Slippage and Impact at Retail Size

What getting filled costs you is set by your size against the size resting in front of you, which is why the same money spent at the same instant is nearly free in one instrument and ruinous in another. Below the size at the touch you pay the spread and nothing more; above it the cost per share starts climbing, and it climbs faster than your size does. The same $30,000, at the same instant, costs $0.30 in one of the two instruments below and $80.00 in the other: a tenth of one per cent of the money at risk against 26.7 per cent of it.

Prerequisites: Lesson 2, for a book and what walking it costs, and lesson 10, for why slippage is the one charge of the four you have to measure rather than look up.

The same money, two names

You have $30,000 to put to work and two candidate instruments. Same idea, same instant, same market order.

Name A is a large-cap index ETF at $500, quoted 499.99 bid, 500.00 ask, with 5,000 shares resting at the offer. Your $30,000 buys 60 shares — 1.2 per cent of the touch. Every share fills at 500.00. Against the midpoint of 499.995 that costs half a cent a share, or $0.30 on the whole order.

Name B is a $30 small cap, quoted 29.95 bid, 30.00 ask, and the offer side of its book looks like this:

Price	Shares resting	Your order takes
30.00	300	300
30.05	300	300
30.10	400	400

Your $30,000 buys 1,000 shares here — 333 per cent of the touch. The fill averages 30.055, and against the midpoint of 29.975 that is 8 cents a share, or $80.00. Split it: $25.00 is the half-spread you would have paid at any size, and $55.00 is the walk, which exists only because you were bigger than the 300 in front of you.

Same money, same second, same instruction: $0.30 against $80.00. And now put both against the risk, which is the only comparison that decides anything. Give each a stop 1 per cent of price away — $5.00 a share in A, $0.30 a share in B — and both trades risk exactly $300. The entry alone has spent 0.1 per cent of that risk in Name A and 26.7 per cent of it in Name B.

Lesson 10 put the whole four-part bill for a trade at 7.7 per cent of risk and moved the breakeven win rate by three points. Name B’s entry alone, before commission, before financing, before the exit, is three and a half times that.

What this does not settle

That the book you can see is the book you will get. Displayed size can understate what is really there, because a large order can be shown a slice at a time, and it can overstate it, because a resting order costs nothing to cancel and most of them are cancelled. Telling the two apart is the subject of lesson 26, and either way the depth you measured a second ago is not a promise about the fill you get now.

Nor that the square-root law is your law. It was estimated on institutional orders taking meaningful fractions of a day’s volume, where the permanent half of impact does the work. Your 60 shares of Name A are a vanishing fraction of that day’s volume, and at that scale the formula returns a number so small it is meaningless beside the spread. The useful part for you is not the curve, it is the boundary: the touch size, which you can read off a screen.

And none of it says the answer is smaller orders. Splitting a purchase into three clips reduces the walk on each, but it leaves you exposed for longer while price moves for reasons that have nothing to do with you — and the cost of that exposure is not on any of these tables. That trade-off, which is the real content of execution, belongs to lesson 57, not to this one.

Every figure here also assumes you cross. A limit order pays no walk and no half-spread, which makes the whole of Name B’s $80.00 avoidable in principle — at the price lesson 3 put on it, which is the fills you do not get and cannot see. This lesson prices one of two choices properly and does not price the other one at all.

And the comparison rests on a stop this page chose for you. Giving both names a stop at 1 per cent of price is what made their risk equal at $300 each, and it is an assumption rather than a finding. A stop belongs where the trade is wrong, and if Name B genuinely needs 3 per cent because it moves three times as much, its risk becomes $900 and the same $80.00 entry falls from 26.7 per cent of it to 8.9. The gap between the two instruments is real. Its size depends on a decision lesson 21 makes and this one borrowed.

Slippage is not a property of you and it is not a property of your broker. It is your size divided by somebody else’s, and you can look up the denominator before you trade.

Problems

1. Double it and check the law. Name B’s book continues with 500 shares at 30.15 and 500 at 30.20. Buy 2,000 shares instead of 1,000 and give the average fill, the total cost against the midpoint, and the cost per share. Then compare the cost-per-share ratio you got against the 1.4 the square-root law predicts for a doubling, and say what the difference tells you about this particular book.
2. Find your own boundary. On the instrument you actually trade, record the size resting at the best bid and the best offer at ten scattered moments during a session. Take the median of the twenty numbers and multiply it by the price. That figure is the position size below which your slippage is only the spread, and above which you are buying the walk. Write it down. It is the boundary every figure in this lesson is measured against.
3. Price one idea twice. Take a trade you would genuinely put on and cost it in the most liquid instrument that expresses it and the least liquid one you would consider, both at the same dollar size and the same percentage stop. Give the entry cost as a share of the risk in each, as the worked example does. If the two answers are far apart, you have just found a decision you were making without noticing.

Sources. Robert Almgren, Chee Thum, Emmanuel Hauptmann and Hong Li, “Direct Estimation of Equity Market Impact” (Risk, 2005), which fits the cost of real orders against the fraction of daily volume they take and finds an exponent near one half. Ananth Madhavan, “Market Microstructure: A Survey” (Journal of Financial Markets 3, 2000), for the separation of temporary from permanent impact and why the two need different models. Nicolo Torre and Mark Ferrari, Market Impact Model Handbook (BARRA, 1997), the practitioner document that put the square root into general use.

You can now price the same idea in two instruments and get different answers for the right reason. The next lesson turns that into a ranking, and applies it to everything you might trade rather than to the two you happened to think of.

Educational only. Trading involves substantial risk of loss. Not financial advice. Past performance does not guarantee future results.