The Signal Was Identical. The Fill, Fees, and Tax Window Were Not.

Two accounts can act on the same chart pattern, in the same instrument, within the same second, and close the month with different numbers. The conventional explanation is psychological — one trader held, the other flinched. That explanation has the disadvantage of being unfalsifiable, which makes it useless as a diagnostic.

A more tractable version is that the entry signal is one input among several, and most of the others are set away from the chart entirely. Four are worth separating: the price at which the order filled relative to the quote at the moment it was received, the fees attached to the sale side, the position size that converts a price move into a dollar result, and the calendar rule governing whether a realized loss is usable in the year it was taken. Three of the four are defined in published federal rules. Whatever is left over after they are measured is the part psychology has to account for.

Figure 1. Four components that separate two accounts acting on the same entry signal.

The Fill Is a Published Statistic, Not an Impression

The quality of a fill is not a matter of opinion. The Securities and Exchange Commission defines it and requires it to be reported. Under 17 CFR 242.600(b)(8), the average effective spread is the share-weighted average of effective spreads, where an effective spread for a buy order is double the difference between the execution price and the midpoint of the national best bid and offer at the time the order was received. The reference point is the midpoint at receipt — not the price the trader had in mind, and not the last print on the tape.

Two related measures sit alongside it. An order is executed with price improvement when a buy fills below the national best offer in force at receipt, per 242.600(b)(45). The average realized spread under 242.600(b)(13) applies the same doubling arithmetic against the midpoint at a specified interval after execution, separating the cost of crossing from the price move that followed; Rule 605 requires it at five checkpoints — 50 milliseconds, 1 second, 15 seconds, 1 minute and 5 minutes. The E/Q ratio, average effective spread divided by average quoted spread as a percentage, compresses the whole thing into one number.

Figure 2. Reference points used by the execution-quality statistics required under Rule 605. Source: 17 CFR 242.600(b) and 17 CFR 242.605.

These statistics are reported under 17 CFR 242.605. The rule was amended in 2024, and the amended version reaches considerably further: it covers odd-lot orders of 1 to 99 shares and orders for less than a single share, requires timestamps at millisecond rather than one-second precision, and breaks execution speed into buckets beginning below 100 microseconds. It also extends reporting to what the Commission calls larger broker-dealers — those introducing or carrying 100,000 or more customer accounts, roughly 85 firms covering more than 98 percent of customer accounts.

The timing matters for anyone trying to use this. The compliance date for the amended rule was originally December 14, 2025, and the Commission extended it to August 1, 2026. Because reports must be made available within one month after the reporting period ends, the first reports under the amended rule — covering August 2026 data — are due to be publicly available by September 30, 2026. Until then, the published record remains the older, narrower format.

A separate rule covers where the order went rather than how it filled. 17 CFR 242.606 requires a quarterly public report on the routing of non-directed orders, naming the top ten venues plus any venue receiving 5 percent or more of the flow, and disclosing the net aggregate payment for order flow received. Under 242.606(b)(1), an individual customer may request the venues to which their own orders were routed over the six months prior to the request.

Two Fee Lines That Scale on Different Things

Two charges attach to the sale side of a US equity transaction, and they scale on different quantities. The first comes from Section 31 of the Securities Exchange Act. The statutory obligation runs from the self-regulatory organizations to the Commission, based on the aggregate dollar amount of covered sales; the SROs then charge their members, and members commonly pass the charge through. The rate is not a constant — the Commission resets it against its appropriation.

For fiscal 2026 the rate is $20.60 per million dollars of covered sales, effective April 4, 2026. For the eleven months before that it was $0.00. That is not a rounding artifact: the Commission set the rate to zero on May 14, 2025, having already collected its full fiscal 2025 appropriation. A strategy running in February 2026 paid nothing on this line; the same strategy in May 2026 paid $20.60 per million sold. Two traders separated by a calendar quarter are, on this component alone, not running the same trade.

Figure 3. Section 31 fee rate per $1 million of covered sales at each effective date, fiscal years 2021 to 2026. Source: U.S. Securities and Exchange Commission, Section 31 Transaction Fee Rate Advisories.

The second charge is FINRA's Trading Activity Fee, and it scales on shares, not dollars. Under the fee schedule FINRA filed with the Commission in SR-FINRA-2024-019, the rate for covered equity sales is $0.000195 per share with a maximum of $9.79 per trade for 2026, rising on a published schedule to $0.000232 per share in 2027 and $0.000249 per share, capped at $12.50 per trade, by 2029. The options rate for 2026 is $0.00329 per contract.

The per-share basis has a consequence that catches people out. Consider two sales of the same notional value, $102,000 — one in a $50 stock, one in a $5 stock.

Sale of 2,000 shares at $51	Sale of 20,400 shares at $5
Section 31 at $20.60 per million: $2.10	Section 31 at $20.60 per million: $2.10
TAF at $0.000195 per share: $0.39	TAF at $0.000195 per share: $3.98
Sale-side regulatory total: $2.49	Sale-side regulatory total: $6.08

Identical dollar exposure, roughly 2.4 times the regulatory cost. The per-trade cap does eventually bind — at $9.79 and $0.000195 per share, the TAF stops scaling above about 50,200 shares in a single sale — but below that threshold the share count is the whole story.

Both numbers are small relative to the spread. On the $50 stock with a one-cent quoted spread, a fill at the offer rather than the midpoint costs half a cent per share, or $10 on 2,000 shares — four times the entire sale-side regulatory bill, and it applies on both sides of the round trip. At 200 round trips a year, the gap between consistently crossing and consistently getting midpoint fills runs to roughly $4,000, against about $500 in Section 31 and TAF combined. The fee line is the one the trader sees. The spread line is the one Rule 605 was written to expose.

Six Cents of Fill Rewrites the Position, Not Just the Entry

A fill is not only a cost; it is the anchor from which every other level in the trade is measured. Move the anchor and the geometry of the position moves with it.

Take a plan built before entry: buy at $50.00, stop at $49.50, first target at $51.00. Risk is $0.50 per share, reward is $1.00, reward-to-risk is 2.00. Now suppose the order fills at $50.06 instead — six cents, about twelve hundredths of one percent of the price, well inside the range a marketable order can travel in a fast tape.

• Distance to the stop becomes $0.56, so capital at risk on 2,000 shares rises from $1,000 to $1,120 — a 12 percent increase that was never sized for.
• Distance to the target falls to $0.94.
• Reward-to-risk falls from 2.00 to 1.68, a 16 percent reduction in the payoff structure produced entirely by execution.

The trader who sized on a $0.50 stop is now carrying more risk than the plan specified in exchange for less reward than the plan specified. Holding dollar risk at $1,000 instead means cutting the position to about 1,785 shares, and the same price move produces a smaller result. There is no version where the fill is neutral: it either changes the risk or it changes the size.

This is also how two traders who genuinely followed the same rules diverge without either making a mistake. One order rests and fills at the midpoint; the other crosses the spread. Both entered on the same bar, but the positions are not the same position, and every decision that follows is being made from a different starting point. Attributing the eventual difference to discipline mislabels an execution outcome as a character trait.

The Calendar Keeps Measuring After the Position Closes

Two structural rules operate after the exit. The first is settlement. Under 17 CFR 240.15c6-1(a), a broker or dealer may not enter into a contract for the purchase or sale of a security providing for payment of funds and delivery of securities later than the first business day after the date of the contract. The standard cycle has been T+1 since May 28, 2024, which means proceeds from a sale become available one business day later rather than two — shortening the interval before the same capital can be committed again.

The second is the wash sale rule, and it has the longest reach. 26 U.S.C. 1091(a) disallows a claimed loss where, "within a period beginning 30 days before the date of such sale or disposition and ending 30 days after such date," the taxpayer has acquired, or entered into a contract or option to acquire, substantially identical stock or securities. Counting the day of the sale, that is a 61-day window centered on the disposition.

Three properties are worth stating precisely, because they are commonly garbled:

1. The disallowed loss is not destroyed. Under 26 U.S.C. 1091(d), it is carried into the basis of the replacement position, which defers it rather than eliminating it.
2. The window runs in both directions. A purchase made 25 days before the loss sale can trigger it just as a purchase 25 days after can.
3. IRS Publication 550 notes that acquiring substantially identical stock in an individual retirement arrangement, including a Roth IRA, counts as an acquisition for this purpose.

The rule applies to losses, not gains. Two traders who exit at a loss on the same day and re-enter the same setup at different times can therefore end the tax year with different deductible amounts from identical trading decisions — one outside the window, one inside it.

This is a description of what the statute says, not tax advice. Whether two particular instruments are substantially identical, how a specific account structure is treated, and how any of it applies to an individual return are determinations for a qualified tax professional. The reason to know the rule exists is to know when to ask.

What Would Invalidate This

This decomposition has a defined operating range, and it fails outside it.

• Long holding periods dissolve it. Six cents of slippage against a several-dollar move held for months is noise. Execution cost matters in proportion to turnover.
• Rule 605 reports describe venues, not people. The statistics are monthly aggregates across all customer orders at a market center or broker, segmented by security, order type and size. Reading a firm's E/Q ratio as a prediction of an individual's fills is a category error — it describes a distribution, and a single order is one draw from it.
• A cost model built on today's rates expires. The Section 31 rate has been $5.10, $22.90, $8.00, $27.80, $0.00 and $20.60 within roughly five years, and the TAF is on a published escalation through 2029. Any spreadsheet that hard-codes either is wrong by the next fiscal year.
• Account structure can null out a component. In a tax-deferred account the wash sale analysis does not operate as it does in a taxable one, because the loss was not deductible to begin with. Non-US traders face a different fee and tax structure entirely.
• The residual may be the whole story. If two traders exited on different criteria — one at a fixed target, one on a trailing rule — that is a strategy difference, not an execution difference. The decomposition is run first so this case can be identified, not to explain it.
• Below one tick, the measurement stops resolving. In a liquid penny-spread name traded in small size, the fill difference between two competent traders may be smaller than the minimum increment. The component is real there but not actionable.

Concrete Framework

1. Record the quote, not only the fill. For each entry, log the national best bid and offer at the moment the order was sent alongside the execution price. Without the reference point, effective spread cannot be computed after the fact — and it is the one execution statistic the regulator has defined.
2. Compute the fill's own E/Q. Twice the distance from the midpoint, divided by the quoted spread. A run near 100 percent means the strategy is consistently paying the full spread; below it means some flow is getting price improvement. Average it over at least 30 trades before concluding anything.
3. Recompute reward-to-risk from the actual fill. Do it at the moment of entry, while the position can still be adjusted. If the realized ratio falls below the threshold the plan required, the choice is to resize or stand down — not to hold the original size and hope.
4. Separate the two fee bases in the log. Carry Section 31 as a rate per million of sale proceeds and the TAF as a rate per share. A strategy that migrates toward lower-priced instruments raises one and leaves the other flat.
5. Re-check the Section 31 rate each fiscal year. The Commission publishes a fee rate advisory, and the rate has changed on dates in February, April and May in different years. Note the current rate and its effective date rather than assuming last year's number.
6. Pull the routing report once a quarter. The Rule 606 disclosure names the venues receiving the flow and the net payment for order flow associated with it. Under 242.606(b)(1) an individual account can request its own routing for the prior six months.
7. Mark loss exits on a 61-day calendar. Record the sale date and the boundary 30 days either side. Treat the calendar as a flag that a question exists, and route that question to a tax professional rather than answering it from a trade log.
8. Run the residual last. Subtract the measured execution, fee and sizing effects from the observed difference in results. What remains is the part that discipline or judgement has to explain, and it is usually smaller than it looked before the arithmetic was done.