Brad Lasater
Every price in this system is a traded price, and that changes which tests are even meaningful
https://bradlasater.com/log/2026-09-07-pricing-with-imperfect-market-inputs.html
2026-09-07T12:00:00+00:00
2026-09-07T19:30:36-06:00
Option NBBO is a 403 on this data tier, so implied volatility is inverted from last trade or day close rather than from a quote midpoint. The consequence is not merely noisier numbers — it invalidates one of the three daily identity checks whenever either leg of a pair is a stale print, and it makes the spread-based cost model uncalibratable rather than approximate.
Brad Lasater
Half my datasets can be backfilled and half cannot, and that decides what a replay may claim
https://bradlasater.com/log/2026-09-07-historical-completeness-vs-point-in-time.html
2026-09-07T12:00:00+00:00
2026-09-07T19:30:36-06:00
Option day bars reach back to 2022-08-31 and can be rebuilt from the vendor archive at any time; chain snapshots exist only from 2026-08-31 because there is no historical snapshot endpoint. That asymmetry, not storage cost, is what limits which backtests this system will ever be allowed to run.
Brad Lasater
The ATM term structure agrees with itself by construction, so I had to work out when its agreement is evidence
https://bradlasater.com/log/2026-09-07-atm-term-structure-before-a-surface.html
2026-09-07T12:00:00+00:00
2026-09-08T10:32:27-06:00
The term structure inverts both the call and the put at the at-the-money strike and writes out both. When put-call parity happens to select that same strike, the two legs invert to identical volatilities as an algebraic identity — so their agreement proves nothing, and only the dates where the two strikes differ carry information.
Brad Lasater