History that matches the live feed exactly.
Point-in-time series on the same schema as production, so a signal that works in research is reading the same shape it will read in deployment.
Block Scholes3 min read
- History since
- 2020
- Tokens
- 150+
- Options venues
- 10+
- Listings catalogued
- 1.8M+
/01 The problem
When a backtest fails in production
A signal that works in research and fails in production has usually failed on its data: history that was quietly revised, instruments that disappeared from the record, or a research dataset shaped differently from the live feed.
/02 Why Block Scholes
Research data built the way traders price
Most crypto research datasets are raw venue quotes: one exchange at a time, noisy, with gaps where a book went thin and instruments that vanish when they delist. A model fitted to that learns the noise as well as the market.
Block Scholes history is the calibrated market: composite prices, forward curves and fitted volatility surfaces, kept point-in-time back to 2020 and served on the same schema as the live feed. A researcher works on the numbers a trading team would have priced from on the day.
- Point-in-time back to 2020, with no lookahead
- Calibrated surfaces at any strike, delta or moneyness, listed or unlisted
- Delisted instruments retained, so a backtest includes the instruments that failed
- 150+ tokens covered, on one schema
- Backtests priced on real listed bid/ask as well as the model mid
/03 How it works
History by API, backtests by MCP
Over the REST API, a researcher pulls the slice a model needs: prices, implied volatility by strike, delta or moneyness, model parameters, forward curves, funding, open interest and volume, at minute or hourly granularity, for the latest value or a date range. The same request with a past date returns the market as it stood then.
Alternatively, a strategy can be described in a sentence over MCP. An agent runs the backtest across options, futures, perpetuals and spot, priced on real listed bid/ask as well as theoretical mids, and returns trade-by-trade PnL, hedge attribution, Sortino and max drawdown.
/04 Backtest to live
Run a proven signal live on the same feed
Once a backtest holds up, the signal goes live on the same data source. The same composite, curves and surfaces stream over WebSocket on the same schema and the same identifiers the research used, so the model reads in production the same data it was tested on, without re-mapping or reconciling a second vendor.
History, live data and backtests come from one pipeline, so a live result can be compared against its backtest like for like, and any drift between them reflects the market.
/05 Product expertise
Support for launching new derivative products
Some research ends in a new product: a new index, a structured payoff, or an options market on an asset not yet listed. The team behind the surfaces works with clients on this: the instrument’s design, the pricing model and its parameters, and how it should be marked, settled and risk-managed once it trades.
This is the same work Block Scholes does for its own indices and surfaces, carried out by researchers with more than 60 years of combined market experience and applied to the client’s product, on data it can already be tested against.
Questions
What teams ask before they build.
Is the research data the same as the live feed?
Yes. Point-in-time series on the same schema and identifiers production serves over WebSocket, so research and deployment read the same shape.Can I pull history over the REST API?
Yes. Prices, implied volatility, model parameters, forward curves, funding, open interest and volume, at minute or hourly granularity, point-in-time back to 2020.Are backtests priced on mids?
Both are available: a theoretical model mid and real listed bid/ask, so a strategy can be tested against the spread it would pay.Can you help us launch a new derivative product?
Yes. We work with teams on new derivative products, covering the instrument, its pricing model, and how it is marked and settled, using the same data and methods behind our own surfaces and indices.
Read next
The data and tools behind this.
- /01
Data & Analytics · Historical Data
Point-in-time history for backtesting and model validation, to three years of depth.
- /02
Infrastructure · Data API
REST and WebSocket access, from hourly snapshots to live streaming updates.
- /03
Infrastructure · Strategy Backtester
Position, hedge and lifecycle simulation running on the same data infrastructure.