Buy-side institutions are formalizing their approaches to transaction cost analysis (TCA), as complexity of markets requires greater precision in calibrating trading strategies.

One of the major decisions with regard to TCA is whether to perform it in-house or relay on a third party.
“We do everything ourselves,” said Nick Nielsen, head of quantitative trading at Marshall Wace. “Our funds are very, very high in turnover. A single basis point of savings will likely amount to savings of about $40 million to $50 million per year. So, the marginal improvement makes quite a bit of impact.”

Marshall Wace is an equity long/short hedge fund that employs a process-driven and fundamental-driven trading strategy across multiple geographies and time horizons.

There are essentially three models for TCA, according to Nielsen. The first, the so-called allocation approach, involves parceling out chunks of an order to different brokers and comparing their performance.

“You could potentially do this with a number of brokers, provided that you have a similar set-up with all of them, meaning that he data is formulaically derived very similarly, and the order flow is handled similarly,” he said.

“With the allocation approach, you’re looking at how a broker performs against the expected market impact,” said Nielsen. “We normalize those statistics for market-wide impact estimates, in order to remove alpha. We look at the entire market tick database to get those market impact estimates. When you look at market-wide statistics, alpha becomes a zero sum game.”

The second model for TCA is to compare performance against a specific benchmark, such as VWAP. “You might give someone a VWAP order and compare their performance against the actual n=benchmark,” said Nielsen.

The third model is for the firm to place the orders itself, and measure the results. “We do a lot of order placement ourselves, and look at the fill rates from the different counterparties,” said Nielsen.

In liquidity seeking, the mounting challenge is to access liquidity in a predictable and quantifiable manner.

“This challenge is ever growing as more and more of the available liquidity moves into the dark,” said Travis Felker, vice president of research and development at Cyborg Trading Systems.

“In response, several tools have been developed that can aid in achieving reliable results, such as parallel posting, dark heat maps, genetic routing tables, and IOIs to name a few,” said Felker. “Increasingly, the knowledge of where liquidity lies and how deep it runs is held at a premium, and widely coveted.”

TCA, to be effective, must take in a sufficient number of data points to provide an accurate picture. A real-time or snapshot approach is therefore unsuitable.

“Real time TCA is irrelevant,” Nielsen said. “You can’t make decisions based on a single data point. You need to have a statistically significant sample.”

Among the world’s largest institutions—those managing more than $20 billion in assets—72% of using TCA employ a third-party vendor in cash equities, according to a Greenwich Associates study.

Among the largest responding institutions, that share approaches 80%. Approximately 40% of institutions say they employ a proprietary, in-house TCA tool for cash equities, and roughly the same share use TCA products provided by brokers.

In contrast, 62% of institutions using TCA in fixed income rely on proprietary in-house systems, with only 38% reporting the use of a third-party platform and 10% using a broker-provided system.

The lack of proven third-party systems is even more apparent in FX, in which only about 30% of TCA users employ a third-party vendor and around 70% use in-house systems. In futures, meanwhile, more than 85% use a proprietary system for TCA.