Vincenzo D'Alto for the Toronto Star
Case Studies

Betting big on stocks in an unusual way: Montreal’s OODA mines twitter data for market success

Vital Stats
Jake Hirsch-Allen: 647.262.9202,
Years Active:

Hugues Demers, a computer scientist, spent years working for defense contractor Lockheed Martin as a data analyst. In 2008, when the company’s research and development department was shut down, Demers and some colleagues struck out on their own. Together they founded OODA, a small company focused on what the collection of computer scientists, physicists and mathematicians knew best – winning defense and aerospace government contracts.

“The thing is, that was our client, the government, and that’s it,” says Demers. “From that observation, we decided we wanted to diversify.”

The Montreal-based company decided to make a foray into the financial sector, and at the end of October, launched a program that mines social media for stockbrokers. Called MarketSense, the software promises to give a “Twitter psychological analysis” which can predict changes on the Dow Jones index. The mood gleaned from the social network will help traders know if there is a “bullish” or “bearish” pressure on the market.

It’s part of a growing trend of programs mining big data. It entails analyzing large amounts of information, usually digital and often taken from the internet — such as social network communications, customer transactions at major retail stores, or even data from galaxy-scanning telescopes — in order to create algorithms that can discover patterns in behavior.

OODA decided to pivot because it saw an opportunity to apply its analytical defence knowledge to the potentially far more lucrative finance world. But, as a contractor focused mostly on government contracts and used to government Requests For Proposals (RFPs), the company has little experience marketing a product. And in the cutthroat tech world, companies can’t afford to make many marketing mistakes.

“When you work on a defense contract, from the get-go it’s very, very constrained, you know exactly what you have to do,” says Demers. When you work on a product, he says, “you don’t have a client. You define the constraints. You define the parameters, and for that you have to understand the user.”

Mike Franklin is the director of a lab focusing on big data – including cancer genomics, traffic prediction, environmental monitoring and crowdsourcing – at the University of California’s Berkeley campus. His work is funded by both government and industry.

“The pace of innovation for data management in industry is much faster than in most branches of government,” says Franklin. “In terms of challenges of going from government to industry, that’s probably one of the big ones.”

OODA does have a bit of experience in the financial world. They’ve already released a product called BullBear, software that takes commodity news from the web and offers analysis about price movement. Some 2,000 people currently use the application. It isn’t bringing the company a lot of cash, but is still viewed as a success in the eyes of its creators. MarketSense is BullBear applied to all markets, meaning it has a much larger potential uptake.

But finding interested companies is tricky for OODA, in part because winning clients is an entirely different process in the defense world. To get government contracts, a company submits an RFP in a more or less open competition. Weeks later, the winning company is chosen and they deliver the work.

In an effort to market a product with broader appeal, make up for its handicap from working in the government world, and better tailor its approach to financial professionals, OODA partnered with Qmetrica – a Toronto-based quantitative analysis company that tests investment ideas. The partnership provides them insights into what financiers want in a product.

They also brought on Jake Hirsch-Allen, a lawyer who works as a business development consultant. He’s taken it upon himself to make MarketSense and BullBear sellable.

Thanks to its days working in defense, Hirsch-Allan says OODA’s founders “have some unbelievable acronyms,” adding that the company’s jargon was “totally incomprehensible.” He nixed the jargon and focused on making the description of the product easy to understand with a snappy elevator pitch.

Hirsch-Allan is also constantly networking, trolling blogs and Twitter for new names to expand OODA’s potential client list. A handful of professional traders and trading houses – seeking a social signal to increase the accuracy of their market prediction – have already expressed interest in using MarketSense.

And Hirsch-Allen has even more reason to be optimistic. OODA and companies like it have a golden opportunity to capture new big data markets.

The U.S. government recently launched a major research initiative to mine big data for insight into a variety of research fields. And in April, Splunk – a company that makes analysis software for customers including Zynga Inc. and Bank of America Corp – became one of the first big data companies to go public. The San Francisco-based business’ earnings jumped 80 per cent on releasing its IPO, and it now has a market value of over $3 billion.

OODA are remaining cautious about the move into a new market, however.

“For the foreseeable future, government contracts will remain very important because they are bringing in the money,” said Demers.

Yet he’s confident that the company has a valuable future in the financial sector.

“MarketSense has the potential to have millions of users,” he says. “Maybe one day that will be the main revenue stream.”


As Interviewed by: Tom Henheffer

There’s a lot of market demand for analysing big data and turning it into information, but it’s a really complex business. Still, OODA are headed in the right direction by partnering with Qmetrica because they need to determine what competitors are offering, have some focus group sessions, and figure out who their target buyer actually is. I don’t think the target is stock brokers -- they’re unlikely to change established research habits. But capital markets traders and retail day traders, people who trade for profit and are looking to gain an edge through information, might pay for OODA’s service -- if there’s a high enough correlation between their information and market outcomes. I suspect their correlation is somewhere around 25 or 30 per cent, and no one will pay for that low level. But if the company can find the right sources of information, which probably go well beyond social media sites, and can fine-tune their analytics to reach a correlation of 70 or 80 per cent, they might have something more lucrative than any project they’ve taken on in the defense industry.

by Peter Conrod, FCMA

OODA are right to capitalize on mining big data, and they’ve brought the right people together to tap the opportunity. But these guys know the defence industry, and the private sphere is very different -- product cycles move much faster, and marketing is very different from responding to RFPs. The company is based in Montreal, but sellsl to the financial industry, so my first advice is to set up a remote office on the Bay Street corridor in Toronto. You can’t underestimate the importance of a face-to-face presence, and collaboration tools, shared work spaces and remote office technology make this relatively easy and inexpensive to establish. A solid virtual presence will also be important. They need a website that communicates their value proposition and expertise, one that shows they have a legitimate product and are going to be serious players in the financial industry. And, because they’re selling something new to an industry they don’t know, first impressions will be very important. In other words, they need to grab some tablets and show software demos to potential clients that demonstrate exactly how MarketSense works and what value it can create.

by Tisha Rattos

by Theo Peredis