Trading results can deteriorate even when a strategy generates valid signals. This guide shows how to audit trading execution quality by measuring arrival price, spread cost, slippage, fill rate, partial fills, rejection rate, latency, and market impact. Use the formulas, order-type comparison, and trade-log template to separate strategy weakness from broker or execution costs.
Overview
Execution quality is the difference between the trade your strategy intended to make and the trade that actually reached the account. A backtest may assume an entry at a signal price, while a live order can experience a wider spread, delayed routing, partial execution, rejection, or a fill at a less favorable price. These differences matter to discretionary traders, algorithmic systems, day trading bots, and swing strategies alike.
The first objective of an execution audit is not to blame a broker. It is to classify the cost correctly. A losing trade may reflect a poor signal, an unfavorable market regime, oversized order, weak liquidity, or avoidable execution friction. Reviewing execution separately from strategy performance makes the diagnosis more reliable.
Start with a sufficiently consistent sample, such as trades from the same strategy, instrument group, order type, and time-of-day window. Compare live results with the strategy's intended entry and exit, but do not treat a theoretical backtest price as automatically achievable. A realistic backtesting process should model spreads, commissions, latency, and imperfect fills where the data permits.
Useful execution questions include:
- How far did the fill move from the price available when the order decision was made?
- How often was the order fully filled, partially filled, rejected, or canceled?
- Which order types and time windows create the highest costs?
- Are costs large enough to change the strategy's expectancy?
How to estimate
Use a consistent benchmark for every trade. The most practical starting point is the arrival price: the bid, ask, midpoint, or last available quote recorded when the trading decision or order submission occurred. Record which benchmark you used because comparing an immediate market buy with the midpoint and comparing it with the ask answers different questions.
Core formulas
For a buy order, execution slippage per share can be estimated as:
Buy slippage = Fill price − Benchmark price
For a sell order:
Sell slippage = Benchmark price − Fill price
Both formulas produce a positive number when execution is less favorable than the benchmark. Convert the result to basis points when comparing instruments with different prices:
Slippage in basis points = Slippage per share ÷ Benchmark price × 10,000
The estimated dollar cost is:
Slippage cost = Slippage per share × Filled shares
For a round trip, calculate entry and exit costs separately, then add them to commissions, exchange or regulatory fees where applicable, spread costs, and any borrowing or financing costs relevant to the account. Avoid counting the same spread twice. If the benchmark is the midpoint, the difference between the midpoint and the fill captures the adverse half-spread plus price movement and market impact. If the benchmark is the ask for a buy, it may better isolate movement beyond the displayed offer.
Measure fill quality, not only price
Price alone can hide operational problems. Calculate:
- Fill rate: filled quantity ÷ submitted quantity × 100.
- Partial-fill rate: orders with some, but not all, quantity filled ÷ submitted orders × 100.
- Rejection rate: rejected orders ÷ submitted orders × 100.
- Cancellation rate: canceled or expired orders ÷ submitted orders × 100. Separate trader cancellations from broker or venue expirations.
- Latency: time from signal generation to order submission, submission to acknowledgment, and acknowledgment to fill. Keep these intervals separate where timestamps are available.
For marketable orders, compare the fill with the available quote at order submission. For passive limit orders, also record opportunity cost: the price improvement from a favorable fill may be outweighed if the order misses a move and must be replaced at a worse price. That makes fill rate and time-to-fill essential parts of order execution analysis.
Inputs and assumptions
A useful execution log should contain enough detail to reproduce each calculation. Export broker records where possible, then join them with the strategy or bot's event log. A practical field template is:
| Field | Purpose |
|---|---|
| Trade and order ID | Connects signals, orders, fills, and cancellations. |
| Symbol, side, and quantity | Identifies the instrument and whether the order bought or sold. |
| Signal timestamp | Shows when the strategy decided to trade. |
| Submission and acknowledgment timestamps | Separates software or network delay from broker processing. |
| Order type and time-in-force | Allows comparison of market, limit, stop, and bracket behavior. |
| Bid, ask, midpoint, and last price | Creates a benchmark at decision or submission time. |
| Limit or stop price | Shows the instruction the broker was expected to follow. |
| Fill price, filled quantity, and fill timestamps | Supports weighted-average price and partial-fill analysis. |
| Status and reason code | Distinguishes filled, rejected, canceled, expired, and replaced orders. |
| Spread, volume, and liquidity notes | Provides context for market impact and unusually poor fills. |
| Commission and other recorded costs | Completes the all-in trading-cost estimate. |
Use the same market-hours definitions throughout the sample. Label regular-session, premarket, and after-hours trades separately rather than combining them. Also separate highly liquid symbols from thinly traded names, and distinguish opening, midday, closing, earnings, and major-news windows. A stock scanner or trading bot may produce signals in all of these conditions, but the execution assumptions are not interchangeable.
Order-type comparison
| Order type | Primary execution benefit | Main execution risk to measure |
|---|---|---|
| Market | High likelihood of immediate execution when liquidity is available. | Slippage, spread cost, and market impact. |
| Limit | Controls the maximum buy or minimum sell price. | Non-execution, partial fills, and opportunity cost. |
| Stop | Activates after a defined trigger condition. | Gap or fast-market slippage after activation. |
| Bracket | Combines entry and protective or profit-taking instructions. | Trigger sequencing, cancellation behavior, and differing fills across legs. |
Broker terminology varies. Confirm how the platform labels submitted, acknowledged, working, triggered, partially filled, canceled, rejected, and expired orders. A status shown in a user interface may not have the same meaning as an API event, so use the broker's current documentation when reconciling records.
Worked examples
Assume a strategy submits a buy order for 500 shares. At the decision point, the midpoint is $20.00 and the ask is $20.02. The order receives two fills: 300 shares at $20.03 and 200 shares at $20.05.
The weighted-average fill price is:
((300 × 20.03) + (200 × 20.05)) ÷ 500 = $20.038
Using the midpoint as the benchmark, midpoint slippage is $0.038 per share, or $19.00 before commissions and other costs. In basis points, the estimate is approximately 0.038 ÷ 20.00 × 10,000 = 19 basis points. The fill rate is 100%, but the order still experienced a partial-fill sequence and adverse price movement between fills.
Using the ask as the benchmark produces a different result: $20.038 − $20.02 = $0.018 per share, or $9.00. Neither benchmark is automatically correct. The midpoint comparison estimates the full distance from the market's center; the ask comparison asks how much execution deteriorated beyond the displayed offer. Record both only if you explain the distinction and apply it consistently.
Now consider a 500-share limit buy at $20.00. The order fills 250 shares at $20.00 and the remaining 250 shares expire. The fill rate is 50%, and the unfilled quantity must be analyzed separately. If the strategy later buys the missing shares at $20.12, the additional $0.12 per share is an opportunity cost associated with non-execution, not ordinary slippage on the original fill. This distinction helps determine whether a more aggressive limit, a marketable limit, or a different participation rule is appropriate.
After calculating each trade, group results by order type, symbol liquidity, time of day, side, and strategy setup. If poor fills cluster around earnings gaps or unusual news, the issue may be market conditions rather than a persistent broker defect. Readers reviewing an earnings-driven system can pair this audit with an earnings gap risk-management workflow.
When to recalculate
Run an execution review at least monthly for active systems and immediately after a material change. Recalculate when the broker changes its fee schedule, order-status fields, routing options, API behavior, market-data feed, or supported order types. Repeat the analysis after changing a trading bot, broker connection, hosting location, position size, limit-offset rule, or order-submission logic.
Also revisit the audit when market conditions change. A strategy that performs acceptably in liquid regular-session trading may incur materially different costs during premarket, after-hours, opening volatility, closing activity, or a sharp news event. Segment the new sample instead of relying on an old average.
Use this monthly workflow:
- Export orders, fills, fees, quotes, and strategy timestamps.
- Reconcile every order ID and investigate missing or duplicated events.
- Calculate slippage, spread cost, fill rate, partial-fill rate, rejection rate, and latency.
- Group results by order type, session, liquidity, symbol, and strategy.
- Compare live execution with the assumptions used in the backtest or trading plan.
- Test one operational change at a time, such as a limit offset or maximum order size.
- Document the result and retain the prior period for comparison.
Execution quality should inform, not replace, strategy analysis. If costs are rising, review position sizing and risk limits alongside routing and order rules; the position-sizing guide can help connect per-trade exposure with execution risk. For algorithmic systems, maintain a separate audit trail for signal quality, implementation quality, and broker performance. That separation makes it easier to decide whether to improve the strategy, change the order logic, compare brokers, or simply avoid conditions where the expected edge is too small to overcome trading costs.