Point-in-time universe (PIT universe)
A point-in-time universe (PIT universe) is the set of instruments that qualify for inclusion at a specific historical timestamp, based only on information available up to that timestamp. Membership is time-varying by design: symbols enter when they become eligible and exit when they become ineligible (for example, after coverage ends). PIT universes are a core tool for reducing survivorship bias in backtests.
Why it matters
Using a static universe (often built from today’s symbol list) can exclude delisted markets and include symbols before they existed. That changes return distributions, liquidity constraints, and even which opportunities your strategy can access. A PIT universe keeps research aligned with what a trader could actually have traded at each point in time.
How it works
To build a PIT universe, you:
- define eligibility rules (venues, quote currency, instrument type, minimum history, liquidity)
- apply those rules “as-of” each date or rebalance timestamp
- ensure any metrics used in selection are computed from past-only data
Many teams store the resulting membership per rebalance so the universe can be reproduced exactly.
How do you build a point-in-time universe for crypto?
Choose your venues and instrument types first, then apply coverage-window checks so symbols are included only when the timestamp falls inside the relevant window. Next, enforce minimum-history and liquidity rules using only past data (not forward-looking windows). Finally, persist the membership per rebalance so you can rerun the backtest with the same universe.
What is the difference between a PIT universe and a static universe?
A static universe is one fixed list used for the entire backtest period. A PIT universe is recomputed through time and reflects listings, delistings, and eligibility changes. Static universes are simpler but are a common source of survivorship bias.
Do PIT universes fully eliminate backtest bias?
They remove a major source of bias related to universe membership, but other issues remain. Execution modeling, fees, latency, and timestamp alignment can still distort results if handled poorly. PIT is necessary for credible results, but it is not sufficient by itself.
Real-world example
You research an altcoin strategy from 2017 onward. A static universe built in 2026 includes many tokens that were listed after 2017 and excludes many that were delisted. A PIT universe adds symbols only after they become eligible and keeps delisted symbols in the sample during the period they existed, producing a more realistic backtest.
How CoinAPI Market Data API helps
CoinAPI’s Market Data API provides symbol metadata via GET /v1/symbols, including coverage boundaries like data_start / data_end and, where available, type-specific windows such as data_trade_start / data_trade_end. These fields help you time-bound symbol eligibility and build universe membership that is consistent with historical data availability. For reproducibility, you can snapshot the metadata and the resulting universe at each rebalance.