Skip to Content
Optimization & WFADay-of-Week & Time-of-Day

Test days and times to exclude

Compare full backtests with new entries blocked on one weekday or during one UTC time slot at a time.

Watch the one-minute walkthrough .

Before you start

Save a strategy and run an ordinary backtest as your baseline. Keep its data, dates, execution settings, and strategy parameters for the comparison.

Day of week works with a supported common data frequency. Time of day needs intraday bars. Free-plan global data is daily-only; Pro and above can use available global intraday data. Availability still depends on the selected symbols. See Data management.

Both test types consume 2 credits per backtest. They do not use the two lifetime free parameter-grid optimizations. A weekday test can use a Free account’s available credits when its data and other settings are permitted.

Configure the test

  1. Open Optimization and select New optimization, or select Optimize on a saved strategy.
  2. In Test, choose your saved strategy and Day of week or Time of day.
  3. In Data, select the symbols. Review the combined trading calendar shown for the selection. In Execution, set a shared Bar frequency, overlapping Date range, and the simulation settings used for every backtest.
  4. Configure Weekdays or Time slots as described below. Review Fixed parameters; these values remain the same across all tests.
  5. In Review and run, name the run, select the comparison metric and check its direction, and review the backtest count and credits. The default metric is Realized PnL.
  6. Select Run optimization. Results update as each backtest finishes.

Day of week

LONA uses the instrument’s calendar to determine the tested weekdays: typically Monday–Friday for equities, ETFs, and forex, and all seven days for crypto. It runs one backtest per trading day, blocking new entries on that day throughout the selected period. With custom data and no exchange calendar, every UTC weekday is tested. Multiple markets use the union of their trading weekdays.

A five-day calendar runs 5 backtests for 10 credits; a seven-day calendar runs 7 for 14 credits. Strategy parameters are not swept.

Time of day

Choose an intraday frequency such as 1h, 4h, 15m, or 5m, when available. Under Time slots, choose Start time (UTC) and Slot width.

The start must align with the data’s bar grid and trading session. Available slot widths are whole multiples of the bar duration. For example, one-hour bars can test one-hour slots, while four-hour bars need slots at least four hours wide.

LONA divides the remaining session into slots. With multiple markets, the displayed session spans their combined UTC hours; some slots may be outside an individual market’s hours. Custom data without an exchange calendar uses a 24-hour UTC session. Where the slot strip permits individual selection, click a slot to skip that test. Each retained slot produces one backtest and costs 2 credits. A session’s last slot may be shorter; review the displayed intervals and estimate before submitting.

Time slots use UTC, not your browser’s local timezone. Each exclusion includes its start and excludes its end: a slot from 08:00 to 09:00 blocks entries at 08:00 but does not block entries at 09:00.

Read the results

The bar chart and Results table compare the selected metric for the full backtest after each exclusion. A Monday bar means “the strategy with new Monday entries blocked,” not “profit earned on Mondays.” A time-slot bar has the same interpretation for its UTC interval.

Strongest when blocked and Weakest when blocked summarize the observed comparison for the chosen metric and direction. Compare these results with your original backtest; the test does not prove that excluding a period improves future performance.

Existing positions can still close during an excluded period. The test does not force positions closed at a slot boundary or prevent holding a position overnight. Blocking an entry can also change later trades, so a bar is not a standalone attribution of that period’s profit.

Use Claude or ChatGPT

With LONA connected, ask:

Using my saved strategy, prepare a Day-of-Week optimization with one test per trading day, blocking new entries on that day. Keep strategy parameters fixed and compare realized profit and loss. Show the dataset, date range and estimated credits, and ask for confirmation before running.

For intraday data:

Using my saved strategy, prepare a Time-of-Day optimization using one-hour bars. Test one-hour UTC slots across the instrument’s trading session, blocking new entries in one slot per backtest. Keep strategy parameters fixed and compare realized profit and loss. Show the chosen dataset, date range, UTC time slots and estimated credits, and ask for confirmation before running.

These runs use lona_run_time_window_optimization, not the parameter-grid tool. See MCP tools reference for the input contract.

Last updated on