What it means
A quoted stock price is not necessarily the price at which an entire order can be filled. Transaction cost analysis (TCA) compares execution with a stated reference price and expresses the difference in money or basis points. It can be done before a trade, using estimates, or after a trade, using actual fills.
The benchmark matters. An arrival price captures the price when an order decision is recorded; a midpoint benchmark uses the middle of the quoted bid and ask at a specified time. Different benchmarks answer different questions and can produce different cost figures for the same fills.
TCA may examine spread, market impact, commissions and other fees. A broader measure, implementation shortfall, can also account for shares that were not filled. State what is included before comparing results.
How it is calculated
For a fully filled stock order, one simple signed, benchmark-relative measure is:
Cost (US$) = s × Q × (P_fill − P_ref) + F
sis+1for a buy and−1for a sell; it has no units.Qis the number of shares filled.P_fillis the share-weighted average fill price, in US$ per share.P_refis the benchmark price, in US$ per share, observed at the specified time.Fis included commissions and fees, in US$.
A positive result is a cost relative to the benchmark; a negative result is a saving. To express it in basis points, use Cost (bps) = Cost (US$) ÷ (Q × P_ref) × 10,000, where Q × P_ref is the order’s benchmark value in US$ and one basis point equals 0.01%.
This formula gives an all-in benchmark difference when fees are included. It does not, by itself, separate spread from market impact or isolate a price move that would have happened without the order. Partial fills require an explicit treatment of unfilled shares.
Worked example
Suppose, hypothetically, an investor buys 100 shares. The recorded arrival price is US$50.00 per share, the share-weighted average fill price is US$50.10, and included fees are US$1.00.
- Price difference: US$50.10 − US$50.00 = US$0.10 per share.
- Difference across 100 shares: 100 × US$0.10 = US$10.00.
- Add fees: US$10.00 + US$1.00 = US$11.00 of cost against the arrival-price benchmark.
- Benchmark value: 100 × US$50.00 = US$5,000. Cost in basis points: US$11.00 ÷ US$5,000 × 10,000 = 22 bps, or 0.22%.
These are illustrative fills, not an estimate for any real stock. Using another benchmark could change the measured cost.
Why it matters before you trade
Pre-trade TCA asks what execution might cost under stated assumptions, rather than treating the last traded price as a guaranteed fill. Order size relative to daily volume, estimated spread, volatility and expected market impact can all affect that assessment. The estimate helps make execution conditions and cost assumptions explicit; actual costs remain uncertain until execution.
Post-trade TCA instead uses recorded fills to compare what happened with the chosen benchmark. Keeping estimated and realized costs separate makes it possible to inspect differences without treating a pre-trade estimate as a promise.
How PreTrAIde uses it
PreTrAIde performs pre-trade analysis on a user-built stock basket before any order is placed. Its deterministic quantitative layer estimates bid-ask spread, participation in daily volume and market impact using an Almgren-Chriss-style model, alongside other risk and execution measures. It produces execution scores and returns a GO, NO-GO or MODIFY verdict with a suggested execution strategy and plain-language reasoning for the basket and each line.
Those inputs relate to estimated transaction costs; they are not a report of achieved fills. PreTrAIde does not receive, transmit or execute orders. Its verdicts support the user’s decision and are not trade recommendations.
Common mistakes
- Omitting the benchmark or its timestamp: a cost figure is ambiguous without both.
- Mixing estimates with actual fills: pre-trade and post-trade TCA use different information.
- Counting the same cost twice: an all-in fill-versus-benchmark difference may already reflect spread and impact; adding separate estimates can double-count them.
- Ignoring unfilled shares: a filled-shares-only formula does not capture the full effect of a partial fill.
- Comparing unlike figures: check whether each result includes fees and uses the same benchmark and units.