Nobody kept the quotes.
So we’re keeping them.
Historical option quotes are not for sale anywhere. Trades and one-minute bars, yes — the bid and ask you would have traded against, no. So the desk records them: an always-on OPRA capture writing real chain NBBO every second, for every day it runs.
This chain is not a screenshot.
Move the spot. Change the expiry. Push volatility around. Every quote, every implied volatility and every greek in the table is being computed as you do it, by the same closed form and the same Newton solver that run on the desk. Pick a bid to load that contract into the fill readout underneath.
| Calls | Strike | Puts | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Vega | Theta | Gamma | Delta | IV % | Bid | Ask | Strike | Bid | Ask | IV % | Delta | Gamma | Theta | Vega |
- NBBO
- 12.35 × 12.75
- Mid
- 12.55
- Your fill
- 12.7000
- Cost · ×100
- $1,270.00
A simulator that fills you at the mid is quietly claiming you were the one providing liquidity. The desk pushes the mid toward the taker by 0.75 of the half-spread and caps it at the far side of the NBBO: $1,270.00 against $1,255.00 — a difference of $15.00 per contract, before the size cap makes you walk for the rest.
A day that nobody recorded is gone. Option quotes cannot be fetched back — not from our vendor, not from any vendor.
So every replayable day is one of two things, and the desk tells you which before you trade it.
The quote
Real OPRA NBBO, captured off the websocket and coalesced to one-second rows.
A mid derived from Black-Scholes over the real underlying tick tape, at the volatility fitted from that contract’s own prints.
Displayed size
Real bid and ask sizes. A fill can never exceed what was actually shown.
None exists. The fill is capped instead by the volume the contract demonstrably absorbed in the surrounding interval.
The spread
Real width, second by second, including the moments it blew out.
Modelled from print deviation and bar estimators, trained on recorded days — and biased wide wherever the fit is uncertain.
IV and greeks
Solved from the real quoted mid, then greeks at that volatility.
Volatility is the fitted quantity: estimated from real traded prices, smoothed across strikes and expiries, interpolated where a contract never traded.
Open interest
Captured at the open, from the chain snapshot.
Not obtainable at a point in time. Shown as a dash rather than a guess.
What clears it to ship
It is the ground truth. Everything else is measured against it.
A blocking test: run the synthesis over days the recorder already covered and prove the synthetic fills are never better than the real-NBBO ones, bucket by bucket.
The recorder is the calendar-critical piece: every day it runs becomes a permanently replayable day, and every day it misses is unrecoverable. That is why it ships before the interface it feeds.
Never better than reality.
Options are where a generous simulator does the most damage: the spreads are wide enough that a mid-fill assumption manufactures an edge no live account has ever had.
Reported the way your broker reports them.
Greeks are computed server-side and sent down the wire, so the number in the chain is the number the fill engine used. There is no second calculation in the browser to drift away from it.
Long calls and puts. That is the whole of version one.
Buy to open, sell to close, auto-close at expiry, on liquid underlyings only. Everything below needs machinery that would be wrong to fake — margin, assignment, early exercise — so none of it ships until it can be built properly.
- Selling to open
- Multi-leg spreads
- Assignment modelling
- Options margin
- Small-cap chains
- Level 2 for options
Recorded, or labelled.
Never quietly guessed.
Coming soon Coming soon. Specified in full, not started — the equities desk is live today and options land on the same account when the recorder has days worth replaying.