A stale quote is a displayed bid or ask that no longer reflects a tradable market.
A stale quote is a displayed bid or ask that no longer reflects a tradable market. In practice, the quote remains visible after the underlying market has moved, the venue has updated, or the opportunity to trade at that price has disappeared. For quantitative research, that matters because a backtest can treat the displayed quote as executable even when live execution could not have occurred at that level.
The mechanism is straightforward. A strategy reads quote data, applies an entry or exit rule, and the simulator fills the order at the quoted price. If that quote was stale, the simulated fill is based on an observation that was visible in the data but not actually available in the live market when the order would have reached execution. The result is a phantom fill: a backtest fill that exists in simulation because of the data record, not because the market was still tradable there.
Cross venue data issues make this problem more likely. Sonar’s research on cross venue data describes how market data from different venues can arrive with different timing and quality characteristics, which can distort consolidated views of the market and create misleading apparent opportunities if the feeds are not handled carefully. That kind of timing mismatch is one path by which a quote can appear valid in a historical dataset while already being obsolete in the live market. When a backtest consumes those observations without modeling quote freshness, venue specific timing, or execution latency, it can overstate how often a strategy would have traded at the observed prices.
This directly affects performance evaluation. Phantom fills can improve apparent entry prices, improve apparent exit prices, increase the number of filled trades, and reduce measured slippage in the simulation. Those changes flow into headline metrics and can make a strategy look more stable or more profitable than it really is under live conditions. Sonar’s backtest overfitting audit tool is relevant here because it is designed to examine whether reported backtest results are robust or whether they are being inflated by artifacts of the research process. If stale quotes create non executable fills, they can become one more source of inflated backtest metrics that the researcher needs to isolate and test.
The statistical consequence is that stale quote effects can contaminate downstream measures of strategy quality. Sonar’s glossary entry on the deflated Sharpe ratio explains that conventional performance statistics can be misleading when they are not adjusted for the realities of the testing process. Although the glossary does not discuss stale quotes specifically, the connection is important: if the return series used for evaluation already includes phantom fills, then even a carefully interpreted performance ratio is being computed on distorted inputs. In that case, the issue begins before the statistic. The data and execution assumptions must be repaired first.
For that reason, stale quotes are best understood as an execution realism problem inside historical simulation. A quote can be present in a dataset and still be too old to support an executable assumption. Once that happens, the simulator may record fills that a live strategy would not receive. That is the core sense in which stale quotes introduce phantom backtest fills by displaying prices that are no longer executable in the live market.
Covered in depth in the Cross-venue market data & signals pillar hub.