TCA – NO ARGUMENT?

FX markets may lack the transparency of equity markets, but the asset class is gaining momentum. As a result attention is turning to transaction cost analysis to achieve best execution. Best Execution talks to Jim Cochrane* about the challenges.

What are the reasons that asset managers use TCA for Foreign Exchange now and will that change in the future?

Foreign Exchange TCA already has a past, present and future: compliance, best execution and process improvement. Of these three, compliance is the most basic (the past), best execution the most popular (the present) and process improvement the most advanced (the future). Peer group and pre-trade analysis will also be a greater part of the future as asset managers pay more attention to foreign exchange. The market will demand analytical systems that measure trades done electronically as well as those traded in more traditional venues such as over the phone or through custodians. Therefore, FX TCA will evolve to not only answer basic questions but will also change with advances in electronic trading such as the use of algorithms. However, the initial phase of FX TCA is still focused on compliance and best execution analysis.

Our firm plans on creating analytics for more advanced FX trading styles, as well as more sophisticated measurements for compliance, best execution and process improvement. Compliance measures can be developed in the future to include not only measures of daily weighted averages but also intraday measures of weighted averages tailored to a customer’s specific trading hours and various intraday benchmarks. It can also be flipped to measure trading performance within the daily or intraday ranges.

Asset managers rely heavily on best execution analysis, measuring trade performance using flexible benchmarks to exact timestamps. This service is extremely versatile, and is limited only by the creativity of the client and the FX TCA provider. It provides an excellent overview of trading costs by currency pair, counterparty, trade size and any other metric provided by the customer. The future of precise best execution analysis is founded in data. It is highly dependent on precise, copious, multi-source data and flexible benchmarks. As a consequence, we have invested heavily in market data to provide the best platform on which to base TCA.

Process improvement helps to measure implementation shortfall within the investment process. Once again, this is only limited by the timestamps that an asset manager can provide through their order management and execution management systems. Examples of these measurements are execution rate versus the portfolio management decision time or the order arrival time to the trader and FX broker. This analysis is the ultimate use of backward looking FX TCA. It provides the asset manager with the data necessary to make positive changes to the investment process.

With regards to pre-trade analysis, banks and trading platforms are beginning to offer this service in simple forms with very basic execution analysis. In the future, traders and portfolio managers will be provided online tools to pick the times of high liquidity and low volatility. FX TCA will help them decide which algorithm to use as well as measure the performance of that execution versus an estimate of market slippage if they had performed the trade live over the phone or in one order on a trading platform. As electronic trading reaches full penetration, algorithmic trading will become as popular in FX as it is in the stock market. Additionally, clients are demanding peer group analysis of their transaction costs. Along with pre-trade measures, these are the next logical steps in any TCA program.

On what do you base your vision of FX TCA evolution?

ITG is a market leader in equity TCA. We have been instrumental in the creation and advancement of all aspects of transaction cost analysis, working with the largest asset managers in the world. In the short time that FX TCA has been sought by the buy-side, those same steps we witnessed in the equity markets are being repeated in the FX market. Compliance review led to best execution measurement which in turn led to process improvement. At the same time, our clients demanded peer group analysis and pre-trade analytics. While we are not recreating the wheel, we are certainly following the same path to discovery. The next step in analysis will be peer group analysis and pre-trade TCA.

What is the biggest challenge to analysing foreign exchange transactions?

There are two: Accurate timestamps and precise data. Without an exchange, it is not possible to know what the official price is in the market for foreign exchange transactions, especially large or exotic transactions. Having a large set of data is the best way to overcome that handicap. The other complication is that if you do not have a precise timestamp for the trade, then the analysis will have to be directed away from precision and onto trends. Negotiated trading and custodial fills are the bane of accurate timestamps. Moving from negotiated trading over the phone to electronic and algorithmic trading can solve this challenge. Refusal to produce timestamps is the last refuge of some custodians. Client pressure on the banks, and perhaps regulation, are ways to solve that issue. The data problem is less difficult to master. We have solved it through negotiation and purchase of a very expensive set of data.

Most FX TCA products analyse spot trades. Is it possible to measure other transactions, including derivatives?

Absolutely. Prices for spot, forwards, swaps and non-deliverable forwards are available from multiple sources. Being able to analyse the data and apply it properly is the only difficulty provided you have the brains trust to develop a system to make the data match the execution, whether it is a cash trade or a derivative trade. FX TCA for options will also be available in the future but will require excellent data and a powerful option pricing engine.

What is the difference between trade cost analysis and execution analysis?

Execution analysis focuses on the trade in near-real time. It gives immediate feedback for the trader that he received the best price in the market given his pricing stream. It can also measure how much he gave away to choose another provider in order to manage liquidity and allocate trades across his board of FX providers. What it does not provide is a look at the monthly or quarterly trends across currency pairs, counterparties and trading styles. It shows you that you chose the best price among the prices that the banks on your trading platform allow you to see. A complete trade cost analysis shows you your trade executions across a long time period against all of the prices available in the market.

Do you worry that basic analysis will be overlooked?

Yes and no. In the beginning I feel that many institutions are merely asking for a “check in the box” solution to satisfy their fiduciary responsibilities. However, as I stated earlier, even basic analysis will reveal trends and raise questions. For example, a client might take satisfaction that they trade within the daily range, but question why they never outperform the daily average. That will lead them to demand better execution and they will need to measure this change in focus. Natural curiosity and a desire for higher returns will lead them to ask more penetrating questions regarding even the most basic analysis.

What market variables will real time analytics measure?

Real-time and near-real time analytics will measure volatility and liquidity in the first phase. Volume will not be possible to measure in the FX market. Liquidity will be estimated using bid/offer spreads. These measures challenge any analytical system because they are highly dependent on a system of opaque data. Once these measures are mastered, they immediately bring up the idea of predictive analysis. Will real-time measures help us predict transaction costs with great accuracy or the direction of the market? The latter has not been achieved yet in the equity market but the former has and can be attempted in the FX market.

Can you measure the difference in performance between a zero cost execution and an algorithmic trade?

Yes. That is possible if you have full depth of book in your data set and a precise timestamp. A zero cost execution would match the best available price, taking slippage into consideration. A 100mm trade in any currency pair would not get done at top of book, but a good system can account for that if the database is thorough. That adjusted price can be used as a benchmark for an algorithmic trade. A concurrent trading algo can be measured against the benchmark trade just described. Algos can be measured against a variety of benchmarks if your TCA system is flexible. It can be measured against the mid-rate at the time of execution, the best price (top of book) or the best expected price based on the depth of book. Each benchmark contains trade cost information which the client can use to evaluate their performance. ITG TCA for FX can tailor the system to meet any of the client’s analytical questions.