Premier League 2017-18: Manchester’s Rise and North London’s Fall


How 2016-17 Expected Goal and Save Values Help Illustrate What’s to Come in the Premier League

Either Pep Guardiola has been taking note of some new metrics or his football sense is just that keen. Regardless, the argument can be made that Joe Hart had the last laugh.

For last season, that is.

The goalkeepers Guardiola enlisted over Hart didn’t only fail the eye test as Manchester City fell 15 points shy of the Premier League title in the manager’s first season. Claudio Bravo and Willy Caballero were a measurable problem for Guardiola, and the manager wasted little time rectifying that this transfer window by bringing in Ederson from Benfica.

STATS has refined some of the most advanced metrics in football, and using them gives insight into just how teams measure up against expectations. Expected goal value uses machine learning and historical tracking data to address how likely a goal is based on the location of a shot, the position of the defenders and manner of the attack. Possibly the best way to think of it in terms of how it measures a player’s worth is that it assesses the individual against the league average.

Man City allowed 39 actual goals. Their expected goals against came in at 36.9, and the plus-2.1 differential – meaning they allowed more goals than they should have – accounted for the fourth worst in the league ahead of only Crystal Palace, Watford and Liverpool. The contributions of Bravo and Caballero with expected save differential – calculated by subtracting expected saves from actual saves – was minus-5.7, meaning they did not save nearly six more shots than the average keeping tandem would have. That ranked ahead of only Crystal Palace’s keepers. With Hart featuring the year before Guardiola’s arrival, City were at a perfectly acceptable shade above the league average at +0.1.

That very last line of defense is, of course, not the only area where City spent this summer. The signings of Benjamin Mendy, Kyle Walker and Danilo figure to address any defensive shortcomings for a side that’s dealt with inconsistency and injury even after bringing in John Stones last year. Without even getting into the addition of Bernardo Silva and having Gabriel Jesus for an entire season, those signings may amount to the changes City need to bring Guardiola yet another trophy in yet another league.

If you’re not yet convinced, let’s now consider improvements they could see from within their established attack by surveying Guardiola’s most trusted finisher and his supporting cast.

Sergio Aguero’s -4.1 expected goal differential – calculated by subtracting an individual’s expected goals from converted goals – last season was the third worst in the division, but he was at +3.6 the previous season. He still scored 20 goals last season, and he did so while wasting chances. Of the seven 20-goal scorers from the past two seasons, he’s the only to post a negative xG differential. That’s a class he’s repeatedly been a part of, so it probably follows that he’s quite unlikely to waste as many chances going forward.

Now consider David Silva (-1.9), Raheem Sterling (-1.8) and Kevin De Bruyne (-1.6). Silva was right at that rate in 2015-16 while Sterling was slightly better (-1.0), but De Bruyne’s was +2.5.

If Aguero and De Bruyne get back up to a level we know they’re capable of and City put them in similar situations to score, the tandem could theoretically account for 11 more goals. Add that to the prospective goalkeeping improvement, and City have the possibility for a staggering overall goal differential increase.

But a similar argument can be applied to the attack of City’s closest rivals, who spent big this summer on efficiency they sorely need. If that works out, the managerial rivalry we saw between Guardiola and Jose Mourinho elsewhere could reach maximum velocity in England.

Manchester United’s potential for improved efficiency is enormous

After a 4-0 loss at Chelsea on Oct. 23, Manchester United found themselves five points back of the eventual Premier League champions.

From there, the Red Devils allowed 17 goals in 29 league matches, conceded more than a goal once and kept 14 clean sheets. On paper, that certainly looks like the defensive capacity to give a club every opportunity to make up those points to win the league.

Yet they drew 13 of those 29 matches, lost two and did not traditionally qualify for the Champions League because of it. Their five goalless finals in that time included Old Trafford disappointments with Burnley, Hull City and West Bromwich Albion.

To say Mourinho’s side left points on the pitch is an understatement, but how to quantify it? United’s goals for differential was substantial. Try -16.3, or third worst in the division behind Southampton (-27.6) and Stoke City (-17.2).

Remember, that’s how we quantify finishing measured against the league average. It was a systematic problem of wasting chances with no player posting an xG differential of even +1.0. Jesse Lingard (-3.4), Paul Pogba (-3.3), Marouane Fellaini (-2.8), Zlatan Ibrahimovic (-2.4) and Wayne Rooney (-2.0) were the most responsible parties. Some of those names are gone, but attackers like Marcus Rashford and Antonio Valencia also produced negative differentials.

In this case there seems to be a need to address styles of play. If Mourinho is able to do that and get even average outputs from key attacking players, the shift in goals scored could be significant enough to make United – not City – the team to bring Manchester back to the top.

Now add a new No. 9 to the mix. United’s potential appeal for 2017-18 only increases with Mourinho’s addition of a player who could sway efficiency back in the right direction – if United use him properly. Romelu Lukaku’s +9.9 xG differential might not be sustainable on quite that level, but there isn’t much of an argument against Mourinho having added a finisher who operates at impressive efficiency levels to an attack that’s already bursting with measurable potential.

Lukaku’s xG differential trailed only Harry Kane last season. That begs this question: If Spurs couldn’t win the league last year, is it realistic to think they can now?

Last season might have been Tottenham’s best chance

Tottenham had some of the most quantifiably effective attackers in the Premier League last season, and it goes well beyond Kane.

Kane scored 29 goals in 30 matches, which is impressive enough on its own, even if you normalize the seven goals he scored in two throwaway matches at season’s end. It becomes even more remarkable when considering his 15.7 xG. His league-leading +13.3 xG differential implies he was consistently finishing chances the average player wouldn’t.

He also led the league in 2015-16, but with a considerably lower and more sustainable differential (+5.2). His mark last season was significantly better than next-best Lukaku, and, for comparison’s sake, any of the big names for the Spanish giants. Lionel Messi led in Spain with a +9.3 xG differential, and no one in the Bundesliga, Ligue 1 or Serie A topped that.

It wasn’t just Kane for Tottenham, though it was an alarmingly top-heavy club performance. Spurs scored a league-best 86 goals, which was +17.6 of their expected goals for. But their xGF differential was thanks entirely to three of the top six individuals in expected goal differential being a part of White Hart Lane’s final season. Heung-Min Son (+6.2) ranked fourth and Dele Alli (+5.3) was sixth.

The three scored 61 of the club’s goals despite Kane missing eight matches. Spurs lost none of them and dropped six points, which still wouldn’t have been enough to win the league. The success without Kane hinged heavily on eight combined goals from Son and Alli.

Tottenham’s depth seems questionable and is already under examination with Walker joining City and replacement right back Kieran Trippier suffering a preseason ankle injury that will cost him the start of the season. How Mauricio Pochettino closes the gap between Tottenham and the top with a fully healthy and optimally efficient side is no small question. Consider any personnel concerns that may arise, or any dip in form, and producing the level of efficiency they’d need to win the league seems nearly impossible.

Hugo Lloris’ stellar 2016-17 furthers that point. Tottenham’s expected goals against was 36.0. They conceded just 26, for an xGA differential of -10.0. Lloris and Michel Vorm combined to post the league’s second-best expected save differential with a +10.4 mark, meaning they saved at least 10 goals a league-average keeper would have let by.

For that to be sustainable, Lloris will have to prove Tottenham’s consistent xS differentials of years past – +0.7 in 2015-16, +0.2 in 2014-15 and +1.1 in 2013-14 – were somehow the anomaly rather than the norm.

That’s a glimpse into how Spurs could struggle to entertain Wembley. But even their North London rivals who perform better at those grounds could have similar concerns.

Cech can’t save Arsenal to the top

Spurs’ goalkeeping was great. Arsenal’s was slightly greater, but that has to be of the greatest concern to Arsene Wenger for reasons beyond his No. 1 being on the wrong side of his prime.

Petr Cech helped the Gunners to a league-best +11.7 xS differential, which was the best in the Premier League over the past five seasons.

It’s difficult to see this as sustainable, particularly since Arsenal’s xS differential in Cech’s first season at the club was +4.4. While it should be of some comfort that they added Ligue 1’s most efficient striker in Alexandre Lacazette (+8.3 xG differential) to balance that on the other end, there are further reasons for concern.

Arsenal gave up 44 goals as it was. If we add seven to that, bringing them in line with their 2015-16 save differential, their goals against last season jump to a tie with West Bromwich Albion for eighth at 51. No club in the last 15 Premier League seasons has conceded more than 50 goals and finished in the table’s top four. Put the Gunners at the league average by adding 11 goals to their save differential, and they’re tied with Burnley for 10th in goals against.

And if winning the league is their ultimate goal, Arsenal have a great deal of work to do with a back line and midfield not so dissimilar to last season’s. No club has ever won the Premier League while allowing more than 45 goals.

Maybe that’s where Chelsea finally come in as the London club with the best shot at the title.

A third title in four years?

The Blues conceded 33 goals last season, which was third to Manchester United and Tottenham. That number matters because it was consistent with the club’s expected goals against (31.8). They didn’t have a keeper constantly bailing them out. Their system worked.

The problem for Chelsea comes with scoring, where Antonio Conte got much more last season than his club may be capable of moving forward. Their 85 goals for ranked second to Tottenham, but their +20.6 xGF differential was a full three goals ahead of that previously discussed unsustainable Kane-Alli-Son-powered mark for Spurs. Granted, Chelsea’s productivity was more spread out among attacking weapons, their stability absolutely thrived in a three-back system, and they added Alvaro Morata’s potentially impressive efficiency.

Those three terms – productivity, stability, efficiency – are telling, and they’re more measurable in football than ever. Conte comfortably won a title in his first season at Stamford Bridge by implementing them in impressively quick order.

But that could ultimately mean little this season considering Guardiola’s established plenty of his own.

Can Morata Replicate His Real Madrid Efficiency at Chelsea?


Let’s start with the positives: Alvaro Morata scored more La Liga goals per minute played than Cristiano Ronaldo and Luis Suarez last season, he required fewer shots per goal than Lionel Messi, and Real Madrid didn’t lose a league match in which he played.

Left at that, Chelsea fans might be thrilled with the idea of a 24-year-old striker who’s already a veteran of Europe’s highest levels lining up in front of Antonio Conte’s midfield. That line of thinking doesn’t properly consider how Morata scored his goals and how he might be asked to finish for the Blues in less conducive types of play. That’s something that can now be quantified with STATS Playing Styles.

On the surface, Morata’s efficiency is hard to not laud. His 15 league goals came at a rate of one every 88.8 minutes, which, among the seven La Liga players with at least 15 goals, was only bettered by Messi (76.5). No one else came close and, for what it’s worth, Antoine Griezmann’s 16 goals required 191.6 minutes each.

The same conclusions can be drawn from Morata’s shooting. He needed just 3.7 shots per goal last season, which was unmatched by anyone with more than 10 La Liga goals – including stars like Ronaldo (6.4) and Messi (4.7).

Getting even more specific by using STATS’ expected goal value metric, Morata was expected to score 10.6 goals. Expected goal value is assigned based on the probability of a goal being scored from the position of the shot. His xG difference of 4.4 was better than not only Ronaldo (minus-1.9) and Griezmann (2.6) but also that of the man he’s replacing – Diego Costa (1.7).

And, most importantly, it resulted in wins. Of the 26 league matches in which Morata saw the pitch, Real Madrid went undefeated and averaged 2.69 points.

All that, yet in the past two seasons Morata has failed to assert himself as a top striker for Juventus and Real Madrid. The why involved here often feels like something a manager sees on the pitch that we can’t always accurately quantify. Again, that’s no longer the case.

There’s some validity in arguing his numbers were better because he was frequently used as a substitute and able to put more effort into his average of 53.9 minutes per appearance than a 90-minute player. However, he scored 11 of his goals in his 14 starts, so there’s something more to it than simply coming on with fresh legs.

His numbers start to come back to the realm of normal when considering nine of his goals came against the bottom six teams in the La Liga table. Then consider that over the last three seasons the bottom six clubs in La Liga have allowed 72 more goals than the bottom six of the Premier League, and Morata’s appeal begins to fall off a bit.

But that’s mostly surface-level stuff. It gets more interesting with club specifics. Here’s why such impressive productivity probably isn’t possible for Morata as a starter in the Premier League.

It can be argued that what makes those elite-level scorers 90-minute players is an ability to score in various playing styles. It’s no surprise that even goal scorers like the diminutive Messi lean on crossing more than any other playing style to score goals. It’s a proven attacking method that will always have its place in football. Seventeen percent of Messi’s shots and 30 percent of his goals occurred in the presence of a crossing style. Ronaldo: 28 percent on shots and 28 percent on goals. Suarez: 22/26. Costa: 33/37.

Morata’s crossing percentages were 44 percent for shots and 52 for goals, which isn’t necessarily a good thing for someone who’s about to change systems.

It follows that his finishing might be limited if he’s not playing for a club that doesn’t distinguish themselves from others in that way.

You guessed it. Chelsea didn’t distinguish themselves from the league when it came to crossing. They were exactly at the league average under Jose Mourinho and Guus Hiddink in 2015-16 and -3 percent under Conte’s league-winning side.

Chelsea’s overall 2016-17 playing styles measured against Premier League averages (0%).

That’d be all well and good if Madrid was also around the La Liga average, but they’re not. Rather, it becomes especially alarming for Chelsea when comparing their playing style under Conte to Real’s. The teams were similar with certain styles, but Real Madrid thrived at 40 percent above the La Liga crossing style average.

Real Madrid’s overall 2016-17 playing styles measured against La Liga averages (0%).

Now let’s get back to the fact that Morata came on 12 times as a substitute and try to quantify what that could mean for him. In the 60- to 90-minute range, Real’s attacking threats went wild, particularly when the score was even or they were losing. Real’s crossing style increased to 143 percent of the league average in those circumstances.

Real Madrid’s playing styles when tied or losing in the 60- to 90-minute range.

Compare that to Chelsea in the same scenario (+24 percent crossing), and the way Morata could be forced to play in late-match situations with tight scores at Stamford Bridge might seem a bit foreign without balls flying into the box at a level he’s used to – and that’s without even mentioning the luxury of those fresh legs he often had with Real.

Chelsea’s playing styles when tied or losing in the 60- to 90-minute range.

Morata may very well show progress as a young striker that wasn’t possible in his reserve roles at his past clubs. He may very well score 15 goals for Chelsea. He may very well score 20 like Costa did last season. He’s just unlikely to do it at last year’s rate.

LMA Chairman Wilkinson: STATS Has ‘Anticipated What the Future Might Bring’


Howard Wilkinson’s managerial career was at its apex around the time the Premier League was formed, yet he remains the last English manager to win the country’s top flight. That came with Leeds United in 1991-92 in the old First Division’s final campaign, and the Premier League’s subsequent 25 seasons have been won by Italian, Scottish, French, Portuguese and Chilean managers.

Wilkinson has seen analytics and video analysis revolutionize the game during that time. What’s remained a constant is the value in beating opponents to the facts, and much of that now hinges on innovations from data providers.

“Without the facts, it’s very difficult to make effective decisions,” said Wilkinson, the League Managers Association chairman. “And although intuition comes into it, intuition only comes with experience. Therefore, the two are linked, and the longer you use data analysis, the better you become at making those intuitive leaps, which are necessary.”

STATS has taken such expert opinions into account in its attempt to revolutionize football analysis with STATS Edge – an intuitive search and analytics application that leverages artificial intelligence so clubs can find specific game clips and analyze complex patterns with speed and accuracy never before possible.

“The reason STATS and their forebears have been around so long and have been successful is that they’ve not only moved with the times, they’ve anticipated what the future might bring,” Wilkinson said. “And that plus the amount of games, players they’ve analyzed has allowed them to build models, which are models for football, not just for that club.”

Crossing Over: Lukaku’s Success Will Depend Heavily on Manchester United’s System

STATS’ playing styles data shows the striker’s goal efficiency could flourish at Old Trafford – if his manager puts him on the end of crosses and allows him to operate in space

To some, Jose Mourinho’s frustration with wanting more from Manchester United’s transfer window will be received as the manager being his contrarian self. His club just spent big to bring on the Premier League’s No. 3 scorer over the past four seasons as he enters his prime.

To others, his complaints are valid given the insipid nature of the Red Devils’ finishing last season and the subtraction of two names in a historical class that newcomer Romelu Lukaku has years of remaining work to join – Zlatan Ibrahimovic and Wayne Rooney.

The second of those names is one of only three players to reach 50 Premier League goals at a younger age than Lukaku, so the traditionally sexy question of whether a high-profile move to Old Trafford from Goodison Park can benefit Lukaku’s career the way it did Rooney’s is going to linger.

Forget all of that for now.

There are more fascinating ways to consider the Lukaku move, and the real predictive analysis has far less to do with Mourinho’s mouth or Rooney’s legacy after a season in which the England great’s withdrawn role was anything but comparable to what Lukaku’s will be.

What’s of more relevance – and can be properly considered now with a dig into quantifiable player- and team-tendency data – is that the 24-year-old Belgium international is entering a United system in which he seems to have the opportunity to succeed on levels similar to those he enjoyed last season at Everton. He can even surpass them if he can become more effective in attacking situations without goal-facing space.

It’s noteworthy that Lukaku went through the league last season with only one goal from the penalty spot and one from a free kick. Virtually all of his scoring threat comes from the run of play, so he’s a particularly worthwhile player to evaluate with playing styles.

Lukaku scored nine of his 25 league goals off of crosses last season, despite operating in an Everton system with a playing style that came in just below the league average of time spent in crossing scenarios.

In a comparison of the 2016-17 styles of United and Everton against league averages, Everton rarely differentiated themselves and were at -3 percent of the league crossing average. Manchester United were positive 10 percent.

The playing style web shows the league average as the 0% differentiation line.

Switch Lukaku’s shirt from blue to red and, without even looking at the data, the first thoughts that come to mind are promising with outside players such as Antonio Valencia and Marcus Rashford – and possibly Ivan Perisic if United eventually agree on a price with Inter Milan – putting balls into the box with frequency that Everton couldn’t match. Playing style numbers back that up.

Where the Toffees did distinguish themselves some stylistically were in counter attack (+14) and fast tempo (+12 percent). Notice United’s similar counter (+15) and a drastic increase in fast tempo (+81), which one would think bodes well for a No. 9 threat such as Lukaku who’s physical profile at least passes the eye test in comparison to Ibrahimovic.

What’s more is the playing style presence during their goals linked up for the top two categories. Lukaku existed in a crossing style during 36 percent of his goals and 20 percent came from direct play. Ibrahimovic: 31 percent crossing and 19 percent direct play. But that doesn’t mean Ibrahimovic was necessarily effective in Mourinho’s system.

One can argue Lukaku did more with less last season than the big Swede by looking into STATS’ expected goal values, which is an efficiency metric determined by the likelihood of a goal being scored based on the position from where a player’s shots were taken. Lukaku’s xG for the season was 15.1, and of the 25 he scored, he needed 4.4 shots per goal. His +9.9 differential was unmatched in the Premier League.

Ibrahimovic’s xG was 19.4. He scored 17 – which on its own isn’t necessarily relevant in comparison with other players considering his late-season knee injury – but it’s worthwhile to note his -2.4 differential and that the goals he did score came at a rate of 7.1 shots per.

Here’s where the question of how Lukaku can still grow as a player comes in and what United might need to do to maximize his efficiency if he ends up struggling in different styles.

There’s little yet to support Lukaku can be the kind of player Ibrahimovic has often been with back to goal and tighter marking while playing for clubs with such possession-based, attacking-third threats. The Belgian has on occasion been criticized for a heavy touch, which may be on display with more time operating in less space within scenarios of sustained threat and build up.

At least in the Premier League, it’s hard to say he’s ever experienced the tight quarters he may see with United, who operated last season 30 percent above the league average of sustained threat and 41 percent higher in build up. But the Everton managerial change away from Roberto Martinez’s style before last season might have stunted any growth that was happening. That, or it might show how Mourinho can tinker his system to get immediate results from Lukaku.

Everton’s build up in 2015-16 was 25 percent higher than the league average, so far closer to Manchester United’s ’16-17 rate than Everton’s under Ronald Koeman (+4). Forty-six percent of the playing style presence during Lukaku’s scoring was crossing, but Everton’s crossing style (+5) didn’t exactly exploit that to great ends. Twelve percent of Lukaku’s goals came from build up.

Now consider his efficiency. In that final season with Martinez, Lukaku scored 18 goals but had an xG of 21.2 with a 6.7 shots-per-goal rate.

Considering his efficiency under Koeman, it follows that Martinez’s system probably didn’t provide the best styles for the striker to convert opportunities.

So for Mourinho, it might be more about maximizing the playing styles in which players such as his new striker can succeed rather than continuing to buy, buy, buy until the end of the window.

It can be statistically argued he didn’t maximize Ibrahimovic after signing him last year, and Mourinho’s first season at Old Trafford ended with the Red Devils eighth in the league with 54 goals (excluding own goals for). That was -16.3 of their expected goal value (70.3). They were outscored by Bournemouth, and the Red Devils’ differential between goals for and expected goals for was worse than every club other than Southampton (-27.6) and Stoke City (-17.2).

Juan Mata scored six league goals, but no one else other than Ibrahimovic topped five. Players not putting the ball in the net – on the surface, that seems like something a manager should be able to vent about. But playing style? That falls on the manager.

Consider the sequence of events and place blame accordingly.

Allardyce: Video’s ‘Small Gains Are So Important’


Sam Allardyce recalls the time around the turn of the millennium when statistical and video analysis came to the Premier League. Analytically minded teams had a substantial advantage over others before match-preparation tools grew in popularity.

In the nearly 20 years since, it’s gone from innovative to mainstream to inundated, but the benefits remain in matters of methodical proficiency. Now more than ever, it’s about streamlining the process to reach conclusions faster and implement the findings into training earlier.

That’s exactly how STATS plans to revolutionize football’s analysis process this summer with the launch of STATS Edge – a search and analytics application that leverages artificial intelligence so clubs can find specific game clips and analyze complex patterns with speed and accuracy never before possible.

“Video clips are massively important. Short bursts of information. Not just on (your own players), but also what we do on the opposition pre-match and post-match,” Allardyce said.

“If you talked about ’99-2000 when statistical analysis came in, you could gain 5-10 percent because not everybody wanted it, not everybody wanted to use it, not everybody had it. So it was much easier then to gain a bigger percentage advantage on how you applied that to your team. But now it’s like everybody’s doing it and the gains are so small now, but those small gains are so important.”

Aston Villa’s Bruce: Video Analysis a ‘Very, Very Important Weapon’ for Every Top Club


From the First Division to the Premier League and an unlikely FA Cup final, Steve Bruce’s nine-club managerial career has often put him in situations where he’s trying to guide a club’s promotion or orchestrate a Wembley upset over a European giant. It’ll be no different next month as Aston Villa begin the nine-month task of attempting at returning to England’s top flight. Regardless of his place in English football, an ever-present has been video analysis’ growing importance in match preparation.

“Video analysis has become so much a part of your work now, you probably scratch your head and think how would we manage without it?” Bruce said.


STATS SportVU: The Independently Validated Optical-Tracking Solution


In its mission to make STATS SportVU the global standard for accuracy among optical-tracking systems, STATS commissioned a 12-month independent study to validate STATS SportVU alongside a global-positioning system and a local-positioning system with the three tested against gold-standard technologies. This paper summarises the results of the study for sports scientists.

As Performance Data Becomes More Complex, STATS Finds Ways to Streamline Analysis


The volume of sports data available to teams and leagues is expanding exponentially. It’s causing sports to move faster and competition to become even more intense. It is increasingly difficult to capture all of the relevant data points available and distill the complex information contained in those millions of data points per season into a series of simple representations that can be quickly absorbed, tailored and shared to enhance the teams’ performance and win more games.

STATS’ vision is to collect all of the world’s sports data and make it simple, meaningful, fast and intuitive. The aim is to bring new levels of context to sport through the application of cutting edge technologies – such as computer vision, machine learning and AI – to help teams game plan, build their teams, and maximize athlete training programs to find their winning edge.

Brentford Use Data and Analytics with Hopes of Premier League Promotion


In the three seasons since Brentford climbed from England’s League One, 23 different clubs have finished in the top half of the Championship. As one of three sides along with Derby County and Cardiff City to finish in the top 12 in each of those campaigns, the Bees don’t seem far from reaching the top rung of English football’s ladder.

Behind manager Dean Smith, the Bees are looking for ways to differentiate themselves to reach the Premier League for the first time in club history. Smith explains here how his staff focuses on two key areas when preparing for an opponent: opportunities and threats.

Player Tracking and Deep Learning Innovations Lead to Greater Interactions with Data


Fans regularly consume STATS’ data without knowing the original source of the information, and that isn’t necessarily changing as the company explores future applications of richer data. What is different in 2017 from 1987, 1997 or even 2007 is STATS’ specialization from often drawing the line at providing sports information to more properly contextualizing it for clients.

One of the leaders making that happen at STATS emerged from behind the scenes to speak with Sirius XM Wharton Moneyball hosts Adi Wyner and Shane Jensen earlier this month. Patrick Lucey, STATS Director of Data Science, reflected on five years of STATS SportVU tracking data collection and provided insight on the answers now coming from growing sample sizes and industry advances.

STATS began tracking NBA games with its optical system SportVU half a decade ago, which happened to coincide with what Lucey marked as a particularly interesting time in sports science.

“Something happened in 2012. Deep learning came along,” said Lucey, co-author of the 2016 MIT Sloan Sports Analytics Conference first-place research paper on patterns of player movement and ball striking in tennis.

“It’s like a perfect storm. I like to think of it as a three-legged stool. First of all, for deep learning to work, you need three things. The first thing is you need a lot of data. Luckily at STATS, we have that. The second thing is you need the computational power. … The third thing is having the people to actually know how to work or set up the network.”

For nearly two years, Lucey has sat atop that stool at STATS and helped expand player tracking into additional sports with an emphasis on the global game of football. Plenty of his current work, including his 2017 MIT Sloan Conference runner-up paper on shooting styles in basketball, has involved finding the proper ways to ask and answer the what-if questions that weren’t confrontable in STATS’ earlier years.

“The big thing – and this is what we can do really well – we can contextualize the data,” Lucey said. “…You can ask us specific questions. You can say, ‘Well, what’s the likelihood of this player making that shot? Or, what happens if I switch that player with another player?’

“So, we can really do fine-grain simulations and ask those what-if questions. … We have the box score or we have these players’ stats, but (I think the next stage in analytics) is to simulate these kinds of specific plays and see what a different player will do in these situations. In a given situation, we can model that context and we can give more precise answers because we understand the data in those situations.”

The particulars of body pose have become one of his key interests. Specifically, advances in player tracking – and the potential for more accurate predictive analysis – lies in an increased emphasis on measurements that can go beyond using a basic center of mass to establish the XY coordinates in optical tracking.

“Once upon a time, we could capture body pose, but we had to do that in a lab setting,” Lucey said. “… Now we can actually do it in the wild.”

Lucey’s team is constantly honing in on the next big thing in its field, and his view for how performance will be evaluated more closely in the future might surprise some. It’s not necessarily about creating fresh analytical metrics. It’s about the relationship between people and technology.

“I don’t actually think new metrics is the way to go,” Lucey said. “I don’t think that’s the future. I think it’s a symbiosis between a human and a computer. So can we develop new technology to help a domain expert do their job better. I think that’s really the next step in sports analytics. Just enabling, creating this kind of technology just to help coaches or analysts or people at home to be able to ask these what-if questions. I don’t think we’re that far off.”