What should disappear when you move a historical replay cutoff backward?
This deterministic example uses synthetic OHLC bars only. It compares an information-safe replay view with an explicitly invalid view that leaves future bars visible.
Earlier cutoffs should reduce what the chart is allowed to know.
Use the slider or arrow keys. The left panel withholds future bars. The right panel intentionally leaks them so the difference is visible.
Future information left visible
The cutoff changes the admissible information set.
At each position, the safe panel shows only bars whose synthetic interval has completed by that point. The leaky panel keeps the same future bars visible in a different treatment so you can see why a polished chart can still violate a historical information boundary.
Read the historical replay Method →Anything computed from the future must obey the same boundary.
Indicators, annotations, profile sessions, labels, and summaries can leak information even when future price candles are hidden. A replay-safe implementation needs availability semantics for every layer, not just the base OHLC series.
Use the Chart Evidence How-To →This example teaches information timing, not market behavior.
- The OHLC sequence is deterministic and synthetic.
- No trading rule, signal, execution model, or forecast is represented.
- The leaky panel is intentionally invalid and exists only for comparison.
- Real chart layers can have more complex availability rules than bar completion.
- A replay-safe chart does not by itself establish that a strategy test is free of all forms of look-ahead or selection bias.
Which non-price layer is easiest to leak accidentally?
After understanding the bar cutoff, inspect how indicators, events, or profile sessions establish when their values become knowable. The next question is about layer-specific availability, not whether the synthetic path looked favorable.