The Simulation Game: How Angstrom Offers a Peerless Pricing Solution for Major US Sports
How can you increase the accuracy of real-time odds? Sion Colley, CEO at Angstrom Sports, introduces a gradual approach to price setting for sportsbooks with play-by-play level predictions of every single underlined event that contributes to the game. Sion also shares strategies for meeting the demands of the US market and his predictions on the future of the industry.
SPEAKERS
Russell Karp, Senior Vice President in the DataArt Media and Entertainment practice, focusing on sports betting.
Sion Colley, CEO at Angstrom Sports. Angstrom Sports is a data-driven sports forecasting business specializing in odds generation, trading, and risk management. They build player-level, play-by-play simulators that produce the most accurate lines in the industry across the major US sports and facilitate zero-latency in-play pricing.
Kevin Twitchell, advisor in the DataArt Media and Entertainment practice.
Watch the full video or read a shortened text version below.
Russell Karp: When we had our previous call, you went into some good depth about some of the interesting analytical strategies you have in place to come up with a product a bit unique on the market, I would say. Can you talk a little bit more about that?
Sion Colley: Yeah. It’s sort of fortuitous that we built our technology in the way that we did. The guys and the core of the team have been involved in the sports betting industry for decades in various parts of the supply chain. But, when they set out to tackle predicting outcomes in MLB games, and then NBA and NFL games better than anyone else in the world, the approach that they thought would be the most accurate and leads to the most efficient pricing was to make those predictions on a very granular basis. So, player-level predictions, play-by-play level predictions. And to the extent that we wanted to trade on a much more aggregated basis. For example, on what we call the main lines in sports betting, so you know who’s going to win and by how much.
The key fundamentals that underpin the way we would come to those values were, if you can predict in a very accurate granular way, that’s the best way you can predict that at an aggregated basis. And that was that. The way we went about it.
As it turns out, it has given us a huge advantage in being able to price not just the main lines very well, which is what we’ve been trading and doing as it’s not been traded for the last 15 years, but all of the derivative markets and every single underlying event that contributes towards what the overall outturn of a game is.
Sion Colley
CEO at Angstrom Sports
And, as you guys will be aware of, I think the industry is becoming more aware, being able to offer a breadth of products that covers not just the main lines but the stuff that people can build narratives around, engage in their team around, and engage in the sports around; which is playable stories, or stats level stories, or engagement that isn’t just who’s going to win the game. That equipped us very well to be able to provide our clients with the type of product their customers are asking for. So, it’s the beginnings of predicting in a very granular way that has given us perhaps a different starting point from the way the industry was pricing up markets at the point when we started. And maybe there are a few guys who are trying to do it more on the line of the way we’ve been doing it for the last 15 years or so of late.
Russell Karp: I’m sure it’s ever-evolving for you, especially with the US major sports over the last 6 to 10 years, maybe a little bit less than that. Look at the NBA; the whole strategy of the gameplay has changed, right? Or when their rule changes. Whether it’s in the NFL or any sports, that’s going to affect your prediction model as well. So, you have to always be on the cusp of what’s current and what’s coming.
Sion Colley: Yeah, we do, for sure. This is a really interesting point because it goes to the heart of the way in which markets have been priced historically and the deficiencies in that approach, how we price markets, and the advantages of taking the approach reading. Now, historically there would have been an assessment in relation to a team’s average performance in certain sets of circumstances and a view that a team, especially a team with X, Y, and Z in it, is X points better than the average NBA team, for example
And that advantage increases by 0.25 points, 0.5 points, or 1 point depending on a number of different factors. That’s a very swift way of determining what the right price is for a contest or a game on a pre-game basis; it’s very efficient. And it worked. Those models have become more and more sophisticated.
What those models don’t deal very well with is if you change the rules or if you change the way a team is playing strategy-wise, you have to rip up the rulebook on all of the averages and all of the analytics that you’ve done on a team level basis up until that point in time. And you have to try and guess what is going to happen in the way teams are going to behave in the new rule’s environment or with a new generally accepted way of playing the game, like shooting threes as much as you can. And the advantage of going at the program in a very granular way is that we don’t have to make these universal changes to these averages.
We can say players with these types of attributes and capabilities are going to become relatively more valuable and are going to rely on relatively more. What happens when you put them in the mix more often? What happens when you increase the usage of them? What happens when you ask these guys to shoot from outside the three-point range 40% more the night than they used to? What happens in those events? Because our models have the answers, you’ve just got to ask the right questions. You know the answer.
The same goes for a rule change. If you are playing seven innings rather than nine because you’ve got to a congested season because a global pandemic has hit, we tackle the rules. And we say ok, put the same players in and tell us what’s going to happen over seven innings and do that on a pitch-by-pitch basis.
And that’s what our approach facilitates, which is a big advantage whenever there’s a chance of a pre-game situation; and a huge advantage in trying to figure out the way a game is going to play out once it goes in-play. And the difficulty with the preeminent process is that it relies on averages and average profiles of scoring patterns.
If you’ve got the Chiefs in seven-point favorites, there will be a generally accepted view of how those seven points are generated over the course of four quarters. But what happens when the Chiefs go down by 16 in the first quarter? How are they going to play from there onwards? And it’s very difficult to do that if you don’t have a very granular way of predicting the outcomes in those sports.
Kevin Twitchell: I’m just fascinated by the whole model. So, just taking a little step here into the business of it. How has your approach been in dealing with customers and launching this in the US? And the other sport I’m most fascinated with and watched too much of is the NHL. I think there’s a lot of complexities in US sports. How do you deal with that? It’s two questions there.
Sion Colley: Yeah. I’ll take your last question first because it links into the way in which we go about modern sports. And I think you raised a great point. NHL is probably one of the more difficult US sports to predict. It’s an invasion game, similar to an NBA; it’s quite difficult.
I grew up consuming European sports. I’m a complete convert, but I have to say NHL is the latest sport for me to become a fan of, and I’m hooked. It is so fast-paced, so fast. And I like the fact that relatively speaking, it’s low scoring. There are a few very significant events that can occur in an NHL game that will drive the outside of the game, whereas the NBA obviously is taking along a score every 15 seconds or whatever it is. But it does present a very different challenge in that respect. Similar to NBA in terms of it being continuous and an invasion game, but it’s a little more challenging because of the greater variance and the lower number of scores versus the NBA, for example. But it’s all possible to be tomorrow.
So, you put in place the framework, the way in which the game is played. You look at the way players react to different situations and the way teams and coaches react to different situations. And then you set situations up in some sort of fictional land, which is where your model is live, and they play the game out a whole bunch of times.
Sion Colley
CEO at Angstrom Sports
And then they’re able to say, “Well, on balance, 67% of the time, this happens, and 22% of the time, this happens.” So, that’s how we take on that type of challenge. And as I say, our approach, our way of generating those insights and predictions, is a lot more accurate than saying, “Okay, well, historically, when team A is the favorite by one point or two points, when they found themselves in this situation, this is what they’ve done.”
And that’s quite an inaccurate way of going through an in-play situation. That’s the difference. In terms of how we tackled our expertise in the US, very early on, the approach of all the license operators or those who were in receipt of a license to operate a sportsbook online or retail looked to the European market and the European providers one way or the other, either the B2C operators or the B2B providers, to help kick-start their operations. Because, as we know, it was coming off a low base.
Relatively few people are employed in the industry and a relative scarcity of expertise in running sportsbook operations. And like some of our European counterparts who participate in other parts of the supply chain, platform, sports pickups, and B2C themselves, we’ve been able to come and talk to the operators who are getting going in the US and to demonstrate our expertise and the fact that we’ve been doing what we’ve been doing for a long time.
And the advantages of our approaches have been listened to because of the, I guess, depth of experience that we’ve got in doing what we do. That said, there are some very significant differences in the way that the US industry is evolving, and, not least, that’s the nature of the sports themselves and the nature of the US consumer. It’s not our job, as a provider of pricing services, to anticipate those needs. There are guys who are best placed to do that than the sports operators themselves. You guys are the ones who are close to the consumer, close to what the user experience needs to look like. We can give it a go; we have our views ourselves. But our job is to try and facilitate their own neighborhood, the vision that their clients and our partners have in terms of the direction they want to take their sportsbooks. Whether that’s their strategy promotions or pricing strategy or the way they want to take their content in terms of product offering. And any other aspects that we can innovate and help innovate around any of that.
So, yeah. How we want to market, we were very fortunate to be a part of the wave of experienced Europeans who were able to bring their experience in what was a nascent market. But we also listen to you, guys. It’s a different marketplace. We don’t have all the answers, but we have lots of ingredients that can create some fantastic solutions for our clients. It’s about listening to what our clients want and trying to enable that for them.
Are you looking to elevate your sports betting operations to the next level? DataArt is here to empower you with cutting-edge solutions that can boost efficiency, expand your market reach, and provide a technological advantage over your competitors. Our team covers every aspect of sports betting software development, from scalable and secure platforms to advanced analytics. Explore DataArt’s strategic sports betting solutions.
By Russell Karp,
Vice President of Media and Entertainment Practice at DataArt
Originally published at https://www.dataart.com.
