Schwab’s first two expanded Rule 605 summaries show persistent order-size gaps in price improvement

Charles Schwab’s expanded Rule 605 summaries for August and September 2026 show a recurring difference in the share of executions receiving price improvement between the smallest and largest order-size categories. Across the four combinations of S&P 500 membership and order type examined, the gaps ranged from 37.16 to 43.32 percentage points in August and from 34.04 to 41.37 points in September.[1][2]

The calculation compares the same published categories across two months. It does not prove that order size alone caused the difference, that any execution breached a duty or that a competing broker would have delivered a better result. It does show why an aggregate execution score can obscure material differences among the orders underneath it. The central contribution is a reproducible reconstruction from the broker’s own monthly summary files.[1][2]

What the primary files measure

The report archive provides downloadable CSV summaries as well as other report formats. This review uses the August and September summary files, matching the S&P 500 indicator, market or marketable-limit order type and order-size category. The smallest category is below $250 and the largest is $200,000 or more. Those are order-size bins, not the market value of a customer’s account.[1][2]

Schwab’s explanatory page describes the price-improvement percentage as a share-weighted execution measure. It should not be rewritten as the percentage of customers who received a better price or a simple count of orders improved. Share weighting allows a large execution to contribute differently from a small one. The report’s price-improvement concept concerns execution relative to the applicable quoted benchmark, rather than the eventual profitability of the investment.[3]

The summary also separates securities in the S&P 500 from other securities. That division matters because the instruments represented can have different spreads, trading activity and liquidity. It is a grouping variable in the comparison, not a complete control for security characteristics. No additional round-lot-only filter is imposed: this analysis uses the summary as published, rather than a third party’s separately normalised subset.[1][2]

The gap persists in each of the four groups

For S&P 500 market orders in August, the share receiving price improvement was 97.1219% in the below-$250 category and 56.5612% in the $200,000-and-above category. Subtracting the latter from the former gives 40.5607 percentage points. In September, the corresponding figures were 97.4366% and 58.0506%, a gap of 39.3860 points. That is a percentage-point difference, not a 39% change in execution quality.[1][2]

S&P 500 marketable-limit orders show a similar pattern: the gap was 43.3171 percentage points in August and 41.3708 in September. Outside the S&P 500, the market-order gaps were 39.3543 and 38.5930 points; marketable-limit gaps were 37.1645 and 34.0403 points. All four comparisons retained the same direction across the two reporting months.[1][2]

The gap narrowed in every selected group from August to September, by between about 0.76 and 3.12 percentage points. That descriptive change is too limited to establish a continuing improvement trend. Two months can have different securities, market conditions, order sizes within the bins and customer instructions. The persistent direction is observed; its cause and future path remain open.[1][2]

Why larger orders can produce different results

A larger order can exceed the quantity available at a particular quoted price and interact with several levels of liquidity. A small order can more readily fit within available displayed or non-displayed liquidity. These are plausible market-structure mechanisms, not explanations demonstrated for the particular orders in these files. The summary does not expose each order’s contemporaneous order book or routing sequence.[3][4]

The two endpoint bins are also wide. An order just above $200,000 and a much larger order share the top category; the distribution inside that category can change between months. The comparison fixes the published bin labels, not the precise order values or securities. A causal study would need a more closely matched set of orders and relevant market conditions.[1][2]

Market orders and marketable-limit orders should remain separate. A limit price can constrain where and how much an order can execute, and that can affect the measured outcome. Combining their percentages without appropriate weights would mix different instructions. The fact that the pattern appears in both categories is useful corroboration of the descriptive result, rather than permission to average the four gaps into a new universal score.[1][2]

Price improvement is only one execution dimension

The summaries provide additional fields, including execution time and effective-spread measures. Those fields answer different questions from whether an execution received any improvement. A high improvement frequency can coexist with a small improvement amount; a lower frequency can coexist with some larger improvements. One percentage cannot establish the full economic quality of execution.[3]

Schwab’s own guide cautions that order-flow characteristics affect comparisons. That counterevidence is part of the interpretation, not an exception to be hidden after a ranking. This article does not compare Schwab’s aggregate score with another firm’s score because matching order type and size alone would not control for all relevant security, timing, customer and routing differences.[3]

The selected summary files do not provide the order-level observations needed here to estimate the statistical uncertainty of the gap or identify customer-specific outcomes. The reported percentages can be reproduced arithmetically without that further inference. No confidence interval, significance claim or estimate of customer money lost is calculated from these summary rows.[1][2]

How the calculation can be checked

For each month, the method selects the row for each S&P 500 flag and order type at the smallest size bin, then subtracts the price-improvement share in the corresponding largest bin. It retains the precision published in the CSV before rounding the prose. It does not add overlapping summary categories or infer missing observations. This matters because the archive also contains an aggregate row covering orders below the largest threshold.[1][2]

The August S&P 500 market example is therefore 97.1219 minus 56.5612, or 40.5607 percentage points. Repeating that operation for September gives 39.3860. The same operation yields the other six results shown in the table. All figures belong to their stated reporting months; they are not current quotes or a prediction of the next trade’s execution.[1][2]

The expanded disclosure regime gives readers more categories to inspect than a single headline average. Its practical value depends on preserving those definitions when presenting the results. A change in reporting format is not itself evidence that execution improved or worsened, and an expanded report cannot be compared casually with a differently scoped older series.[5][6]

What further reporting would add

A longer series could test whether these category gaps persist across market conditions and whether their magnitude changes with the distribution of securities and order values. Order counts and closely matched order-level evidence would help assess robustness and causes. Neither a sustained gap nor a narrowing one should be treated automatically as a verdict on best execution without that wider context.[1][2]

For now, the established finding is bounded and consequential for how execution claims are read: in both published months, the largest order category had a materially lower share of price-improved executions than the smallest within each group examined. Readers can reproduce that difference. They should keep it attached to the report’s definitions, rather than convert it into a universal broker or customer-outcome score.[1][2]

Report an error or submit an update