Transaction cost analysis, is exactly like its moniker, it analyzes the cost of execution. This intuitively seems simple. If the broker charges a $1,000 for execution, isnt the execution cost $1,000? Well, not really. To accurately measure the cost of execution you need to measure more than broker commission and fees. Execution cost is typically segmented into two major categories: explicit and implicit costs. Explicit costs are fairly easy to measure in a transparent market. They are brokerage commissions and spreads. If the broker charges you $1,000 and you bought at the offer, your explicit cost would be $1,000 plus the spread.

Calculating implicit cost however is much more difficult. Implicit cost is typically split into market impact, delay, and missed trades. Market impact is the affect the order had in the market, or how much a buy order has pushed the market higher / sell order has pushed the market lower. Delay, is the amount the market has moved between the acquisition decision and the execution price and missed trades are the prints (other executions) that you unfortunately were unable to participate in.

If you guessed that in the equity market, commissions were the lions share of the cost, well you would be wrong. Actually, the way Plexus (now a subsidiary of ITG) measures it, the largest aspect of transaction cost in the US Equities market is delay, not commissions, impact, or missed trades. The lost time between the decision to buy and the actual purchase happens to be deadly.

Importance of Data

However, if by just reading this you guessed that transaction cost analysis requires a significant amount of data, you would be absolutely right. Equity TCA requires that you have the universe of your decision-point time-stamped orders, when they were received by the trading desk, how the trader segmented the order, how

.if you are trying to mathematically analyze your FX trading cost down to the gnats eyelash - the data is just not there to do a sophisticated analysis.

orders were executed or delegated, all of the executions associated with the segmented order, and all of the market data associated with every single order in the market, or at least the volume weighted average price of all of the asset that traded that day. This is a significant amount of data. The question is, is this data available in the FX market, can it be developed, and by who? And is this data really important to developing new and maybe more interesting TCA in the FX marketplace.

For the most part, the majority of underlying data is not available for traditional TCA in most over-the-counter marketplaces. There are no unified markets or trading venues in over-the-counter markets, so there is no unified execution tape. The market is not only bifurcated it is tri or even quad-forcated (if that is even a word). There is an inter-dealer market where dealers trade, a credit-intermediated dealer market where hedge funds trade, dealer-to-client electronic RFQ systems to the buy-side, and the phone or single dealer market where the bulk of the dealer to customer business is done. These systems are not linked together and there is no central trade reporting.

Because there isnt a unified market it is difficult to benchmark your trades. Who can definitely say that the price from your dealer was or was not the best? While it may have appeared to be best according to the market data platform, there may have been an RFQ transaction, a dealer to customer trade, or even a phone-based order that may have been better priced. This fragmentation also makes it difficult to measure both impact and missed trades, as there is no way to calibrate your executions against the market.

So the idea of FX TCA looks bad huh? Well if you are trying to mathematically analyze your FX trading cost down to the gnats eyelash - the data is just not there to do a sophisticated analysis.

To more fully obtain accurate TCA information the industry must develop an execution tape, a list of all executions across all execution venues (inter-dealer broker, intermediated inter-dealer broker, multi-dealer RFQ platforms, and dealer to customer transactions) that provides the time and date of the transaction, the product, quantity, and the price with a credit normalization field that allows for consistent data analysis across the counterparty creditworthiness. Credit normalization of FX TCA will be important as unlike equities, FX prices are predicated upon counterparty creditworthiness.

The chance of this occurring in the immediate future is limited. The market is segmented, the dealers and trading platforms have a vested interest in keeping the data private, there is no consistent protocol in which all parties communicate, and there isnt even a regulatory body to force everyone to play nicely in the sandbox.

Improved access to data
However, that said, life is changing. The buy-side, though automated execution platforms and market data providers has greater access to data than ever before. Platforms, such as EBS Prime provides intermediated hedge fund access to the inter-dealer market and platforms such as HotSpotFX and LavaFX enable the buy-side the ability to trade with each other through a two sided marketplace. Technology can grab, capture and store streaming quotes to provide measurable time series. And we are also seeing increasing demand for benchmarks such as EBSs SmoothRate, which is about as close to a VWAP rate for FX as one could hope for in a fragmented marketplace.

That said, firms are also beginning to move away from average price executions and move toward an implementation shortfall approach to measure transaction cost. Implementation shortfall ignores VWAP and benchmarks the order against a decision price, which was the price at the time it was determined to trade the asset. So if the portfolio manager, treasurer, or fund manager decided to buy a currency when it was at a specific level, that specific level would be the benchmarked price, hence acquiring it lower would be positive and acquiring it at a higher price would be counted as a negative. The challenge with measuring trading cost in this manner is that while it does measure how well you did against your order, it does not help you understand the liquidity characteristics in the market and how to better trade that product the next time.

Conclusion
While more sophisticated implementation shortfall methodologies use volume and time and sales data to measure market impact and missed trading opportunities, one needs to start somewhere. And that starting point seems to have occurred. With greater client demand, there will be greater pressure on dealers and execution platforms to open their data, provide unified access, and generate the metrics needed to better understand execution, and only then will firms truly be able to measure trading cost in a sophisticated manner.