Who this is for — Anyone who needs to distribute an order over time or assess an execution against a time average without confusing the benchmark's name with the algorithm's actual behavior.
TWAP stands for Time-Weighted Average Price. The term is used for two related but distinct objects:
- a benchmark, meaning an average of prices observed at defined time intervals;
- an execution strategy that divides a larger order into child orders distributed across a time window.
The benchmark is used to compare an outcome. The algorithm decides how to seek fills. A platform order labeled “TWAP” does not promise that the final price will equal the benchmark, nor does it imply that every child order will have the same quantity or execute successfully.
In plain terms — “Average the price every five minutes” is a measurement rule. “Send portions of my order over two hours” is an execution rule. They may use the same clock, but they answer different questions.
The TWAP benchmark
For a series of prices observed at equally spaced times, a simple form is:
TWAP = sum of observed prices / number of observations
An operational definition must state which price is sampled — last trade, midpoint or a bar's typical price — as well as the frequency and the beginning and end of the window. Changing any of these changes the benchmark. TWAP is not therefore a single “true average” independent of the data; it is a measure constructed under a stated methodology.
VWAP instead gives greater weight to prices associated with higher volume. TWAP and VWAP coincide only in special cases. A session in which most volume trades during high-price periods can produce materially different values. A time-scheduled algorithm should therefore not be described as a guaranteed way to achieve VWAP.
The TWAP execution algorithm
The initial instruction, or parent order, specifies at least the side, quantity, instrument and time window. A limit price, start time, urgency, permitted venues and cancellation conditions may also apply. The algorithm generates or manages the child orders used to work the quantity.
A teaching example might divide 1,200 units into twelve slices of 100. Real implementations may instead:
- vary or slightly randomize child-order sizes;
- begin passively and become more aggressive near the end of a bucket;
- redistribute unfilled quantity into later intervals;
- observe a limit price and leave incompatible quantity unfilled;
- pause, cancel or terminate when the duration expires.
Coinbase Advanced documentation, for example, says its estimated suborder size
may vary, some children are randomized and unfilled quantity may be
redistributed. Coinbase Prime requires parameters including start_time,
expiry_time and limit_price for its TWAP order. These are product examples,
not a universal definition for every provider.
Instruction, algorithm and regulatory classification
A user may select “TWAP” as an instruction in an interface, but a system behind that selection determines timing, quantity or child-order management. Article 4(1)(39) of MiFID II defines algorithmic trading as trading in which an algorithm automatically determines individual order parameters with limited or no human intervention. It excludes systems used only for routing orders without determining trading parameters.
ESMA's 2026 Supervisory Briefing identifies timing, price, quantity and post-submission management as examples of parameters and emphasizes pre-trade controls, governance, testing and outsourcing. The exact classification still depends on the activity, entity and applicable framework: the presence of a TWAP button is not a substitute for a regulatory assessment.
What it may achieve — and what it does not guarantee
Spreading quantity over time can avoid demanding all available liquidity in one instant and reduce temporal concentration. It does not guarantee:
- completion of the parent quantity;
- a better price than an immediate order;
- an execution price equal to TWAP or VWAP;
- lower market impact or slippage;
- anonymity or freedom from information leakage;
- protection from a market moving away during the window.
A longer duration can reduce urgency and immediate impact, but increases exposure to price risk and opportunity cost. A shorter duration concentrates the children and may consume more liquidity. Passive orders avoid immediately crossing the spread, but introduce non-execution and adverse selection risk. No single clock setting is optimal across all instruments and market regimes.
Common mistake — Assessing TWAP only from the average price of completed fills. When a material portion remains unexecuted, ignoring the residual quantity and subsequent market move conceals opportunity cost.
Evaluation example
An investor must buy 12,000 shares between 10:00 and 12:00 subject to a maximum limit. The algorithm schedules child orders across the window, but the market trades above the limit for forty minutes. Some children remain unfilled and are retried later; 9,500 shares have been acquired when the order expires.
The average price of those 9,500 shares can be compared with declared benchmarks, but the assessment must also include fill ratio, residual quantity, commissions, spread and the cost of non-execution. Saying only “better than TWAP” does not describe the full result.
Control checklist
- Define parent quantity, window, price limit and termination rule.
- Read how the provider sizes, randomizes and catches up child orders.
- Establish benchmarks and timestamps before execution.
- Record quantity-weighted fills, residuals, cancellations and costs.
- Separate algorithm outcome, TWAP benchmark and VWAP comparison.
- Verify controls, outage handling and behavior during market suspensions.
The full context is in the Orders, execution and market microstructure hub. Economic components are covered in transaction-costs and execution-quality.
Sources
- Coinbase Help, Advanced Trade order types — TWAP orders — parameters, suborders, redistribution, passive/aggressive behavior and randomization in the documented implementation.
- Coinbase Developer Documentation, Prime — Order Types — TWAP parameters and the distinction from VWAP orders in Coinbase Prime.
- EUR-Lex, Directive 2014/65/EU, Article 4(1)(39) — algorithmic-trading definition and the exclusion for routing alone.
- ESMA, Supervisory Briefing on Algorithmic Trading in the EU, 26 February 2026 — automated parameters, governance, testing, outsourcing and pre-trade controls.