Jeff Bacidore, who joined ITG as head of algorithmic trading nearly a year ago, has created algos on the buyside and on the sellside. After seeing both sides of the fence, he’s found there’s little difference between the two.
Prior to joining ITG, Bacidore was head of algorithmic trading on the buyside at Goldman Sachs Asset Management. While his current job has many similarities to the last one, he said building algos for the sellside is somewhat different, since he has to make sure they are flexible enough to appeal to a broader audience.
Fortunately, he has the experience to help him understand the varying needs of different traders. Prior to working for GSAM, Bacidore was a vice president on the sellside at Goldman Sachs. He had previously been a research director at the New York Stock Exchange, and actually began his career as an academic, earning a Ph.D. from Indiana University where he did his thesis on decimalization.
His approach in developing algos is to be quantitatively driven, relying on trading data to determine effectiveness rather than simply trying to build an algo that mimics the typical actions of a human trader.
"Some providers view algorithms as just being like robots, and say, ‘As a trader, if I were in this situation, this is what I would do,’" Bacidore said. "We’ll actually study what is-according to the data-the best thing to do."
Recently, Bacidore spoke with Traders Magazine about some of the latest developments in algorithmic trading and how trade data is helping to make algos better. That’s something that he has found rewarding, describing his tenure so far at ITG as "fun."
Traders Magazine: Are smart order routers overtaking algos as the new hot area of trading technology?
Mr. Bacidore: Across the Street, the focus has been more on the micro-level stuff, like the smart routing order placement, because as market structures evolve, you really have to constantly change that, whereas the scheduling you don’t have to. But I don’t think that all the good work’s been done, especially on the portfolio trading aspect of it, where I think people are just scratching the surface. The initial pass of just incorporating portfolio risk into the algorithm, a lot of the players have put that in. Where people are competing now is to make that logic a little bit smarter, so it’s not just risk minimization-it’s actually a much more thoughtful process.
TM: How else are firms developing smarter algos?
Mr. Bacidore: One area is measuring execution quality. We’ve been seeing a lot more interest recently on quantifying the performance of the algorithms.
TM: Don’t most firms want to use an independent trade cost analysis?
Mr. Bacidore: They’d like to have an independent third party quantify all of it, but for practical reasons, a lot of buyside firms aren’t able to collect the data on such a granular level as the algo providers do, so we’ll provide a supplemental report that will show on a much finer level of detail things like where the orders are executing and adverse selection reports and things like that. People are using a lot of third parties for compliance reasons, but I think they’re relying on the brokers to get that real detailed, granular, deep dive into the algorithms. You can try to do that through a third party, but there are so many layers in between the execution and getting the data to the third party, that sometimes it’s better to just work directly with the broker, and then use the third party to make sure everything’s above board and consistent.
TM: Any other advantages to getting the information from the broker?
Mr. Bacidore: There’s also timing. If a client came to us and said, "Tell me about a trade yesterday," we could provide them with detail. Even on trade day, in a lot of cases, we provide them with a great level of detail, whereas if you went to a third party, that’s just generally not the cycle, especially if you want to do something custom.
TM: Where else are we seeing innovation with algos?
Mr. Bacidore: Where you’re seeing a lot of the innovation is more in the futures and FX. With FX, people will often refer to it as the Wild West of trading, just because there’s not as much regulation to fall back on in terms of protections across markets. With that said, it’s a very liquid, deep market, and even though it’s relatively new to this space, it’s ripe for the picking.
TM: What about algos specifically designed for illiquid small-cap stocks?
Mr. Bacidore: Our approach has never been to have one algorithm for a small cap and one algorithm for a large cap. It’s really been, let’s have one algorithm, and the algorithm will decide how to optimally trade that within the context of that name. Traders have a lot of brokers, a lot of different algorithms, and one thing they struggle with is when to use which algorithm. Our view is not to broaden out the suite. It’s really to make a set of strategies. So if someone says I want to trade implementation shortfall, they really just choose one algorithm, and we’ll decide the special handling if it’s illiquid.
TM: We’ve been hearing a lot about changing tick sizes, either exchanges offering sub-penny pricing or certain stocks being allowed to trade at nickel spreads. Would that affect algos?
Mr. Bacidore: With a tick of one penny, if you have a stock that’s trading at a $1.50, that’s still a very, very wide tick. Our algorithms are attuned to handle those types of environments. And then you have cases like Google, where a penny tick is virtually nothing. So if regulators came to us and said they’re going to make the tick bigger or smaller, it’s just the number of stocks that are going to be in that bucket now.
