“And buy a decent suit. You can't come in here looking like this. Go to Morty Sills, tell 'em I sent ya.”
-- Gordon Gekko, Wall Street, 1987.
Broker algo offerings have essentially converged to a common set of “off-the-rack” algos – VWAP, Arrival Price, Liquidity Seeking, Dark Aggregation, etc. For a given algorithm type, the overarching strategy or objective is usually identical across brokers. For example, VWAP algorithms aim to minimize deviations from the VWAP benchmark by tracking a historical volume curve. Arrival price algorithms aim to minimize slippage to the arrival price, subject to some aggressiveness level chosen by the trader. Of course, there are real differences across broker algorithms. But these differences arise mainly in the implementation of the strategy, not in the underlying objective of the common algorithms.
You can have any color you want...as long as it's black....
Consequently, buyside traders really do not have as much flexibility as it may seem in finding the best strategy when selecting among “off-the-rack” algos. Traders often wind up choosing an algorithm not because the algorithm’s objective matches their own, but rather because that algorithm most closely mimics the behaviors most closely aligned with their own high-level objectives.
For example, VWAP algorithms were originally designed to track a historical volume curve as a means of minimizing deviations from the VWAP benchmark. But many of the heaviest users of VWAP algos, like quant funds, are unconcerned with VWAP as a benchmark. Rather, they use the VWAP algorithm because it does a good job of lowering execution costs, as it can be effective at maximizing spread capture and reducing impact by forcing the algorithm to spread the order out over time. But how many traders’ true optimal strategy is achieved by tracking a historical volume curve?
Arrival price algorithms provide another example. Traders benchmarked to the arrival price often gravitate to arrival price algorithms because these algorithms aren’t rigidly bound to an arbitrary schedule and can be more opportunistic than, say, VWAP algorithms. But do Arrival Price users always want their orders front-loaded to reduce variation around the arrival price? Or might they be willing to trade at a more even pace – or even backload when nearing the close – whenever doing so would achieve a better execution?
I'd rather light a candle than curse your darkness...
Fortunately, brokers are increasingly offering clients the ability to customize their algorithms to meet their client’s specific objective. Once a service provided only to top accounts, algo customization is becoming more mainstream, as broker platforms have become increasingly scalable as well as much more functional. And to be clear, what we’re talking about here is not the simple tweaking of an existing algorithm or some simple set of trader-defined algo switching rules passed off as a “customization”, as had been the case in the past. Rather, we’re talking about brokers allowing clients to more clearly define the core objective and the underlying behavior of the algorithm, i.e., true algo customization.
In addition, some brokers have invested in more robust analytic frameworks to help assess the value of these customizations by providing clients detailed post-trade analysis, including breakdowns of how the various components performed, how often certain tactics were engaged, etc. Many brokers even allow clients to run controlled “experiments”, where the buyside can test out multiple variations of an algo side-by-side and then compare performance. The broker and client can then try other enhancements to improve performance even further. Customization makes this performance feedback loop much more valuable than before, as the number of possible improvements has grown dramatically.
With greater choice comes greater responsibility
But the ability to customize has increased the burden on buyside traders to ensure best execution. In the past, a buyside trading desk could rationalize using a VWAP algorithm by noting that the VWAP algorithm’s performance was the best among the “off-the-rack” algos at trading passively and reducing execution shortfall. But now the obvious question becomes “Have you tried to develop an algorithm that can trade passively like VWAP, but perhaps be less rigid and more opportunistic when appropriate?” And it’s no longer enough to say, “We compared Arrival Price algorithms across brokers and determined the Top 3 brokers were X, Y, and Z”. The response will likely be “But have you tried creating a custom algorithm, perhaps working with Brokers X, Y, and Z, that can do even better?”.
Some firms have already accepted the challenge, hiring quants to work with the brokers to customize the algos, develop an experimentation framework, and analyze performance, often in the form of an “algo wheel”. Small and midsize firms have generally outsourced some or all of this work to consultants, like The Bacidore Group (yet another shameless plug!). But regardless, customization has afforded the buyside a tremendous opportunity to further enhance their execution quality. All that is required of the buyside is to engage their brokers and to invest modest resources to analyze the resulting data.
Of course, it may turn out that the “off-the-rack” algorithms do as well as the customized algorithms, or even better. But you’ll never know how good a custom suit feels if you always buy off-the-rack…
“And use a decent algo. You can’t keep trading like this. Go to your broker, tell ‘em I sent ya.”
-- Jeff Bacidore, The Bacidore Group, 2019.
The author is the Founder and President of The Bacidore Group, LLC. The Bacidore Group provides research and consulting services to buyside and sellside clients, including assistance with broker algo customizations, "algo wheel" design and analysis, custom TCA frameworks, custom algo development, and bespoke quantitative analysis.
Copyright 2019, The Bacidore Group, LLC. All Rights Reserved.