What slippage is

The gap between the request and the fill 

Between reading a quote and receiving a fill, a sequence of things happens and each of them takes time. The instruction leaves the device, crosses a network, arrives at the firm, is checked, and is matched against whatever interest is available at that instant. The market did not pause for any of it. Prices in a heavily traded instrument change many times a second, so the quote that was read and the quote that exists at the moment of matching are two separate observations, and a fill can only ever be the second one.

Slippage is the difference between those two observations. It is measured in the instrument's own increment, and it becomes money through the contract size in exactly the way a spread does. It is not a rate, not a fee and not a term of an account. It is the record of a price that changed during an interval measured in thousandths of a second.

A second and quite different situation produces the same measurement. An order larger than the interest resting at the best price cannot be filled entirely there, so the remainder meets the next level. The market did not have to move at all for that to happen. Both are reported as slippage because both are described by the same subtraction.

It runs in both directions 

The sign of the difference is not fixed. A price that moved while an instruction was in flight moved in whichever direction it was going to move, and the instruction meets it there. Where the new price is worse than the one that was read, the fill costs more than the arithmetic anticipated, and that is the case most readers already have in mind. Where it is better, the fill costs less, and the conventional name for that case is price improvement. The two are the same subtraction with opposite signs.

The symmetry in that block is structural, and it is not a statement about how often each case occurs. Nothing in the mechanism prefers one sign. Which price a given order meets depends on which way the instrument moved in the interval, and no property of the order decides that.

Not a spread, and not a commission 

Three things reduce what a trade returns, and they are different kinds of thing. Stating the difference precisely is what keeps a later comparison honest.

• A spread is a property of a quote. It is visible before an instruction is sent, it is met once on entry and once on exit, and its direction never varies, because a transaction always meets the far side of the pair.
• A commission is a charge. It is a published rate applied to a stated basis, debited as its own entry with a date and an amount, and its size is known before a position opens.
• Slippage is neither. It has no published rate, no entry on a statement and no fixed direction. It is a property of the interval between an instruction and its execution, and it exists only after the fact.

The consequence is practical. A spread and a commission can be added together in advance; slippage cannot. A cost stated before a trade is a statement of terms, and what a trade turns out to have cost is a measurement taken afterwards. A later lesson in this module assembles the terms into a single round turn, and this is why that assembly is the known part of the cost rather than the whole of it.

What widens the gap 

The size of the difference is decided by two quantities: how fast the price is moving, and how much interest is resting near it. Several conditions move one or both.

• Speed of movement. Volatility is the rate at which a price changes, and slippage measures a change across an interval, so the two are directly connected. The faster a price is moving, the further it travels in the same fraction of a second.
• Depth of resting interest. Liquidity is the quantity available at and near the current price. Where there is a great deal of it, an order of ordinary size is absorbed close to the level it asked for. Where it is thin, the same order reaches further.
• Scheduled releases. In the seconds around a data release or a policy decision, quoting parties withdraw and replace their interest rapidly, so both quantities move at once, and both move the way that widens the gap.
• Unscheduled events. A headline removes resting interest faster than it can be replaced, with no warning in any calendar.
• The size of the order relative to the interest resting behind the quoted price, which the next section takes in isolation.
• The order type. An instruction that asks for immediate execution accepts whatever price is available. An instruction that names a price cannot be filled outside it, which converts the possibility of a worse price into the possibility of no fill at all.

When the order is larger than the price 

A quoted price holds for a quantity, and that quantity is finite. An instruction for more than the amount resting at the best level is filled in parts: as much as is available there, then the remainder at the next level, and so on until it is complete. What is recorded is one average price, weighted by how much filled at each level, and that average is worse than the best level by construction. No price moved at any point in the sequence.

This kind of difference is a property of size rather than of speed, and it can occur in a completely still market. It is also why every set of contract specifications states the size a quoted price holds for: an order beyond that size is not the transaction the quoted figure describes.

Gapping 

A gap is the extreme case of the same mechanism. When a market reopens after a break, or when an event removes every resting bid and offer across a range of levels, price moves from one level to another without trading in between. An instruction resting inside that range is not skipped. It becomes executable at the first price that exists on the other side of it, which can be some distance from the level named in it.

Requotes, the other answer to the same fact 

When the price has changed by the time an instruction arrives, a firm has two structurally different responses. One is to execute at the price now available and report it, which is slippage as described above. The other is to decline the instruction and return the current price for acceptance or refusal, which is a requote. They distribute the same fact differently: the first resolves the order and leaves the price uncertain, the second fixes the price and leaves the execution uncertain.

Practitioners disagree about which arrangement is preferable, and the disagreement is genuine. One tradition holds that a requote returns control of the price to the sender, since nothing executes without a second acceptance. Another holds that a requote spends time in precisely the conditions where time is what moved the price, so the returned quote can itself be stale, and an order meant to be immediate can go unexecuted through the whole of a fast move. A third arrangement, a stated maximum deviation from the requested price, caps how far a fill may land from the request and, by that same setting, makes no fill possible once the market has moved beyond it. Each converts one uncertainty into another, and none removes both.

Why it cannot be quoted in advance 

A spread and a commission are terms. They are set by a firm, published, and true until the firm changes them. Slippage is set by nobody. It is the outcome of a race between a message and a market, and its size on any order depends on the state of the book at an instant that had not yet arrived when the order was sent. Stating it in advance would mean describing a future that no party to the trade controls.

What can be measured and published is the execution process rather than its result on any one order. YAL reports an average execution speed of 19 ms and fill quality of 99.6%. Speed narrows the interval in which a price can move, and infrastructure is the part of the mechanism a firm can act on. What the market does inside that interval is not.

Where practitioners disagree 

The first argument is whether slippage belongs in a cost figure at all. One tradition includes it, on the grounds that the distance between the price that prompted a decision and the price that was realised is a cost whatever produced it. Another excludes it, on the grounds that it is market movement rather than a term of business, and that a measure mixing a firm's charges with the market's behaviour cannot be compared between firms. The two are answering different questions, which is why both survive.

The second concerns whether measured slippage is symmetric in practice as well as in construction. Orders are not sent at random moments. They cluster around the events that also move prices, so a measured distribution can be skewed even where the mechanism producing it is not. One reading attributes such a skew to the mechanism, another to when orders happen to be sent, and no published average separates them. This lesson takes no position on it, and no figure here settles it.