The Exabel platform has native support for point-in-time data. For every time series in Exabel, it is possible to retrieve that time series as it was available at a particular point-in-time (based on its "known time").
This is critical to avoid look-ahead bias when evaluating a time series dataset through backtests, applying a dataset towards predicting company fundamentals, or simulating portfolio strategies. Under these scenarios, it is important to use only data that would have been available at that point-in-time, when making a prediction or portfolio allocation on a historical date.
Current support for point-in-time
- Data Import: support for specifying the point-in-time (
knownTime) for time series being imported
- Data Export: work-in-progress to query & export of time series as of a given point-in-time
- Signal Explorer: evaluation & plotting of signals as of a given point-in-time
- Prediction Models: point-in-time is supported for the Prediction Model (ML) and Classification Model types.
- Alpha Tester: point-in-time evaluation of both signals and screens at each rebalance date
- Portfolio Strategies: point-in-time evaluation of both signals and screens at each rebalance date
How does point-in-time work under the hood?
To understand Exabel's native support for point-in-time data, refer to the time series page.
Updated 9 months ago