TWAP gives equal weight to equally long periods. In trading, an automated tool uses a schedule to place smaller orders. The outcome depends on actual prices, filled quantities and the service’s rules, not just the schedule.
Jana wants to buy bitcoin without sending the whole order to the market at once. She encounters TWAP and could easily mistake the name for a predetermined purchase price. It actually refers to two related things: an average used to assess a trade afterwards, and a method of splitting an order. The average can be calculated without buying anything; Jana only knows her own price once trades have taken place. [1] [2]
Imagine two equally long periods represented by prices of 100000 and 100030 USD/BTC. Both receive the same weight, so the average lies halfway between them. If the first period lasted twice as long, its price would receive twice the weight, producing 100010 USD/BTC. We therefore need more than an arbitrary collection of prices: we must specify what is measured and how much time each observation represents. Equally weighted regular samples are one way to estimate a time average, not a precise record of every price move between them. [1]
With an execution TWAP, Jana sets the total quantity and the period over which it should trade. The service then submits smaller orders. These need not all be identical: Kraken describes regular intervals with randomized slice sizes. The abbreviation alone therefore does not tell Jana what each purchase will look like. She needs the rules of the particular tool. [3]
Kraken distinguishes Market and Limit modes. Market seeks available execution when each slice is submitted; Limit adds a maximum acceptable buy price or minimum sell price. A slice can be skipped if the market does not meet that condition. Jana protects her price constraint but may finish with less bitcoin than planned. The schedule cannot guarantee both the price and the full quantity. [3]
Before starting, it helps to know what the service does with the remainder. Kraken describes canceling it when the time window ends. IBKR offers allowPastEndTime, which permits trading beyond the stated end. This IBKR documentation concerns US equities; it is not a promise of the same feature for bitcoin. Different provider behavior explains why knowing the algorithm’s name is not enough. [2] [3]
Splitting a trade may reduce the pressure a single large order would exert. Yet other buyers can arrive while Jana waits and push the price higher. TWAP is therefore not automatically cheaper than buying immediately and guarantees no particular outcome. Assess the quantity actually traded, its prices, fees and what remained unfilled. A smooth average line does not reveal those differences on its own. [3]
If Jana stops the running schedule, she should not expect earlier purchases to disappear. Kraken says pending orders are canceled, while completed trades remain in the account. Afterwards, distinguish the quantity already bought from the remainder that will no longer be purchased. “Stopped” describes the schedule’s status, not an empty trading history. [3]
A time average needs a clearly defined market, price field and start and end of the period. Another exchange or another window can produce a different result. TWAP weights time, whereas VWAP gives greater weight to prices at which larger quantities traded. Neither is universally the more correct number: they answer different questions. Neither alone tells you whether bitcoin is cheap or its price will rise. [2] [1]
Why a purchase may miss the average
In a fictional model, two equally long periods have representative prices of 100000 and 100100 USD/BTC. The time average is TWAP = (100000 + 100100) / 2 = 100050 USD/BTC. Now suppose 1 BTC is actually bought at the first price and 3 BTC at the second. The average purchase price becomes (100000 × 1 + 100100 × 3) / 4 = 100075 USD/BTC. More is bought at the higher price, so the result exceeds the time average. The example excludes fees, is not a current market quote and predicts no future price. It simply contrasts time weighting with weighting by the quantity actually bought.
For the clearest picture, read this entry together with Volume weighted average price, Market microstructure, Market impact, Slippage, Limit order, Market order. The reverse links also lead from Volume weighted average price.
01Must every smaller purchase be the same size?+
No. A simple schedule can use equal slices, but a service may vary them, skip them because of a limit or leave some unfilled. Checking the result requires actual quantities and prices, not just the number of planned steps.
02What remains after canceling TWAP?+
Completed trades are not reversed. Further execution stops according to the service’s rules; check bought or sold quantities, account balances and any open orders. Cancellation is not a refund for earlier purchases.