Liquid ETF Competition Winners Spotlight
In November 2019, QuantConnect launched the Liquid ETF Competition. The challenge for participants was to look for alpha within a basket of 72 ETFs selected for their independence from the market, their variety of underlying assets, and other idiosyncratic variables. The competition was sponsored by an institutional client looking for alpha from this specific basket of ETFs and offered a $10,000 prize pool to direct and incentivize development.
The submissions to Alpha Streams went live on January 20 and the algorithms traded through March 20. The winner was Randy Cooper and second place went to Grant Forman. We took the opportunity to spotlight these two QuantConnect users to learn more about their respective backgrounds, as well as their approaches to research, development, and trading.
(Editor’s note: The answers below have been edited slightly for clarity and length.)
QuantConnect: What made you decide to participate in this competition? Do you plan to participate in future competitions?
Randy: I was getting into quantitative trading on my own and learning Python, but it took so long to code an idea from scratch, and on top of that getting good data is not always easy. When you have only a limited number of good data sources for asset classes, it’s difficult. I realized I should be spending more of my time on researching and less on the infrastructure.
I looked into what tools were available and thought QuantConnect did a good job of providing me with resources to both help me save time and make the development and deployment process easier.
When I saw QuantConnect’s competition, I thought it would be a good idea to enter to get me into coding within the QuantConnect platform and learn how to do it with a specific goal. The ETF competition in particular was interesting because I could learn more about ETFs.
Grant: I’m probably different from a lot of other people who have participated in the competition, as I’m pretty new to building algos. I came to QuantConnect about a year ago, and it’s been a very steep learning curve. It was only near the beginning of the first challenge in October of last year that I began submitting to Alpha Streams. I find building algorithms is fun, and I’m not doing it to get revenue.
This competition had a number of hurdles, one being the requirement for the algo to have an 80% Probabilistic Sharpe ratio (PSR) or higher. Passing that threshold and just getting into the competition was the thing that drove me. I figure that after your algorithm gets accepted into the competition, then anything can happen. I will definitely participate in other competitions in the future.
QuantConnect: Both of you mentioned being new to self-directed quantitative trading. Can you tell us about your background?
Randy: I was always good at math and physics, and my early life was spent in academia. I got into finance after my postdoc at UC Santa Barbara. Life was beautiful there, but I knew something wasn’t right with my current career path. I had met an early employee of Adobe who had retired early, and that inspired me to become financially independent.
One of my postdoc colleagues in Santa Barbara knew someone at Goldman Sachs, so, 30-odd interviews later, I ended up being at Goldman from 2010 to 2017, where I moved around a lot. Along the road I got exposure to different asset classes — mortgages, equities, bonds, currencies. I was all over the place, but it gave me a very wide exposure to asset classes and all of the different types of market participants. I was a quant strategist and never really did algo trading. It was more risk modeling and data cleansing. After 2017, I retired, became a private investor, and continued this whole convoluted track to where I am right now.
Grant: I really don’t have any experience in coding. My background is as a scientist, so I guess I have taken a more scientific approach to building algos.
About 10 years ago, I moved out of the laboratory and into the corporate world. But my passion for research and development is hard-wired, so QuantConnect has been my outlet for that. I joke that this is my mid-life crisis — but seriously, it’s become a very challenging and fun hobby. It’s either this or a Lamborghini.
QuantConnect: You hear a lot that there is both an art and science to finding alpha. Can you share your general process for creating an algorithm strategy? Do you start with a lot of ideas or one thesis?
Randy: I keep a running list of ideas and try to eliminate them as efficiently as possible, to fail quickly. I always ask, “why should I expect to make money?” If you’re buying and holding, you can easily explain your positive expectations based on risk/reward. For things that I can’t explain, I need to see a huge amount of data to support it.
Grant: When I’m building algorithms, I try to be as flexible as possible in my approach. If an idea isn’t working, I try adding on tools that are provided by QuantConnect. After a little more experience, one thing I’m more aware of is integrating multiple layers of risk management into the core of the algo, which increases the chance of it being sustainable over a variety of market conditions.
QuantConnect: What tools do you find most helpful when learning about algo development? Is there a tool you would want added to QuantConnect to help you learn more?
Randy: I’ve done a lot of reading, especially early on. I’ve looked at dozens of books, podcasts, and QuantConnect’s “Research to Production” posts, etc. I’ll typically get the seed of an idea from all of these sources.
I’ve learned there is a lot of material out there and much of it is bad. I’ve been able to toss out some trading ideas and sources of trading ideas sooner because I can tell which are good and which have zero or negative value. Being able to distinguish that is important because a lot of them are not based on analysis or statistically reported conclusions.
Grant: QuantConnect is making it easier for me to learn how to code. I’ve done a few Bootcamps and Github has been a very good resource as well. I’d recommend to others to not be afraid to ask the Quant Community and QuantConnect team questions, because they’re a great bunch of people and are happy to help. I’m grateful for all of the tools available, because otherwise all of this wouldn’t have been possible. It’s a great way to lower the barrier to entry for people like myself.
If you have an urge to solve puzzles while learning and honing new skills in a community of like-minded people, check out our community of developers, engineers, and tinkerers.
We are currently hosting a weekly Alpha Streams competition, Quant League, with $1,000 prize pools. Your challenge is to create algorithms that perform well over a 5-year backtest period, perform well in out-of-sample trading, and reconcile well with their backtests. You can trade any asset class or strategy. As there are funds with many different mandates using Alpha Streams, we wanted a competition specially designed to let you unleash your full creativity without worrying about specific themes or universes.
We hope to see your alpha on the leaderboard!