Transaction Cost Analysis (TCA)

Transaction cost analysis (TCA) is the practice of measuring how much it cost to execute a trade or a set of trades, and of judging whether the execution was good. It works by comparing the prices actually obtained with one or more benchmark prices, and by breaking the difference down into its causes. The results are normally expressed in basis points of the value traded.

TCA is used chiefly by an institutional investor such as an asset manager or pension fund, whose orders are large enough to move prices, and by the brokers that serve them. It began in equity markets and has spread to foreign exchange, futures and bonds. Its purposes are to control each transaction cost, to compare brokers and algorithms, and to provide evidence of compliance with best execution duties.

• Pre-trade. Before an order is sent, a model estimates its likely cost and risk from its size, the security's liquidity and volatility, and the time available. The estimate guides the choice of execution strategy.
• Intra-trade. While the order is being worked, progress is monitored against the benchmark so that the approach can be adjusted.
• Post-trade. After completion, realised costs are calculated, compared with the pre-trade estimate and with peers, and attributed to decisions such as timing, venue and order type.

The choice of benchmark defines what is being measured.

• Arrival price. The market mid price at the moment the order reaches the trading desk or broker. Cost against arrival, summed with the cost of any unexecuted portion, is implementation shortfall, a measure proposed by André Perold in 1988.
• VWAP. The volume-weighted average price of all market trades over the period of the order. It shows whether the order did better or worse than the average participant.
• TWAP. The time-weighted average price over the period.
• Open or close. The day's opening or closing price, relevant to funds valued at the close.

For a buy order, cost against a benchmark is:

• $P_{\text{exec}}$ is the average price at which the order was executed.
• $P_{\text{bench}}$ is the benchmark price.

For a sell order the sign is reversed, so that a positive number always means a cost.

Suppose a fund buys 100,000 shares. The arrival price was $25.00 and the average execution price was $25.06. The cost against arrival is $(25.06 - 25.00) / 25.00 \times 10{,}000 = 24$ basis points, or $0.06 \times 100{,}000 = 6{,}000$ dollars. Over the same period the market VWAP was $25.08. Against VWAP the cost is $(25.06 - 25.08) / 25.08 \times 10{,}000 = -8$ basis points, a saving. The same execution looks poor on one benchmark and good on another: the price rose while the order was being worked, and the order bought slightly below the average.

Beyond the headline number, TCA separates the components of cost. Spread cost reflects the bid-ask spread paid. Market impact is the price movement caused by the order itself. Timing cost arises from price drift during a delay or a slow execution, and opportunity cost from shares that were never bought or sold.

Results are grouped by broker, algorithm, venue, trader, order size and market conditions. Analysts look at where orders were sent by order routing systems, how much was filled in dark pools, and how prices behaved immediately after each fill. A price that consistently moves against the order just after execution suggests that the order was detected by other traders.

Rules on best execution make some form of TCA a practical necessity. In the European Union, MiFID II requires investment firms to take all sufficient steps to obtain the best possible result for clients and to monitor the effectiveness of their arrangements. In the United States, brokers must regularly review the quality of execution they obtain for customer orders. Asset managers use TCA reports to demonstrate oversight to clients and regulators.

• Benchmark weaknesses. A large order is itself part of the VWAP, so the benchmark moves with the order and flatters it. VWAP can also be met by trading passively while the price runs away.
• Noise. The cost of a single order is dominated by market movement unrelated to execution quality. Meaningful conclusions need many orders and adjustment for conditions.
• Incomplete orders. Measures that ignore unfilled quantity reward traders for abandoning difficult orders.
• Data. Accurate timestamps and complete market data are required, usually from a tick database, and are harder to obtain in markets without a consolidated record of trades, such as bonds.
• Incentives. Traders judged on one benchmark may trade to that benchmark at the expense of the fund's actual return.

• European Securities and Markets Authority
• Financial Industry Regulatory Authority (FINRA)