HotStreak’s data science and machine learning provide critical technology for real-time gaming.
SANTA MONICA, Calif., March 16, 2023 /PRNewswire/ — One of the most popular growing forms of sports gaming is micro-markets.
Micro-markets, which HotStreak began offering in 2020, are a subset of live sports gaming that focuses on what will happen next in the game. It’s our speciality at HotStreak and what makes us unique in the industry.
Micro-markets, at a basic level, are predictions of what’s going to happen next in the game. With HotStreak, for instance, in NBA you can predict what will happen during the next 3 minutes, or in NFL the next drive, or in MLB the next at-bat.
For customers, micro-markets are popular because they add a new dimension to sports gaming and watching the game. While watching the game live, they can make entries based on what’s happening in real-time, such as who’s on the court, how they’re performing, or the teams’ strategies at that moment. (This also increases engagement because new markets are opening and closing so frequently.) As a result, the considerations for making micro-markets entries are also very different from those when predicting an entire game.
Customers like combining these micro-markets into parlays. But combining these micro-markets in the same game requires advanced technology because it requires accounting for various complex levels of correlation between players. “At HotStreak we were the first to offer these micro-market same game parlays in 2020. We’re still the only company offering these micro-SGP games,” said Greg Dean, chief executive officer at HotStreak.
Same game parlays, which many companies offer, have traditionally focused on the entire game’s outcome. But HotStreak is different because it offers parlays of multiple micro markets during the same game—which other companies don’t offer.
“At HotStreak, we’re focused on providing the most cutting edge gaming products for our customers. That’s why we’re building micro-SGP,” said Gianni Settino, chief technology officer at HotStreak. “To do this, HotStreak needs a different level of technology from other daily fantasy companies. So HotStreak has spent years developing areas like data science, artificial intelligence and machine learning.”
It’s a complex challenge to offer micro-markets, due to several reasons. The biggest challenge is that correlation is very different for micro-markets compared to pre-game markets. In addition, micro-markets happen in real time and the scale of so many markets is too large for humans to manage so many different variables. Let’s look deeper at these reasons and how HotStreak has addressed these challenges.
The biggest challenge to creating lines and odds for live micro-markets is correlation. Correlation is how much one players’ stats will change depending on another player’s performance on the court or field at the same time.
Some types of correlation are intuitive based on how players interact. When Devin Booker has a higher number of three pointers, Chris Paul typically has a higher number of assists, and vice versa. So combining Booker’s threes and Paul’s assists together may cause payouts to decrease compared to combining Paul with another player.
But other types of correlation are not intuitive. Why would one player on one team scoring more affect the points or assists of a specific player on another team?
Even if you know there is correlation, figuring out precisely how much correlation there is—in odds or lines for micro-markets or micro-SGP—presents another challenge. Also, correlation can change depending on many factors, such as the time left in the game or which team is winning. For example, teams may draw up specific plays during certain points of the game, which could affect correlation. Or NBA teams may have players go one-on-one during specific times in the game. And so on. When you add a third or fourth player to your entry, the complexity increases tremendously as you attempt to account for correlation.
In-game vs Pre-game
More importantly for our topic of micro-SGP: correlation for in-game micro-markets is very different from correlation for pre-game lines. This makes sense because pre-game markets covering the whole game can balance out certain effects of correlation as one player plays with more players throughout the whole game.
But for micro-markets—which account for just 3 minutes in the NBA or one drive in the NFL—the effects of correlation can increase—or decrease—much more, because it’s a short amount of time.
Certain players have an especially big difference between correlation for pre-game versus in-play micro markets. For example, Jalen Brunson and Miles McBride have a fairly negative correlation for points and assists for pre-game covering the entire game, our data shows. This could be because McBride is a back-up who often subs in for Brunson. But for micro-markets during the last five minutes when the score is within 5 points (“clutch time”), they have a positive correlation between points and assists. One possible explanation: during whole games, when one is playing well, that causes the other to not get opportunities. But when they actually play together in key clutch moments, they help each other out.
Another example: our data shows that Stephen Curry and Draymond Green, a long time points-assists pair, have very little positive or negative correlation for points and assists for the whole game. But for micro markets in clutch time, they have a fairly strong correlation. This could indicate that for whole games they both help the whole team perform well or poorly, but in clutch time they especially help each other.
“Calculating correlation is very complex and is something that HotStreak has worked hard to measure for all players for each of its markets across different sports,” said Taylor Tanita, head of data science at HotStreak.
Another challenge with in-game micro-markets is the increased scale of markets. With markets in many different stat categories—points, rebounds, assists, three pointers, etc—for many different players, you need technology to create these lines and odds in real-time. It quickly becomes challenging to have humans price 10 different markets (up to 20 total including combinations) for 10 players in each NBA game.
Also, because micro-markets are by definition live, it’s even more necessary to have technology pricing these markets. HotStreak’s micro-markets offer odds and lines on players during games—for example, whether an NBA player will score more or less than 2.5 points in the next 3 minutes, or whether a batter will get a hit in his next at-bat. These odds and lines are constantly changing throughout the game based on changing conditions.
In-game real-time lines and odds are a new addition to the sports gaming space with only a few companies having this capability. And micro-markets that HotStreak offers aren’t being offered anywhere else.
To illustrate this, let’s look at how HotStreak prices constantly changing markets each second during games. There are 2,610 seconds in an NBA game which HotStreak offers. That means there are 2,610 instances in each game of lines and odds on each player. If there are ten players offered in that game and 15 markets for each player, you have 391,500 unique and different lines and odds for each game. In a given night there may be up to 15 games being played. That’s almost 6 million unique markets every day.
These markets are being created dynamically on the fly, based on changing conditions in the game. None of this is predetermined, provided by a vendor, or based on a human trader’s gut instinct.
Closely related to scale is the issue of speed. Micro-markets are constantly changing in the game based on a variety of factors. This real-time environment is where machine learning and technology really shines.
With 150 or more markets each second per game, the technology has to respond to conditions in real-time, then process those variables and then present customers with odds and lines in microseconds. As a result, the technology runs entirely online and is designed to be lightweight and lightning fast. Again, at this speed, it’s automated so humans can’t change what it produces.
Because these lines and odds are changing so fast, risk management is a key issue. At HotStreak, we’ve developed various forms of technology that provide risk mitigation that adjust to changing conditions during games.
At HotStreak we believe that micro-SGP is the future of sports gaming. As the only direct-to-consumer provider of this technology, we have seen first-hand how popular this is with our users and it has driven a meaningful portion of handle. With a deep investment in data science and machine learning, we are excited to build the best products that customers want.
HotStreak provides industry-leading daily fantasy sports games that are powered by artificial intelligence, machine learning and data science. The first to offer live in-game micro markets direct to players everywhere, the company is backed by Polychain Capital and angel investors such as former FanDuel co-founder Nigel Eccles. For information visit hotstreak.gg or blog.hotstreak.gg. For more information, contact: firstname.lastname@example.org.
View original content:https://www.prnewswire.com/news-releases/inside-hotstreaks-industry-leading-micro-market-technology-301773472.html