The Big Data Edge for Hedge Funds

The term Big Data has been around for some time but questions remain exactly how hedge funds and other buysides exploit the vast stores of data that has never been available to them in the past and continue to grow each day.

The term Big Data has been around for some time but questions remain exactly how hedge funds and other buysides exploit the vast stores of data that has never been available to them in the past and continue to grow each day. Traders spoke with Irene Aldrich, a former prop trader and now managing partner with Able Alpha Trading, a quantitative and high-frequency trading services firm. She shared her thoughts on what Big Data can bring to the buyside and how it has empowered the firms in their relations with their broker-dealers.

Traders: What are you hearing about your clients use of Big Data? Is this marketing hot air or is Big Data an established resource that theyre actively exploiting in their day-to-day trading?

Irene Aldridge: Oh, yes, absolutely. I think there are three major trends when it comes to big data. The first remains just information gathering. Theres a ton of information that is generated by everyone and basically everything is now tradable. Theres tradable news; there is information provided by Bloomberg, Reuters, and Dow Jones; and there are news feeds. Theres also the internet that is like a data mine of information — theres unlimited numbers of stuff.

The number one trend is really to synthesize and gain an edge by having some custom understanding of how the news is incorporated into the market and be the first to synthesize the news from the global universe of news.

Traders: What is the second trend?

Aldridge: Trend number two is more traditional. Its really how you gain alpha. Its more of the high-frequency trading and high-frequency trading portfolio management is how do you respond to new market developments in a very fast time and so intra-day definitely.

Traders: Whats the third trend?

Aldridge: The third trend is really optimal execution in a sense that many buy-side clients are now questioning whether they need a broker or whether they should rely on the broker as much as they did before. The reason for this is they can hire very talented personnel who can develop quantitative models that are, frankly, superior to models that many of the brokers have.

It used to be that brokers used to have an oligopoly essentially where they would control the access to the markets, and they didnt really need superior technology to compete, because all they needed to do was just stay on par with the other guys. So there was always a look left, look left and look right and see what everybody else is doing so they would have to have the same thing at least that everybody else is having, and that was sufficient.

Traders: Are your buyside clients moving away from the broker model?

Aldridge: The clients that Im dealing with will often say, well, its not good for us because it doesnt give us enough edge. So theyre developing in-house what are essentially optimal execution models where if they need to run through a billion dollar transaction, theyre not going to give a block to the broker anymore. Theyre going to break it up into hundred-share little pieces or two hundred-share pieces and give it to all of their seven brokers that they have relationships with in such a way that not one broker can reconstruct what theyre doing.

I used to work for the sell-side and often what happens on the sell-side is, at least in our case, the broker would receive a block trade from some firm, and they would immediately start calling all of their clients trying to generate business and saying, okay, this guys buying this, so you must be buying that too because these guys always know what theyre doing and theyre buying, so, therefore, you all have to do that. And it dilutes the profitability of the original person who gave the block trade.

Its an additional reason why people are now doing optimal execution in-house; at least theyre building optimal execution capabilities. Were talking about the largest hedge funds in the U.S., in the world.

Traders: Theyre acting as their own brokers?

Aldridge: In many cases they still have direct market access, so under the direct market rules, some of them have an affiliated brokerage so that they do it themselves. They have the main operation, and then they have a separate company, which is a brokerage, which they control.

In some cases they do that. In some cases, they just have relationships with ten brokers.

Traders: Is there a big picture trend to Big Data use on the buyside?

Aldridge: I think the overarching trend above all else is secrecy. People are trying to protect their ideas like never before, and because the competition is extremely fast and I think this is a you snooze, you lose kind of environment but on an even faster scale. Peoples desire for absolute secrecy of what theyre doing is underlying all of this.

Traders: You say that the Big Data sources are everything from Bloomberg to Reuters to other market data feeds to the internet, and Im assuming social media like Twitter as well. How are these small hedge funds able to turn on a dime and react very quickly to this data? How do they find a drop of water coming out of this fire hose?

Aldridge: Well, this is where the edge comes in. So everyone develops their own models, most people do not use any commercial software.

The reason for that is because when you buy and bring IT in-house, when your deal is trading millions and billions of dollars and all in a second – even a minute, even an hour – and you bring in some enterprise platform, you dont know whats inside it. If you dont know whats inside it, theres a high probability that when that when something goes wrong, its going to go wrong at the very worst possible moment, and basically you will lose money.

So most people rely on open-source solutions. They dont use anything off-the-shelf unless its extremely trusted and its been there for a long time. But when you talk about the complex event processing (CEP) firms, most of them are not built for fast trading applications. They never have. They come out of these legacy of systems and legacy events, and theyre just not responsive enough. Theyre not fast enough. Theyre these bulky things that are, frankly, just too cumbersome. They just take too much memory. They take too much of everything, of computer power, so theyre not something people rely on.

In most cases what people do is they identify some little phenomenon, and this is all it takes. You dont need to process the universe of events to come up with a trading signal, right? You just need to identify some name that no one else has identified.

Traders: Tell me a little bit about Able Alpha Trading.

Aldridge: Well, I do a bunch of stuff. I dont build trading tools per se. What I build is completely custom trading signals that I write off, and the contracts that I have with people, its not that I only cannot disclose, but they also own all the intellectual property. Whatever I develop for them its theirs.

Able Alpha is more consulting. I have a separate business, which is ABLE Markets where I provide a kind of canned tools that theyre also bought in hedge funds on more of the subscription basis. These include fees for identifying various events, like I have some that process internet news and they deliver signals based on that, and as well as like predicting flash crashes and a bunch of other stuff.

Traders: You dont run a portfolio, do you?

Aldridge: Not at this point, no. I was in prop trading.

Traders: Do you think youll ever return to that in the future?

Aldridge: Yes, Im certainly very interested in that.