Megan Lanham was working on a master of business administration degree at MIT when the idea congealed: all this data-wrangling and modeling could be well-applied to sports betting as effectively as air traffic flow or the likelihood of Constellation Pharmaceuticals’ stock cratering.
Boston and MIT, in particular, are fertile ground for this type of thinking. MIT’s Sloan School of Management annually hosts a stathead conference that a few years ago was heavy with sports betting flavor, and Massachusetts was in the process of legalizing sports betting when Lanham was matriculating. The broad thought wasn’t unique, but Lanham (right) sought to hone what would eventually become the company she co-founded in 2022, Rithmm, into a system that would customize the sports prediction model sold to subscribers based on what they considered important as mid-to-sharp-level bettors.
Artificial intelligence would be crucial. So Lanham needed a quant. Luckily, many of those are bouncing around Cambridge, and Lanham was introduced to a like-minded partner in mechanical engineer Brian Beachkofski at a birthday party in 2019. He’d once even taught the “Data Models and Decisions” class, where Lanham’s thoughts drifted to sports betting. They gave it a go.
After a year of testing and $4 million in investments secured, Rithmm – as in algo-rithm and an extra ‘m’ for ‘models’ – launched in September.
Gaming Today spoke with Lanham about the process.
GT: Why is Rithmm here?
MEGAN LANHAM: We built Rithmm to advance analytical models for sports bettors. Sharps use predictive models all day long to make their bets. They have a team of quants, a team of data scientists, and engineers. We wanted to be that team on the backend while, on the front end, the user has a much easier experience.
You don’t have to be a quant, you don’t have to be a data scientist to understand this. We will build the algorithms. What we need from you is what your instincts are and what you think is important in a game and in a bet.
And then we take your instincts, we combine it with our algorithm, and within 10 seconds to one minute, provide you with a list of picks for tonight’s game. What we’ve set out to do is give access to people, what the pros use, giving access to everyday bettors.
GT: How do you do this?
LANHAM: We find inefficiencies in the market. Our algorithm does that. What's really cool about this is we've seen, outside of the mainstream games, everybody bets the favorites. What our models do is find these inefficiencies that are more obscure games, and it's allowing people to bet on very different types of teams. And we've seen enthusiasm around that.
For example, let's say for football, you think passing touchdowns per drive is important. You think red zone conversions are important. You think takeaways per possession, or per play, is important. We take those, we combine them in our algorithm and our modeling, and then we backtest that with three years of data. And then it shows you how well your model back-tested. And literally, it's green is great, yellow, there's some potential there. Red, we would tell you to stay away from.
So it allows you to take your instincts and use really incredibly analytical models, and backtest them and see if your instincts are actually validated and would produce a good bet.
Former NFL, USFL QB Espouses Rithmm System
GT: Who's your customer?
LANHAM: Typically it's 25-to-45-year-old males that have a weekly budget, that are betting weekly, that are betting typically on the NFL and some college football. Those are the people that have been our entire market right now. Obviously, we'd like to expand that. But that's who we have focused on - that age group, that demographic, and that type of bettor.
GT: Was Massachusetts the perfect breeding ground for this?
LANHAM: I was in a "Data Models and Decisions" predictive analytics class and doing modeling for my company at the time and many other companies. And I was fascinated with the ability of how you could predict. And I was really excited to hear, to see what the betting community was doing, because I was betting. I had coached before and I was betting. And I was like, 'Oh, wow, let's see what the betting community, how they're using these predictive analytics.'
And that's where I saw a really big white space between, 'Oh, pro gamblers or sharps are using this,' and then there's this really small community that's having to scrape their own data using Excel sheets and just incredibly complicated work, time-consuming, and possibly not always accurate because the amount of data you need to consume.
And there's the rest of us in the middle who are asking our friends, 'Who do you have tonight?' or researching for hours, but really not understanding what is relevant because they're using math. And that's where we saw the white space for Rithmm. This should be available for people.
That's why it was born in Massachusetts. Because I was in school there, and my co-founder, who was a quant, from MIT. It's just been a beautiful community.
Go Deeper on AI and Sports Betting with the Gaming Today Playbook Podcast:
- How sharp can AI become in predicting sports outcomes?
- How detailed can a personalized model be?
- Do prediction services or sportsbooks with the future AI war?
GT: How did you and Brian Beachkofski firm up Rithmm as a concept at a party?
LANHAM: Someone said, 'Hey, this guy, I think you should go talk to him. He got his MBA at MIT, too. And so I went over and said hello. And then I was telling him how hard this class I was in was. And he was like, 'Oh, wow. I actually taught that class.' And I was like, 'This is too much.'
And I said I've got this idea, and I shared it with him and he said, "I can do that."
I said, "You can?"
He was working for the military. He was in charge of their software division for the Air Force doing predictive analytics for very important national security. I said, "We have to have a really interesting front end for people, simplified and personalized. I can take care of that. And you can take care of the code."
GT: How accurate is Rithmm?
LANHAM: How accurate we can get? We have people that are at sharp value or the models that they've created are at sharp value. So anything above a 55% win rate, you've got a 10% [return on investment]. Those are the types of analytics we want. And we are doing incredibly well for the beginning of the NFL.
When you think of the NFL, it's the most efficient market. And we have just been beyond excited about how our models collectively are doing. So, we look at how many people are tracking their bets and how the models are performing. We can also then look, if you didn't track your bets, how your model is performing. Those things we can then help with people.