The quants take on fintech

April 15, 2017 10:00 AM

The broader investment universe has begun to embrace quantitative trading strategies across all asset classes, and entrepreneurs have recognized that the talent required to created systematic trading algorithms is different from what is required to create a successful investment business. With that in mind, two innovative companies are tapping a wide universe of quantitative traders — including academics, professionals and tinkerers — to create investments strategies. 

Quantiacs and Quantopian are both bringing quantitative trading strategies to market by accessing a wide universe of quants and filtering thousands of strategies before offering and/or using the best. 

“We connect trading programs that have been written by freelance quants with institutional investors on a profit sharing basis,” says Quantiacs founder and CEO Martin Froehler. “We don’t charge our investors management fees, we only change incentive fees; our fee model is zero [management fee and] a 20% [incentive fee]. We split our income 50/50.” 

Quantiacs, based in Mountain View, Calif., has more than 2,000 programs on its platform from a group pf 6,500 quants all over the world. It offers free market data and a testing framework to research and build strategies and submit them to the platform to test on live data. It has futures market data dating back to 1990, or the inception of a market, on 50 of the most liquid futures contracts in world. 

“They use our framework for the whole process. Most of our quants work with machine learning methods; we support the most popular machine learning languages,” says Froehler.

Quantopian focuses on equity markets — though it is developing a futures module — and is offering a fund using the best strategies in its stable. Quantopian has pricing and fundamental data on more than 8,000 symbols, including equities and exchange-traded funds that go back to 2002. 

“We have over 100K quants in 180 countries that can come to this platform, meet with other quants and help develop their trading programs,” says Jonathan Larkin, Quantopian’s chief investment officer. 

The main difference, other than Quantiacs’ focus on futures and Quantopian’s focus on equities, is that Quantiacs (see Quantiacs model,” above) will offer access to their traders’ programs as a standalone investment whereas Quantopian is offering a fund based on a portfolio of individual strategies (see “Quantopian model,” below). 

The opportunity

We have pointed out for years that successful money management depends on multiple factors, and that the building of profitable market signals — whether based on a systematic or discretionary approach — is only one element of success. The logic behind both Quantiacs and Quantopian is to allow quantitative programmers to focus on building the best quantitative trading models they can and build an investment business structure around those strong models. 

“We provide a fully hosted research environment where quants can come and have a sandbox for exploratory data analysis and to look for signals,” Larkin says. “Key to that is we have over 50 data sets that are fully integrated into our platform. As any quant knows, there is a lot of data crunching, storing, etc., and we do all of that for our users.”

Both firms stress that the numerous quantitative researchers who use Quantiacs and Quantopian’s platforms to help develop their systems retain proprietary ownership over their work. 

“We don’t look at the code, we are only licensed to use it,” Froehler says. 

“As a quant you own that algo outright,” says Quantopian’s Larkin. “Quantopian doesn’t take it from you and you get paid a royalty. You can learn and get paid to develop algos anywhere in the world.“

Froehler adds, “Our software is free and open-source so they can develop on their own computers and once they have finished their research process they upload the trading algos to our platform and we start building up a forward test. We evaluate the algos every day with new market data; this way they are building a live simulated track record of the strategy.”

At that point, according to Froehler, the strategy can’t be modified anymore. Once an algorithm is uploaded, it is complete, though the same quant can create a second program. 

“The vision of Quantiacs is to democratize quantitative finance,” Froehler says. “We want to make the industry accessible to people who have no access to it at all. Think of all the talent out there. A brilliant mathematician in the suburb of Mumbai, this guy doesn’t have a chance to be hired by a top quant fund. There are many more that don’t want to work at a quant fund for whatever reason and we empower them with the same infrastructure they would find at a quant fund: beta software, educational material and ultimately with investment capital.”

Both firms are making a big data type bet that if they provide a broad open canvass to the universe of quants — be it college professors tinkering, or college undergrads discovering programming for the first time — that universe will produce strategies that are profitable and accessible at a cheaper price than hiring the current best in class quants. 

Froehler’s example of the mathematician in Mumbai may be prescient. Quantopian, five years into their platform, has quants participating from 180 countries. 

“The platform is free. We provide the research environment, the development environment, the back tester, the community forums, everything is free to them,” Larkin says. “Data we make available is mostly free. Some providers provide history and charge for other data.”

Educating tomorrow’s quants

Both platforms are extremely open and welcoming, which attracts a broad range of participants from neophytes to experts, and can require some hand holding. 

“We provide them with full platform capabilities,” Larkin says. “We have a formalized education section on the site; we have approximately 50 lectures across a range of topics: quantitative [development], finance, programming [and] trading. We encourage traders to [utilize these tools].”

Quantopian also has an active forum where traders can bounce ideas and trading concepts off each other. “In 2016 we had more than 10,000 posts in our community forum. Users can post and share things if they choose, they can ask questions from the community, they can look for collaborators,” Larkin says. “We post sample strategies that the users can then clone and use as a base to do further research.”

Quantopian also has an internal academia team that offers lectures on a variety of quantitative subject matters. “We partner with external academics [as well as offering classes] in-house,” Larkin says. “We also hold meetings, hackathons and workshops around the world where people spend time learning our API platform and quant trading techniques.”

Froehler adds, “Any quant can sign up, the only requirement is some basic programming knowledge but we are also happy to help with that.”

Quantiacs walks the user through the backtesting and optimizing process. “As soon as they are satisfied with their research results they upload their algo to our platform and we start building up their live track record, forward tests on live markets,” Froehler says. “Ideally only a portion of the data is used to [optimize]. And then they would test it on out of sample data. And then we would test further on our live data.”

On the other side is the institutional investor. Quantiacs is a 4.7 exempt CTA with a focus on institutional clients. Those clients can choose from the 35 program posted on Quantiacs’ platforms. Customers see the complete track record of each strategy including returns, volatility, Sharpe ratio and a host of other metrics. 

Once a trading program has been developed, optimized, proven itself on live data, Quantiacs may choose to seed it with proprietary money. Typically a strategy will trade prop money for six to nine months before being offered to Quantiacs’ customers. But even before a strategy gets an allocation, the developer will earn a 10% performance fee on the proprietary results. Once an allocation is made, they earn the same amount, splitting the 20% incentive fee 50/50 with Quantiacs. 

Quantopian’s compensation structure is similar, but it does not trade its own prop money and it does not allow outside investors to see the various algorithms. Quantopian acts as portfolio developer, selecting roughly 20 to 25 algorithms, to trade as part of one market-neutral fund. 

“We have an evaluation process where we screen against all of these strategies and we monitor out-of-sample performance to validate the strategy, and if the strategy meets our risk and performance criteria we reach out to the author and enter into a pay-for-performance license,” Larkin says. 

Quantopian’s approach assumes, even strong algorithms may only have a limited shelf life. They will build a steady stream of algos to use in their funds. “The genius of our business is that strategies don’t last forever, so in order to be successful over time as a fund you have to have the ability to have constant innovation, constant research, constant production of new ideas,” Larkin says. “We manage that [end to end] as part of our investment process. It is fully expected that a strategy put in place today won’t last forever, it will last a few years and if the market changes or the edge that they found gets competed away, we have built into our business model for that and part of the value add we provide our clients.“ 

Quantopian is well on its way thanks to the $250 million allocation from Point72 Ventures, but Quantiacs  already has institutional monies (approximaltey $2.2 million) trading. Quantiacs is lean with five employees whereas Quantopian has 50, basically matching the proportional size of the asset classes they are serving. 

“Our vision is to build the largest quantitative research platform in the world that has the most programs and has the best Sharp Ratio [and] to create a network of distributed contributors to our platform that crowdsources trading models,” Froehler says. “We are 100% systematic, we have a clear focus on futures, we have a clear focus on institutional investors. On the client side and we are very open to machine learning and [artificial intelligence] methods. This differentiates us.“

Larkin  adds, “We believe that Quantopian taking on the operational, platform and implementation responsibility while leaving it to the authors to do the alpha research is the right way to go. In our case, we are offering this platform as an opportunity to not just professionals but to anyone with an Internet connection around the globe.”

Tomorrow’s traders

Both see a nearly unlimited capacity as they bring on more qualified quants. Quantiacs plans to eventually create structured products based on offering combinations of strategies on their platform and Quantopian plans to add futures. 

What Quantiacs and Quantopian are doing appears to be a nexus of several trends in modern finance. It involves both the belief in a quantitative approach to trading along with the benefits of crowdsourcing and even big data. They are looking outside the normal places to find trading and programming talent and finding value. 

It involves providing novices the tools to create, compete among themselves, and learn from each other in the hope that the end result will produce better returns. Given the number of participants they are attracting, they are well on their way.

About the Author

Editor-in-Chief of Modern Trader, Daniel Collins is a 25-year veteran of the futures industry having worked on the trading floors of both the Chicago Board of Trade and Chicago Mercantile Exchange.