Consistent across time, sports organizations strive for success and to be the best. However, winning on the field is not sufficient to succeed as an organization as sports entertainment evolves. From fans changing the way they attend and watch games to sponsors changing the way they market to those fans, sports organizations need to look holistically at their operations and the connection between fan engagement, sponsorship, game/player analysis, and business operations. The team that evolves to an analytics-driven, integrated organization and connects these facets will win on and off the field.
– Good afternoon everyone. My name is David Cunningham. I’m the CEO of InsCipher. In your agenda, it probably is listed as Data Factory. We’re in the process to rebranding our company over to and InsCipher. With me today is Young Bang, the Chief Growth Officer of Atlas research.
And we’re going to present to you using machine learning to evolve sports entertainment.
From an agenda perspective, I’m going to go through our introductions here in a quick second here, but Young is going to go to our overall approach around sports analytics, and then I’m going to go through how we’ve implemented this for some of our clients and case studies.
Again, my name is David Cunningham, CEO of InsCipher. I have over 20 years of Information Technology experience, supporting both the federal government and commercial sectors and really been focusing on the pivots of technology over the past 20 years from service architecture to enterprise integration and then focusing the last 10 years on Cloud computing, and big data primarily, and bringing decisions and insights into our organizational clients and accelerating the ways that organizations use data to make better decisions on their organization. And with that, I’ll introduce Young Bang. – Hi, everyone. My name is Young Bang. I’m the Chief Growth Officer at Atlas Research. Atlas Research is a healthcare transformation and digital consulting company supporting federal as well as commercial clients.
My background has all been high tech, ai, digital, Cloud more recently, but going all the way back to the consulting days and the .com days, I’ve always been in the high tech sector, and before that I was in the army. So when you think about analytics, sports analytics, everyone really thinks about the bottom left here, the game and player analysis and while we will cover that as well, we’d like to think about things in a multi dimensional approach here.
And when you tie in fan engagement, sponsorship and business operations, especially when you think about the volume of data that’s out there these days, when you think about crowdsource information, open source information, phones have really taken off producing data, video that’s out there. When you think about even cable and programming, there’s more sports content than there are music, news or weather combined. And so when you aggregate all that information, you can target obviously the game and player analysis or the Moneyball or Moneyball 2.0 type analysis. But when you really think about that, you can really have an enhanced customer experience targeting the people or your consumers, as well as focusing that and targeting potential sponsors for ad revenue generation or you really streamlining your own operations as well.
And when you think about fan engagement, one of the things here that we like to think about is, how do you actually marry up the data that you already have about your fan? And social media example is another area, but we talked about the digital portions as well. But when you combine those type of information as well as other aggregated data sources about the person in general, you can actually get a better view of the person, we like to call that a 360 view of the fan in this case, or your customer a lot of times, and then you get obviously the information about who they are or where they’re coming from. But more importantly, you get an understanding of their interest, where they’ve been, where they’re coming from physically as well as literally, as well as where their interest might be from a promotional standpoint or advertising standpoint, or just are they’re interacting with phone applications while they’re actually watching the game or are they actually using other information to really augment their experience? So that really, when you start thinking about that, that’ll help you think about how else can you actually evolve that piece into looking at sponsorship. So from a sponsorship standpoint, we can build on everything that we learned about the fan and aggregate all those different data sets. And upselling and targeting the customer experience, not only to everyone in the audience, but also the individual in the audience and the fan that’s actually watching that. When you look at certain things here, the intuitive thing is obviously targeting your Nikes, your Under Armour, those type of organizations that are already sports advertisers, as well as just logical organizations you’re gonna be thinking about, but for example, you should be looking at other organizations. So for example, there’s a certain airline that spent 94% of their budget and ad spend devoted entirely to sports. There is another telecom carrier that spent an increase of 700% just specifically in sports. So again, when you marry the 360 experience about the fan together to upsell those type of things, you can do analysis and if you’re not the same carrier, hey, you might want to upsell that fan with your promotions and those type of things. And so I think again, thinking through the intuitive portions of your clients that you’re already targeting and how to get more revenue from them, but also thinking about the fan experience and upselling other sponsors for that experience as well. Those are intuitive portions, but one of the other things that we thought about and actually have supported different clients was in the New York City area, we won’t talk about the specific organization, but it is a major league baseball team. They wanted us to look at different information associated with obviously fans. But one of the things that we did was looking at the mode of transportation in particular, we looked at taxis. So if you’ve lived in New York, everyone who’s a native knows that taking the subway is the best way of transportation, but when you think about that, say you live on the Lower East Side or whatever, you could take the subway up to a certain point but a lot of people will transfer over and get into a cab to get to Shea Stadium or to get to Yankee Stadium, those type of things. When you look at that type of information, there’s a really easy way of looking at things. Here’s an opportunity to upsell. So logically, if you’re an example of a Rideshare company or Uber or Lyft, or whatever, you can target the certain cab business. Or say you’re a cab yourself, and you want to actually upsell to the individual that’s riding your cab for different things. A lot of times, there’s concerts that go on after the stadiums, or there’s restaurants or bars in the adjacent areas that are actually having live concerts and bands and those types of things. And again, there’s cross promotion opportunities there. There’s ways to upsell and target those type of things as well as hey say, maybe I’m from Long Island, but I haven’t eaten at the Momofuku. Maybe there’s an opportunity to get Uber Eats who just dropped off a passenger in the area to pick you up and also have your food. So again, lots of possibilities there when you combine data sets, opportunities for sponsorships, and individual as well.
So when you think about the game portion and the player analysis, yes, everyone does the whole game analysis and player analysis.
So we could talk and now assume about, hey, based on this quarterback, and in this down, and in this coverage scheme, they’re more susceptible to call a passing player or a running player and those types of things.
So that’s intuitive. I think everyone does that these days. It’s kind of table stakes. And even when you look at baseball, and they kind of clock the pitches and show the placement of, those are all table stakes. One of the things that I think that is a really good opportunity that not a lot of people haven’t focused but we have, is looking at certain type of data that it’s not used as commonly. So let’s focus on video. When you look at the content that’s generated from TV and media we’re talking about, every game is televised. And so think about those commercials that you see with those markers of those athletes. And they’re in the gray suit or the gray backgrounds, and they’re shooting them for the next EA games. And that’s really to capture their biomechanics. So one of the things that we’ve taken is things and techniques that we’ve learned from that scenario, but also placed them into video. And so when you combine computer vision and machine learning into video, there’s a lot of things you can actually do that’s not quite as intuitive. So instead of using the little white balls for the biomechanics, we could actually use computer visioning, to watch video, to look at certain things about, say a pitcher. And we’ve actually helped a certain organization looking at scouts for pitching. And so while we’re watching the video, we’re using computer vision and machine learning to look at certain things like external rotation and the mechanics, the arm cocking, the release and follow through to really help them think through their scouting and their draft of who will have a higher probability of success based on the mechanics, or maybe the limited amount of time that they want to spend on mechanics or their pitching cups.
So take that one step further.
Another thing that we’ve done beyond that is actually taking the pitching coach and the pitcher themselves and looked at other things to see if their mechanics are actually following through based on what the coach is saying, or the pitching coach is saying, that’s intuitive. The other things that we’re looking at are certain things, like we talked about external rotation of the arm. We’re taking the video and looking at external rotation as well as internal rotation, plan that over the elbow flexion in degrees, and then we’re tracking that over cumulative innings in the season, and then looking at average trunk separation, so it sounds like a lot, but if you think about those types of things, we can actually predict when a pitcher should be taken out. Not just obviously after the game because you can see the performance dropping, but looking at when potentially, they might need to be pulled out of the season to save them from, say Tommy John surgery. So I’m actually a local DC person. So if you think about the national several years back, they controversially pulled out Strasburg from pitching because they had a fear that he might need Tommy John surgery again.
So they pulled him out.
And that may or may not have impacted the playoff or the World Series. Now, obviously, in the subsequent years, they won the World Series. But if you can actually use the predictive power of video and other data to predict when a pitcher should be pulled out of a season, or maybe they don’t have to pulled down, they could be prolong their season. It would actually impact potentially someone’s season or a team season as well as potentially future seasons as well. So again, I think there’s a lot of areas here in the data side here that we can do and look at that could be focused on not your typical game or player analysis, but other ways of looking at it instead of play prediction, but looking at more non-intuitive ways, and other data sets to look at that. When you look at business operations here, this is probably the other dimension of things that a lot of organizations aren’t thinking about. And David’s gonna talk a little bit more about this in depth. But there’s opportunities here, when you think about how do you look at data?
For a certain similar organization up in… plays baseball up in New York, they asked us to look at a couple things. And they asked us to look at season tickets, pricing, and attendance throughout the season. And so one of the things that we did was looking at how do we optimize price of tickets, and as well as how do we increase attendance during the season, even though they might have sold out their tickets. And a lot of the things in analysis that we’ve done is looking at secondary pricing and secondary markets of tickets. So we’ve looked at other sources of information to look at variable pricing of the tickets. So again, on certain teams that might come in, or certain pitchers that might come in, they might have variances of attendance as well as pricing. And they asked us to look at how do you actually optimize that? But then the other portion of that was how do we look at demand out in the secondary markets? So New York is famous for scalpers or stub pubs or those type of things. So we looked at other data sets to say, “Hey, in these types of things optimally to either increase attendance, you should set the price here or optimally to get the most dollars that you can, because you’re actually leaving this much room on the table, because in secondary markets are selling for this much, maybe you increase the price for these types of series. So again, a couple of examples here that I’m talking about, but David’s going to really talk about how technically we did it, as well as some of the more the details behind that. – Thanks, Young.
So as you talked about the overall sports analytics solution, and when we start to think about what we’re talking about in terms of data, we’re talking about data that we already have, we’re not talking about collecting new data sets, but being able to really enrich and use the datasets that we already have. But there’s also a kind of traditional pre post, pre processing, and post processing. And when we think about that in terms of data, we think about in terms of once we get data into the system, how do we pre process it and get it ready to be used and post processing. Once we’re there, we start to see how we can start to look at quality of data, start to enrich the data as well. But the way that we think about it too, is that the pre processing of data is everything that happens before game. The post processing is everything that happens after a game. And I think what we really want to focus on as well is what happens during the game and in order to provide both that, I’ll say an off band analytics before a game and after a game and in band analytics around real time analytics, while the game is happening. We’ve built our InsCipher platform to be able to support any of those types of scenarios using open source technologies and standards to really accelerate the point from data collection all the way to decisions in an organization. And the way that we’ve done that, is by providing a scalable architecture that rides on top of Spark, and Delta lake and ML flow on any environment, whether it be on Premise, Cloud, Hybrid, Edge devices, etc. All based on Kubernetes and all based upon a non Hadoop ecosystem, because what we found through our experience is that leveraging the Hadoop ecosystem has its goals and its sweet spots when it comes to supporting organizations. But moving to a much more modern architecture, Cloud native or Cloud compatible using object storage at the base, provided us the flexibility and the scalability to really grow with our clients and be able to support the speed at which our clients were asking us for decision making. When we think about how we do that, we really have one integrated approach to secure operations.
And we see that from a three pronged approach. One is our dev sec ops enabled governance process. And that runs across not only the way that we develop InsCipher and provide it to our clients, but what’s embedded inside of our platform so that they can securely deploy, test, validate, and deploy and monitor machine learning algorithms that go out onto the network. That is embedded in every step of our process and embedded in the way that we think about our delivery to clients. From the left hand side, we think about our data ops approach and how do we start to bring in again, not necessarily new datasets, but how do we start to bring in the datasets that an organization already has that are freely available from a social media perspective, from a news perspective, from a streaming media perspective, etc. How do we bring that in, profile it and understand it and prepare it to be used by those machine learning programs and algorithms. Once they’re online for us and be able to extract the features and ready to provision it out to the machine learning algorithms, this is where we really focus our ML ops process, which is powered by ML flow and thinking about ways we go from very quick and dirty, Jupiter notebooks as an example, all the way up to a robust machine learning algorithm that is introspecting video, looking at images, looking at different points of view around what players are doing to hit on the kind of the Strasburg example from Young, all the way to looking at our salesforce system to our ticket master and how we start to bring all those pieces together to really drive insights into an organization, so that they can make decisions both prior to a game happening, after a game happening, and during a game’s happening as well.
So some of the case studies that we have, and I apologize, we can’t get into a lot of the nitty gritty details because of non disclosure agreements. But one of the clients that we had come to us, again local to the DC area, really wants to drive sponsorship and drive sponsorship engagement into the organization. They wanted to understand, where the sponsors were coming from, they wanted to understand who they should start to target from a sponsorship perspective. And really their lens was, again, all focused on sponsorship and what we talked to them about was that is one lens into your organization. And you need to be thinking about how the fans interact with those sponsors, how the game interacts with the fans, interacts with those sponsors, being able to start to unleash the data in your ticketing system, to your CRM system, to your financial system, how all those bits and pieces play together and start to drive your organization forward from a sports entertainment perspective. So when we started to look at the data and started to focus them on a much broader picture from sponsorship, one of the easiest things that we started to look at was sentiment analysis. And this is not just pre game or post game sentiment analysis, but also leveraging the infrastructure that’s in place in their arena or stadium or field around their local hotspots, in game applications, being able to take advantage of credit card swipes on concessions, all these sorts of things to really give them an understanding, hey, this is what people are thinking about the game, this is what people are thinking about the sponsors that are coming in and this is also thinking about how you in real time can start addressing concerns that a fan may have as an example, on the PA system and give them that understanding that hey, we’re listening to you, we understand that one area of the stadium is having concession issues. We’re on it, we’re addressing it, we’re there proactively taking it versus reactively dealing with it after the game. The other aspect that we looked at as well was around a ticketing and making sure that when fans are coming into a stadium, we know where they’re coming from, but also where they’re not coming from. And I think that’s the most important thing that we looked at with this individual client is, not only looking at ways to understand where their fans are, but more importantly, where their message is not getting to. Because if you think about most organizations, they have a very committed fan base. But everyone, especially in this day and age where streaming sports online and on your TV gives you a much more, I’ll say, all inclusive experience, driving state fans into the stadium is critical for the success of these sports organizations. And so what we looked at is not only where fans are coming from, but more importantly where the messaging isn’t getting, how do we start to drive leads through sponsorships, through cross promotions, through just pure fan engagement in those areas. And then to support that as well, we started looking at what is the connectivity between the fans and the companies? As you’re fully aware of, a lot of these sports organizations that have say a baseball stadium, they’ll have a baseball game and then the next night they’ll have a concert. Or if you’re in an ice hockey stadium, you have an ice hockey game followed by an NBA game, followed by a college basketball game, followed by a concert and monster truck and all these sorts of different opportunities for cross promotion of companies and events. How do we start to understand what that network map looks like between the fans, the company’s, the potential cross promotion for other sports organizations, as well as the potential targets for sponsorships and campaigns to ultimately increase the return on investment by the sports organization and increase the return on investment for the sponsor. So again, we looked at this, we brought together all of the existing datasets that are in place, leveraging spark to really run all of the analytics backbone, using object storage from a technology perspective, no dupe ecosystem whatsoever is in place and provide again that batch processing or post or pre game analytics, as well as that in banned real time analytics during the game. And with that, I’d like to thank you for coming and hearing our discussion around and how to enhance sports analytics using machine learning.
David Cunningham has more than 19 years of Information Technology experience supporting advanced technology programs and major transformational programs within the Department of Defense, Intelligence Community, Civilian Agencies, and Commercial Clients. He is consistently at the forefront of technology evolutions to drive customer success and meet critical missions and business objectives. From architecting complex geographically dispersed systems to evolving and strategizing the transition into data-driven organizations, David has provided exceptional support to meet his clients demands for industry-leading cloud and data solutions.
As the Executive Vice President of Growth at Atlas Research, Mr. Bang uses his broad knowledge of the Federal market (Health, Defense, Intelligence, and Civil) to guide the firm’s growth in existing, adjacent, and new markets. His extensive experience in business development, capture, IDIQs, MACs, and BPAs positions Atlas Research for future growth and scale. He is a recognized expert and speaker in health IT, artificial intelligence, data science and big data, DevSecOps, orchestration, containerization, systems engineering, the software development lifecycle, cloud computing, product development, portfolio management, biometrics, acquisition process, telecommunications, logistics, supply and maintenance, and manufacturing.