Understanding the Player

podcasts Aug 23, 2020

 

Marc Sophoulis

Welcome to another episode of Crunching the Numbers with Marc Sophoulis and Shane Liyanage from Data Driven Sports Analytics. Shane, thank you so much for joining us once again.

 

Shane Liyanage

Hi, everyone. Hopefully you've had a great week and started to master isolation.

 

Marc Sophoulis

Obviously we're still in quite a challenging situation Shane. We're a long, long way away from each other, but hopefully we're bringing people in touch with the game and keeping people in touch with the game of tennis.

 

Shane Liyanage

That's the whole purpose. I think we can all learn a lot from this period. And I think by the end of it, there's gonna be a few different ways to communicate and learn that we're all gonna take forward.

 

Marc Sophoulis

Absolutely. And obviously, we're brought to you by The Tennis Menu and Data Driven Sports Analytics too. The Tennis Menu is my website at the moment where we're doing a lot of development based activities and you do a lot of your data and obviously analysis of the game, which is fantastic. And hopefully people can take a lot from what we both do and be able to use that in their own tennis and their own coaching and in their own development from this day on.

Today we thought we'd dive into a little bit of a topic Shane, which probably is close to our hearts because we've just sort of come off the Australian Open period where we had some great success with Nicholas David Ionel from Romania. He won the Australia Boys doubles and has gone on to do some fantastic things with his game. And look, at this stage he's about 500 in the ATP rankings at the age of 17. He's almost top 10 in the world in juniors. We used a lot of the data that you brought to me, which I thought was really good to be able to get his game to another level. So do you want to talk us through our topic today of knowing your player through data.

 

Shane Liyanage

Yeah, thanks Marc. And I think it's good to perhaps start with the information that Data Driven Sports Analytics has a match library of over two thousand matches, particularly from the last four years. It's at various levels - ATP, WTA, Challenges, Futures and even college tennis. And this is not just a video library. The data from those matches is collated into a database that I've built. And that is perhaps the starting point in terms of; when I'm working with a player, I'll look to that to see whether I have anything on them already. My aim is to get about 50 matches on a player before working with them, but that's not feasible in every instance. With David unfortunately we didn't have any in our database to start with. So I had to go and acquire 15 matches and then had to annotate them using a custom tagging panel. Just on the tagging part, the focus really is on the key events that I know that the coach is looking for, but also tagging in a manner that the information can go into the database and been useful for further reporting, analysis work or down the track as well. We do some predictive analytical work and to make sure that we're tagging the right information to be able to do all of that down the track. I touched on the fact that there's a dashboard. So I've built a dashboard that sits on top of the database. So every time the match goes into that database, the dashboard updates with the information and a coach can actually log in and see a dynamic report, so they can filter what they want, get a PDF, or some coaches who are not quite as comfortable using a dashboard, they prefer me to send them a PDF of static data on their player. I just want to stress here. This report really is just a conversation starter. It's a report from me to the coach saying, "here's your player from a data perspective. Let's talk".

 

Marc Sophoulis

The important part about what we do before we get working with a player, similar to what we did with David is we want to get as much knowledge of this athlete as we possibly can. For me as a coach I need to know what this player does really well and what they can really improve on, because you can have an opinion. I think data really makes sure that we can take the opinion out of what we do and make it more realistic, but then also understand where the data comes from and why it is actually happening. There's two aspects - There's the data, which is the science, and the art of what we do with the data that the two most important things when we do this sort of stuff.

 

Shane Liyanage

Yeah, absolutely. And I think throughout our podcasts and various social media you'll see that we use that art/science, needing to be together to deliver that holistic message and also enable that interpretation to be better for the athlete.

 

Marc Sophoulis

Absolutely. We want to talk a little bit about how you went about this specific work we did with David. And basically from the word go when I was talking to David and negotiating our deal and looking at how we are going to work together. You were doing the work in the background, and it was quite funny because I even come to terms or agreements with the management and the family. And you'd done so much work already on David and his game. Do you wanna take us through the kind of thing you did and what you gave me?

 

Shane Liyanage

Yeah. And I think that situation was maybe a little bit unique because it was only on match day of his first match that we knew for sure that we were working with him. But with the work we do, the timing's really important. So when there was even a rumour that we were potentially working, I had to start at that point, particularly because David being a junior and not having a big presence on the ATP Challenger Tour at that stage. There just wasn't a lot of matches that we'd already got in our database. I had to start from scratch really with him. So as I mentioned before, there was some video tagging work, but then I think it's a matter of using the data from there. So I had to look at things like,  what was his preference on serving? What was his return position? Where did he like to go on 30-30? How was he doing in the 0-4, the 5-8, the 9+ rally categories? How was he starting matches? So there was a whole lot, and the list that I've just mentioned isn't exhaustive, but trying to bring all of that together in a succinct report and then really, even once I send that report to Marc, it's about discussing that and making sure both of us are clear in terms of interpretation.

 

Marc Sophoulis

Yeah, absolutely. I think that's probably the key to utilizing data or a data analyst like yourself and then obviously the coach, because I think it's a matter of collaboratively coming up with what is needed and what is next. You gave me some return of serve patterns of David and you gave me this data that said that when he's on the deuce side, when he returns forehand down the line, he has more success. I looked at that and went, yeah, that's really good. That to me is not the right pattern I want him to play. He might win a lot of points in that situation, but generally he's probably hitting the ball late because of his grips and his swing-lines or his court position are making play down the line. It's not a conscious effort to go down the line. So we had to collaborate the data with the thoughts and the art together and make sure that we didn't just say, well the data says that down the line is the most successful spot, we're going to go there. There's always a reason why that happens and it could have been just the late return of serve and the server wasn't ready for it, or whatever it might have been. But I think it's important to make sure that the data matches up with the game style of the player and the patterns of play that you want to play as well.

 

Shane Liyanage

I think that's a really big point Marc, because the data that we had there was historical data. It's not necessarily where you want the player to be in two, three, four years time. So that was a clear indication where, yes he historically had success playing a certain way. But if he wanted his game to be robust and hold up as he went through the levels, we needed to change something. And that really was only figured out when we talk to each other and had that vision for the future.

 

Marc Sophoulis

Well one of major things I think you gave me in this situation when you did the research, was you gave me this research around his starting of matches. You gave me research that he was winning only 0.8 of a game in the first 4 games of every match. What does that mean to you in data terms?

 

Shane Liyanage

Generally, that means you're down a break, if not two breaks. In terms of the score line, it's either 3-1 or 4-0. So he's immediately on the backfoot. I suppose there was a couple of things. So that's the scoreline. But we also dug a little bit deeper to try and understand why. His his first serve numbers were a bit low, he was hitting more unforced errors. He'd return positions were not really consistent with the way he returned the rest of the match. There were all those elements, but again, the art and science approach that we took identified, yes there was some data things that we could fix, but there were other parts which physically, mentally were required to try alleviate that negative start.

 

Marc Sophoulis

Yeah, and I think that's where art and science came in, because I felt like his pre-match preparation was really poor as I've mentioned in previous podcasts, and we had to make sure the pre-match was sorted to make sure he was ready to play those first four games from a physical and mental perspective. That was one thing. But also, you are right. I think the serve data for me was the most important. So what I didn't know about David was he didn't like to serve first because he felt really nervous in the first game. So, everytime he served first, he lost the service game. And basically we had to work out why he was doing that, why he thought like that. He tried to hit every serve with full speed in the first two or three games, as opposed to serving for patterns and looking for the plus one. We totally changed the thought around trying to hit aces to trying to serve to a pattern and then utilize his strengths on the second ball. And that was another little element of that, because I felt like you gave me data based around; If David hit three forehands in a point, he won the point 93% of the time. And without knowing that data I wouldn't have been able to grab the patterns and go "Hey David, this is what you historically have done. This is what we need to do to play our best tennis". So that data for me helped us to be how to create a winning game plan in those first four games.

 

Shane Liyanage

There's a few things as well, so obviously at this stage, largely the communication is still between me and you. So between the coach and the analyst and I know you drip fed bits and pieces to David, particularly more around the video. But maybe you'll want to tell our viewers how much of it you were using at this stage with David.

 

Marc Sophoulis

It's interesting because data can be one of those things where you look online right now and there is so many data analysts out there throwing all these numbers, throwing everything at you. You know, "this is what the man of rallies are. How the percentage of 0-4 is where you win more of your points. These are succesful patterns of play". But what do you use and what don't you? And that's the art of what you do as a coach. Most players can only accept one, with maybe two bits of information at a time. So every single day I drip fed one bit of information and then the next day I would reinforce that bit of information. And if he was ready, I would lay the next bit of information to him. And I think it's about layering your messaging, which is really critical to the data filtering as well. You take the data, you use one, you practice it that day, you see how it is the next day. You reinforce it if it's still not quite where you want it. You continue with that. If it's good or if he understand it, you drip feed the next little part in that you reinforce the next day and you drip feed again. And that's how it looks from a coaching perspective. And I think a lot of coaches out there try and do it all at once. Even players, they go "I'm going to grab all this information. I'm gonna utilize all this". It's very hard to do that. So I think it's a matter of drip feeding for sure.

Let's take a look at the first four shots of the match with David. I think that was the next element. And we talk about data. Do you want to just go through Shane, what are the rally lengths? Obviously you did a great graphic for me for a presentation I'm doing at the moment. The rally lengths of 0-4, 5-8, 9+, and then the amount of time that those points are played for.

 

Shane Liyanage

The 0-4  was roughly 67% of the time. And it's usually around the four to five second mark that the rally went there. So what that tells us is that's where the majority of the points are played. So I think it was big as well going back to the David example. On serve he was winning only 68% of the 0-4 and then on return, he was winning only 20% of the 0-4. So that was a key area for us immediately to go, well, what's going wrong in that 0-4 and part of it was he wasn't getting returns back into play. Part of it was perhaps he was returning to close the lines and not trying to attack through the middle of the court. You can probably go into a bit more detail on what we did.

 

Marc Sophoulis

To me that stood out the most with David and especially this example that we're utilizing today. Knowing that 67% of points finished within four shots and seeing his rate was round 45-50%, he was only winning in the zero to four ratio overall. That was never going to win him a match and it wasn't going to make him win consistently in matches. So we had a massive focus of those first four shots in practice. So everything we did in practice was based around aggressiveness, having an intent, having a mindset of those first four shots - I'm going for it. So basically, we served to spots and used the first shot as a forehand. And on return we returned middle and then took first strike on the forehand as much as we could. That was our focus in our practice session as much as we possibly could. And we tried to focus in on those first four short, sharp, first five seconds of point play that we possibly could do. I knew he was a great endurance athlete, didn't have to do a lot of work on the endurance points, the 9+ and 7, 8, 9, 10 balls was ok. It was that starting point where his mind set and his game just wasn't on. So we changed his training to really replicate those first four to five seconds of play. That to me, Shane was probably the biggest difference. And it has been in a lot of players that I've worked with over the last couple of years. It's been the biggest change in getting successful athletes as opposed to athletes just can go out there and play.

 

Shane Liyanage

And just to end on that, the 15 matches of data that I gave you was before he played. And since then, he's winning overall close to 56% percent of the 0-4 points. So that's a significant improvement in that part of his game.

 

Marc Sophoulis

So that 10% improvement has seen him excel and win a future in Greece since that time and obviously just changed his mind set around playing the faster points better, the shorter points better, which is a huge thing. Just finally I wanted to touch on how we talk a lot about winning that 0-4 ratio. We looked at, what do we need percentage wise  to win in the 0-4 ratio for us to be successful in matches, because obviously that's where almost 70 percent of points are finished. And this is a really big stat for a lot of people out there to really focus in on.

 

Shane Liyanage

I think we touch on this almost weekly. It's a really important point that we do want to stress, because it's such a huge proportion of the overall points in a match from data that I've collected in the Data Driven Sports Analytics database. This is two thousand plus matches. The 0-4 category is 67.4% of overall points in a match. So it's a big proportion. And for the global tennis summit that Marc was doing a presentation, I actually gave him some scenarios which used a hypothetical player or hypothetical scenarios where a player would win, let's say, 30% of their 0-4, 60% of their 5-8, 80% of 9+. What's their sum total point winning percentage in that first scenario? A lot of people thought "OK, that's great". That's something that you put money in and sign up to that scenario, but you only win 42% of points. So in 99 out of 100 matches, you probably lost that match. And then to sort of flip it in another scenario where you're winning 60% of your 0-4, but you're only winning 30% of your 5-8 and 30% of the 9+. If you manage to win 60% in the 0-4, you're winning over 50% of your total points in the match and yet you're probably a winner. So it's something we gravitate to, but it's something really important that you focus your attention on areas that you can maximize the most benefit.

 

Marc Sophoulis

It's just over half the amount of points, really, isn't it? It's pretty incredible that you can think that way. If we as coaches and players can really structure our training based around those first four shots, we're obviously going to have a better understanding of how to win matches and be more successful. This is a process that you and I work together on, and it's quite easy to have a data analysts and a coach working together. But even if you're out there, get as much data as you can of your athletes, make sure you understand them before you make any kind of thoughts around game planning or game structure because if you don't know your player first, it's going to be extremely hard to plan for your matches and plan patterns of play and have a successful outcome for your athlete.

 

 

So the other thing I wanted to mention just to finish off is, it's really important to merge the art and the science. It's not one. It's not the other. Grab the data. Utilize just the little bits and pieces that you need, the little gold nuggets that are in that data and know that you don't have to use every single number that you find in your data analytics. I know that's not music to your ears Shane. When I tell you that, you know, you put in hours and hours of work into data and then I use two or three things, but it's it's really important that we just use the gold nuggets, because players can't digest too much information all at once.

 

Shane Liyanage

To the data analysts or the data scientists working in teams, I just want to stress to you that the information will not be wasted. I think you capture that and the analysis will still be there. At some point down the track, that analysis might be useful and it may be you need a block of six months or three months or something to actually sit down and work through what are the actionable insights from that analysis. But don't be disheartened that the coach doesn't immediately use some of the work. I can say from my experience with Marc is there's a lot of times where a lot of analysis work goes in and he'll just pick out two things and you feel like, OK, I just spent so much time. But it's an important part in the process to understand that the analysis is really important and useful, and you do that accurately and with robust mathematical techniques, but also understand that not all of it will be used at that specific time.

So this podcast was really looking at getting an understanding of a player with data, and that precursor, that initial data collection work that you do to set the scene. Next week on our podcast, we're going to start to go through what happens in a tournament and how we use data for game planning and tournament review and things like that.

 

Marc Sophoulis

Shane as usual. Thanks so much for your input and what you do for tennis, but also for myself as a coach out there and obviously at Data Driven Sports Analytics you can find all of Shane's research. He puts it up there. It's fantastic to read and even though I speak to you a lot Shane, I do learn a lot from what you post on social media, so thanks for the work you do.

 

Shane Liyanage

Thanks, Marc. And see you next week.

 

Marc Sophoulis

Thanks again. You've just joined us again on Crunching the Numbers for another episode. That was knowing your player through data. You can find me on The Tennis Tenu at www.thetennismenu.com and all over social media. I really appreciate you listening in and hopefully you've taken something from this podcast. We hope to continue to bring you all the latest in tennis data and keep you sane in these tough times with COVID-19.

 

Marc Sophoulis

Please stay safe. Stay healthy. See you next week.

 

For a more detailed look at Crunching the Numbers, don't forget to checkout Shane and Marc within the Coaches Cupboard, where they do a deep dive into the latest trends in the world of tennis.