tradeinformatics

Empire State Building

350 Fifth Avenue, Suite 4500

New York, NY 10118

 

info@tradeinformatics.com

© 2019  tradeinformatics

EVALUATING BROKERS

How do you optimize broker selection?
A reasonable question without industry consensus.  Does a broker wheel or broker scorecard help?  The real answer is less a better scorecard or fancier switching engine and more for the buy-side to take back control of the execution process itself.

Broker Wheels and Scorecards

The idea of a broker wheel has been popularized over recent months as a way to answer this question – how do you optimize broker selection?  With a broker wheel, the trader sets rules to adjust flow allocation based upon predefined criteria, perhaps the results of scorecard.  This misses the point. The problem is not that certain strategies are in any way substantially better than others – in our experience they all grade quite similarly under comparable conditions – it’s that it’s impossible for commoditized strategies to perfectly fit every use case.

 

Broker scorecards, on the other hand, can be revealing - knowing which brokers follow instructions and which don’t is invaluable. Yet, while the scorecard is an effective way to gather feedback and monitor performance, it doesn’t go far enough toward effecting positive change. In practice, the buy-side already holds much of the responsibility for broker performance, but they’re typically not in a position to effect change, even if their scorecard tells them what the problems are. Traders may try to do what they can with what they have, but in the end it’s difficult to fit a square peg in a round hole.

 

A Framework for Evaluation

 

The first step toward understanding what makes an execution strategy effective is to acknowledge that a broker’s performance is in large part predefined upstream by the portfolio manager and buy-side trader. While it’s true that every order is different and every broker’s interaction with the market is unique, that uniqueness is overwhelmingly a function of upstream order attributes and timing decisions. Any evaluation mechanism must therefore acknowledge that many aspects of broker performance are in effect predetermined by the circumstances of their orders, and evaluation must focus on how a broker performed relative to the instructions and circumstances they were given.

The goal of evaluation then is not “How good is my algorithm?” but rather “Did my algorithm do what I told it to do?”  Answering this question requires employing metrics that correspond with the use case and urgency of the buy-side trader sending the orders. It also means comparing an execution strategy or a broker against other strategies or brokers that tend to be employed in similar situations. TI believes a bracketed broker scorecard, like in the table below, is the best practice for generating a clear evaluation mechanism incorporating these methodologies.

Useful Measures for Evaluation

 

While many performance measures help us contextualize broker performance, we highlight three such measures here and in the scorecard above because they’re applicable in almost every use case.  It’s worth our emphasizing that this list isn’t exhaustive or applicable in every use case.

 

First, we measure performance versus an interval volume-weighted average price. Were the prices received by the broker better than those of other market participants?

 

Second, we look at performance against a participation weighted price corresponding to the aggressiveness of the strategy. Did the broker follow instructions with regard to aggressiveness? 

 

Finally, because every fill carries with it a degree of information leakage we look at fill size. Higher fill sizes, normalized into average print size as a percent of average daily volume, indicate an ability to receive more back than gets signaled away.

The Takeaway

 

In the end, the answer is to transition much of the execution intelligence from the sell-side to the buy-side, giving the buy-side trader or portfolio manager the ability to dynamically and systematically adjust execution parameters in response to both real-time observations and historical data analysis. Returning to the original question then – how do you optimize broker selection - the real answer is less a better scorecard or fancier switching engine and more for the buy-side to take back control of the execution process itself.