What Is a Fill or Kill Order?
Fill or kill order is an instruction to execute the entire quantity immediately or cancel the order completely. It permits no partial execution and no resting in the book, which makes it a tool for taking a known block of liquidity without revealing intent.
Most order types are about price. Fill or kill is about completeness — it exists for situations where half a position is not half as good, but actively bad.
How it works
Submit the full quantity with a fill-or-kill instruction. The venue checks whether the whole amount can execute immediately at your price or better.
If it can, it does, in one go. If it cannot, the entire order is cancelled — not reduced, not queued, cancelled.
It never sits waiting. The decision is made in the instant the order arrives, which is the property that distinguishes it from every order type that waits.
Why leaving no trace matters
A resting order is public information. It sits in the order book, visible, telling everybody that somebody wants to trade a particular size at a particular price.
And that information is tradeable against you. Others can position ahead of a large visible order, which makes the rest of your execution worse.
A killed order told nobody anything. It probed for liquidity, found none, and vanished — which is the whole reason institutions use the instruction.
A worked example
A trader needs 50,000 shares as a single block to hedge an existing exposure. Half a hedge leaves them with a directional position they did not want and now have to manage.
The book shows 50,000 available at 100.05 or better. A fill-or-kill order at 100.05 takes all of it in one execution.
Now suppose the book only shows 30,000. The order is killed. The trader owns nothing, has revealed nothing, and can try a different venue or a different approach.
Compare that to a plain market order. It would have taken the 30,000 and then walked up through worse prices for the remaining 20,000, leaving a half-hedge at a bad average and a visible footprint.
The order types it sits beside
Immediate or cancel fills whatever is available right now and cancels the rest. It accepts partial execution; fill or kill does not, and that single difference is the whole distinction.
All or none demands the full quantity but is willing to wait for it. It accepts resting; fill or kill does not.
So fill or kill is the intersection of both constraints — full size and immediate — which is why it fails more often than either. You are asking the market for two things at once.
And that failure rate is the point. A killed order costs nothing but a message, and the information that the liquidity is not there is genuinely worth having before committing to a worse execution.
The original data
On this site’s shared series a round trip costs 0.0098, about 2% of the median bar range of 0.493. The ninetieth percentile bar is 1.101 and the largest is 2.338.
That cost applies to the size at the top of the book. An order large enough to need a fill-or-kill instruction is an order large enough that the quoted spread understates what it will actually pay.
And direction runs average 2.01 bars with a longest of 11. An order that must execute now, in full, is betting that the current moment is better than the next one — and over a two-bar average run, that is a bet about very little.
Who actually uses it
Hedgers with a fixed requirement. An exposure of a specific size needs an offset of a specific size, and a partial offset is a new problem rather than a smaller version of the old one.
Arbitrage between two instruments. If the trade only works when both legs execute, a partial fill on one leg is an outright position nobody wanted.
And algorithms probing venues. A sequence of fill-or-kill orders across multiple venues is a way of asking “is the size here?” without committing anywhere until the answer is yes.
Almost nobody buying shares for a personal account needs it. For an ordinary order, a partial fill is perfectly acceptable and the instruction’s only effect is to make the order fail more often — which is worth stating plainly, because the order type sounds decisive and choosing it for that reason is a mistake.
And the venue decides what it means. Exchanges differ on whether a fill-or-kill order may sweep several price levels to complete or must fill at one price, and on how long “immediate” is in practice. Two brokers routing the same instruction to two venues can produce different outcomes from identical inputs, which is why the definition is worth checking rather than assuming.
When it fails
The characteristic failure is using it for ordinary trading. It reads as the disciplined choice — all or nothing, no half measures — and applied to a normal-sized order it simply produces cancellations. The size was always available in pieces, the pieces were always acceptable, and the instruction rejected them on a principle that had no bearing on the situation. The trader ends up repeatedly resubmitting an order that a plain instruction would have filled on the first attempt.
A second failure is confusing it with all or none, which is willing to wait and behaves completely differently.
A third is assuming a kill means there was no liquidity. It means there was not enough at your price at that instant, which is a much narrower statement.
A fourth is not knowing whether your broker supports it. Many retail platforms do not offer it, or translate it into something else.
And a fifth is repeatedly resubmitting after a kill. Each attempt is itself a signal, and a rapid sequence of them reveals more than the single resting order you were avoiding.
Related
Order types covers the full set and what each instruction gives up. Good till canceled order covers the opposite instruction — one that waits indefinitely. And order book covers the visible record this order avoids entering.
Almost nobody trading a personal account needs this order type, and it is worth understanding anyway — because the reason it exists explains what large orders are actually afraid of, which is being seen.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.