92 / 97
Time-Weighted Average Price (TWAP)
Written TWAP throughout.
Working definition
The average of an instrument's price sampled at equal intervals over a window — used both as an execution benchmark and as the schedule for an order that trades evenly through time.
TWAP names two things that are easy to confuse. As a benchmark it is the arithmetic mean of prices sampled at equal time intervals across a window. As an algorithm it is an execution schedule that slices an order into equal pieces released at equal intervals, so that the fills track that mean.
Its contrast with the volume-weighted average price is instructive. VWAP weights each interval by the volume that traded in it, so a VWAP schedule trades more when the market is busy and less when it is quiet — closer to how liquidity actually arrives, and correspondingly harder to run in an instrument whose volume profile is unstable. TWAP ignores volume entirely. That makes it predictable and easy to reason about, and appropriate where no reliable volume profile exists; it also means it keeps sending equal slices into the thinnest part of the window, paying more market impact per unit there than anywhere else.
As a benchmark it carries the self-reference problem shared by every measure computed from the same window as the execution. An order large enough to move prices moves the average it is judged against, and beating a benchmark one is helping to set is not evidence of good execution. Sampling choices matter too: a TWAP over one-minute samples and a TWAP over tick data are different numbers, and neither is canonical.
The data underneath inherits every property of the feed it came from. Sampled from a censored feed, a TWAP silently omits the excursions that were dropped; sampled from a series padded by a carry-forward artifact, it averages repeated prints that record absence rather than trading.
Why it matters
TWAP is a schedule and a yardstick, not a cost measurement. It says how an execution compared with the window it happened in — a question whose answer can look excellent while the decision that prompted the trade decayed. Implementation shortfall, anchored to the decision price rather than the execution window, is the stricter test, and the one that prices what waiting cost.
Commonly confused with
Neighbouring concepts that get used interchangeably, and the distinction that actually separates them.
- VWAP
The volume-weighted average weights each interval by what traded in it, so a VWAP schedule trades more when the market is busy. TWAP ignores volume entirely, which makes it predictable and appropriate where no reliable volume profile exists — and means it keeps sending equal slices into the thinnest part of the window.
- Implementation shortfall
TWAP is measured against the window the execution happened in; shortfall is anchored to the price at the moment the decision was made, before any trading. The first asks how the execution compared with its own window, the second asks how much of the decision survived. Only one of them prices what waiting cost.
- The benchmark versus the schedule
TWAP names two different things and the sentence rarely says which. As a benchmark it is an arithmetic mean used to grade an execution; as an algorithm it is a schedule that slices an order evenly through time. An execution can run one and be graded against the other.
- Market impact
Impact is a cost TWAP does not measure and can worsen. Equal slices arriving during the thinnest part of a window pay more impact per unit there than anywhere else, and the benchmark reports nothing about it.
How to measure it in your own data
A definition you cannot test is a definition you have to take on trust. This is the shortest honest route from the concept to a number you computed yourself.
- Records you need
A price series over the window with the sampling interval explicitly stated, and the execution's own fills with timestamps. The interval is not a detail: a TWAP over one-minute samples and a TWAP over tick data are different numbers, and neither is canonical.
- What you compute
The arithmetic mean of prices sampled at equal intervals across the window, then the fills compared against it. State the interval alongside the result, because a TWAP quoted without one cannot be reproduced or challenged.
- What the answer tells you
Treat a favourable comparison carefully where the order was large. An order big enough to move prices moves the average it is being judged against, so beating a benchmark you are helping to set is not evidence of good execution. Check the underlying feed too: sampled from a censored series a TWAP silently omits the excursions that were dropped, and sampled from one padded by carry-forward artifacts it averages repeated prints that record absence rather than trading.
Questions and answers
What is the difference between TWAP and VWAP?
Weighting. TWAP treats every interval equally regardless of activity; VWAP weights by the volume that actually traded, which tracks how liquidity arrives. VWAP is closer to reality and harder to run where the volume profile is unstable, which is exactly when people fall back to TWAP.
Does beating TWAP mean I executed well?
Not necessarily, because of a self-reference problem the benchmark cannot escape. It is computed from the same window as your execution, so a large order moves the yardstick in its own favour. Worse, a benchmark drifting along with your delay rewards trading slowly, whatever the delay cost the decision behind the trade.
Which sampling interval should I use?
There is no canonical answer, which is precisely why it should be stated. Different intervals give different TWAPs for the same window and the same trading. The defensible practice is to fix the interval in advance and publish it with the number, rather than choose it after seeing which one flatters.
Can a bad data feed distort a TWAP?
Yes, and silently. A censored feed drops the excursions, so the average is computed over a calmer market than the one that existed. A feed padded with carry-forward prints averages repeated values that record absence of trading rather than trades. In both cases the TWAP looks perfectly ordinary.
Related terms
Derived from the links this entry makes and the entries that link back to it.