The Fee That Routes Your Order

Between the button and the fill, one decision is made that nobody races for and nobody shows you: which venue your order is sent to. It is settled by a published fee schedule, in advance, and on a 300-share order the difference between the best and the worst outcome under those schedules is $1.65, on one side of one trade. That is more than half of what lesson 53 charged you for the whole round trip. The fee is real, it is capped by rule at three tenths of a cent a share, and you never pay it, which is exactly why it decides where your order goes rather than you deciding. Add it to lesson 53’s arithmetic and a penny-wide round trip at 300 shares costs $4.80 rather than $3.00, with the extra 60 per cent invisible on your statement because somebody else settles it. Nothing in that paragraph is about being fast.

Prerequisites: Lesson 53, which priced a penny-wide round trip at 300 shares at $3.00 and counted only the spread, lesson 2, for the queue behind the top line and the 900 shares of 7,300 that the quote actually describes, and lesson 3, for the fills you get not being a fair sample of the fills you could have got, which is the whole of what a routing decision changes.

One round trip, with everything in it

Take lesson 53’s trade again: 300 shares, a stock quoted a penny wide, in and out with market orders. That lesson charged you $3.00 and said explicitly that it was counting the spread and nothing else. Here is the rest of it.

Component	Per share, per side	Per side, 300 shares	Round trip
Half the spread, crossed	0.0050	$1.50	$3.00
Access fee at the Rule 610 cap	0.0030	$0.90	$1.80
Total	0.0080	$2.40	$4.80

Four dollars eighty rather than three dollars, on a penny-wide stock, in the most ordinary trade there is. The fee adds 60 per cent to the crossing cost, and the $3.00 lesson 53 charged you is 62.5 per cent of the true total. That lesson said it was counting one charge of four; this is the second of them.

Two things follow, and they point in opposite directions. The first is that your broker absorbing that $1.80 is a real service, and the reason it can be given away is the arrangement lesson 53 derived: uninformed flow is cheap to quote against, so somebody will pay for it. You are not being robbed by commission-free trading; you are on the receiving end of a subsidy whose size you can now put a number on.

The second is that the subsidy is exactly what makes your routing invisible. A broker deciding where to send your order is choosing between the rows in the first table, and the swing between those rows is $1.65 on a 300-share order, which is larger than the $0.90 the crossing itself costs, because moving between crossing and resting changes the sign rather than the size. The decision is worth real money to somebody, and the only place its consequences show up on your side is in a fill quality you would have to measure to see.

Which is what the last problem below asks you to do, and the reason lesson 53’s third problem asked the same thing from the other direction. The order-routing disclosure your broker publishes every quarter names the venues, the share of orders sent to each, and whether payment was received. Read against the execution-quality report, those two documents answer the only version of this question that is about you.

What this does not settle

That the numbers in the first table are your venue’s numbers. The three tenths of a cent cap is fixed by rule and the rest are typical values from published schedules, chosen mid-range. Real schedules are tiered by monthly volume, differ by security price band, and change several times a year, so a firm doing size pays and receives different amounts from the ones above. The cap is the only figure on that table you can rely on without looking anything up, and the shape of the table is what survives: crossing costs, resting is paid, and inverted venues reverse both.

That the fee schedule is the whole routing decision. It is the part with a published number on it, which is why this lesson can price it. The other inputs are real and mostly unquantified from outside: how much size a venue shows, how often it fills at the midpoint, how stale its quote is when your order arrives, and what the broker’s smart order router does when the first venue only fills part of the order. A page that priced the fee and stopped has priced the visible input and left the invisible ones alone, which is not the same as showing they do not matter.

That the subsidy is a gift. Your broker absorbs the access fee and is paid for the flow, and the two are related in a way this lesson has not measured. Lesson 53 derived that uninformed flow is worth paying for and left the division of the nine tenths of a cent open. This lesson has now put $1.80 of cost on one side of that division for a 300-share round trip, which narrows the question without answering it, because what the wholesaler receives is not published in a form you can set against it.

That queue position is measurable before the fact. The 5,000-shares-ahead arithmetic assumes you can see the queue, and displayed size is not the whole book: hidden orders, reserve size and midpoint interest all sit in front of or beside you without showing. So the 25 trades is an upper bound on how visible your wait is and a lower bound on how long it might actually be, and lesson 2’s 900 shares at the quote were 900 shares that chose to be seen.

That any of this changes what you should do in the next hour. It does not. The routing decision is your broker’s, made in advance, and the only levers on your side are which broker you use, whether you cross or rest, and, on some platforms, whether you direct an order to a named venue yourself. Two of those three are decisions you make once a year rather than once a trade.

And the largest gap is that this lesson has priced the cost of crossing precisely and left the cost of resting exactly where lesson 53 left it, which is unmeasured. The queue arithmetic above says how long you might wait; it says nothing about which waits end in a fill you wanted. A resting order that fills quickly usually fills because the price is coming to you and then through you, and one that never fills has often been saved from a loss. Those two effects work against each other, they are the same adverse selection lesson 53 conceded, and the honest position is that this page has now measured two of the three costs in a round trip and still not the one that would decide whether patience pays.

Problems

1. Read your own broker’s routing disclosure. Every broker publishes one every quarter, naming the venues it sent orders to, the share that went to each, and whether it was paid for them. Find yours and write down the top three venues and the payment column. Ten minutes, and it turns a claim about the industry into a fact about your account. Most people who have opinions about order routing have never opened this document for their own broker.
2. Add the fee to your own cost sheet. Lesson 53 had you compute spread times shares times round trips times 252. Now do it again with three tenths of a cent added to each crossing, twice per round trip. On a penny-wide instrument that raises the figure by 60 per cent; on a wider one the proportion falls, because the fee is fixed per share and the spread is not. Work out where the crossover is for your instrument: the spread at which the access fee stops mattering. Half an hour, and the answer tells you whether this lesson is about you at all.
3. Measure the fill you did not get. For twenty resting limit orders, record the size showing at your price when you posted, whether you filled, and where the price went in the next five minutes. That gives you three numbers: your fill rate, the average queue you were joining, and what the unfilled ones would have been worth. The third is the cost of patience that neither this lesson nor lesson 53 has priced, and twenty observations will not settle it either, but it is the only version of the number that belongs to your instrument and your broker. An evening to set up and a month to fill in.

Sources. U.S. Securities and Exchange Commission, Regulation NMS, Rule 610(c) (2005), for the three-tenths-of-a-cent cap on access fees, which is the one number in the first table above that does not vary by venue or by month. Larry Harris, Trading and Exchanges: Market Microstructure for Practitioners (Oxford University Press, 2003), for order types, queue priority and what a router is choosing between, which is where the queue arithmetic comes from. James J. Angel, Lawrence E. Harris and Chester S. Spatt, “Equity Trading in the 21st Century” (The Quarterly Journal of Finance, 2011), for how a market of many competing venues came to exist and what fragmentation did to the cost of trading. Robert Battalio, Shane A. Corwin and Robert Jennings, “Can Brokers Have It All? On the Relation between Make-Take Fees and Limit Order Execution Quality” (The Journal of Finance, 2016), for the finding that brokers routing to maximise rebates delivered measurably worse limit-order executions, which is the divergence this lesson describes and does not price.

One decision sits between your button and your fill, and it is made by a fee schedule rather than by a race. Crossing costs three tenths of a cent a share at the cap, resting is paid about a quarter of a cent, inverted venues reverse both, and a wholesaler charges nothing because no exchange was used. The swing across those outcomes is $1.65 on a 300-share order, more than half the spread cost of an entire penny-wide round trip, and it belongs to your broker rather than to you. Adding it to lesson 53 makes that round trip $4.80 instead of $3.00. None of it required anyone to be fast, which raises the obvious question: what does speed actually buy, and for whom. Lesson 55 puts a number on it and the number is smaller and stranger than the folklore suggests.

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