Backtesting Metrics

Overview of how the autotrading backtest engine operates and what each of the different performance metrics mean.

Option Alpha provides a backtesting tool for idea generation and discovery.

The backtester uses end-of-day (EOD) data, meaning there is only one data point per day per option symbol. Each data point is OPRA (Options Price Reporting Authority) data, captured at approximately 3:45 PM Eastern time every day.

Backtesting data exists back to January 2007. An "All-Time" backtest start date is 1 January 2007. All backtests are pre-run with 1 contract; allocation settings are applied after the fact.

If Avoid Earnings is set, the backtester won't open a position if there is an earnings event during the expiration cycle of the position about to be opened.

Allocation

The allocation rules are straightforward. Each backtest is given a Starting Capital and at least one (up to three) allocations.

Allocators are rules-based, just like recipes. Unless the criteria are met, a new backtest position won't be opened until capital is freed.

Parameter Selection

Backtesting parameter selection works in a slightly more restrictive manner than the bots. Each parameter mapping, such as delta, is chosen in the same closest to manner that the bots would use.

However, each parameter type is bound by a unique range. If the next closest value to the chosen parameter value is outside of that range, the backtester will simply not open a position on that day.

Date selection

The backtester's selection for the date to open positions will vary depending on the other entry and exit conditions you set.

For example, if when using Sequential frequency, the dates will be one after another and this could vary depending on the duration (DTE) of the contract. If using Weekly the backtester will seek to open a position every seven days, if available.

Ultimately, the various entry and exit conditions, as well as the trade frequency, will all determine the date when positions are opened.

Statistics

For more information on the performance metrics used on the backtesting results page, visit the Performance Metrics page of the Handbook.

Profit

Performance Ratios

Win/Loss Metrics

Position Averages

Expected P/L Per Position

After a backtest has concluded, we can calculate the historical expectancy (as opposed to projected expectancy based on theoretical probability, for example). In other words, given enough occurrences, how much money do we expect to make per trade?

The generic formula is:

No tags found.
Was this helpful?

FAQs

Related articles