Timestamp Granularity Breaks Liquidity Measurement
Does timestamp precision change a measured spread or liquidity estimate?
Preserve each quote and trade?s original timestamp precision, then recompute spread or liquidity measures after controlled rounding to see how ordering changes. Results from high-frequency equity records are a warning; effects on coarser bars require separate measurement.
The published timestamp result does not directly validate every bar-data extrapolation.
Evidence map
| Aspect | Finding |
|---|---|
| What it is | The finding that in fast markets the precision of the timestamps determines whether liquidity measures are meaningful. |
| Key result / formula | Holden and Jacobsen study the standard United States quote and trade data and show that with timestamps recorded only to the second, the volume of events within each second is high enough that trades and quotes cannot be reliably sequenced. |
| Why it matters for backtesting | The direct lesson for Stochastly concerns bar construction and alignment. |
What it is
When many events share the same recorded second, the order in which they occurred is unknown, and each measure that depends on matching a trade to the quote in force becomes unreliable.
Key result / formula
The conventional remedies, such as shifting trades backward by a fixed interval before matching them to quotes, were calibrated for a slower market and introduce substantial errors in a fast one. The consequences they measure are large: effective spreads, price impact measures and the classification of trades as buyer or seller initiated are all materially distorted, and the distortions are systematic rather than random, so they bias research conclusions rather than merely adding noise. They propose interpolation methods that improve matters and they quantify how much the choice of method changes the answer, which is the number a researcher needs in order to know whether their result depends on it.
Why it matters for backtesting
A bar file whose timestamps are recorded to the minute or the second carries the same ambiguity within each stamp, and the ambiguity matters most for the very operations a backtest performs: deciding which bar a signal belongs to, and which price a fill receives. A user combining two sources with different timestamp precision, or different conventions about whether a stamp marks the start or the end of an interval, will produce an alignment error that looks like predictability. The check is to verify that both series agree on where a known common event falls, and the rule is to use coarser bars when the timestamp precision is uncertain, since the error is a fixed number of seconds and its relative importance falls as the bar lengthens.
Source
Holden & Jacobsen, "Liquidity Measurement Problems in Fast, Competitive Markets: Expensive and Cheap Solutions", Journal of Finance 69(4), 2014, 1747-1785. Primary source