clock menu more-arrow no yes

Filed under:

Skill vs. Luck and just where Leeds are going wrong.

New, comments

Soccermetrics is like taking a magnifying glass to look at small things in more detail. Long have pundits talked about football being a game of 'skill' and 'luck'; proponents of soccermetrics put forward measures for these. Skill becomes seen in terms of 'shot situations' and luck in terms of 'shot and save percentages'. Welcome to the world of TSR and PDO.

You need to be in the right positions to earn the luck
You need to be in the right positions to earn the luck
Clint Hughes/Getty Images

"'Yes Roger Hunt misses a few, but he gets in the right place to miss them.'"

Bill Shankly.

What's Soccermetrics and why do we care?

Have you ever seen the baseball film 'Moneyball' where the manager of the Oakland As hires a college graduate who is a stats nerd? You should, that basically explains two things: a.) what 'soccermetrics' is and also b,) who I am. I am not in the film, although I once told Leslie Joseph "I don't care who you are, we queue behind the person in front of us" in Whittards in some place in that there London. In my defence she was rude and I was hungover. I digress, it basically sums up what I like about stats. Moneyball that is, not my rudeness to Leslie Joseph.

Don't get me wrong, I don't like numbers; that's for Maths teachers and Physicists. I do like how they can show trends, pick out things that normally might be missed, can be used as 'number pictures' to pick out things and illustrate ideas. Basically, that's how I see 'soccermetrics' - the measurement of aspects of football that allow us fans to gain a better insight into how a particular player plays. For example (no stats I promise), even though Mirco Antenucci has had more shots than Souleymane Doukara, has scored more goals and has a better shot accuracy that Doukara has a much more impressive conversion rate for his 5 goals when you only consider the number of shots on target.

Soccermetrics - can they inform the fans?

'Soccermetrics' can be boiled down to measuring aspects of the game and putting them into a quantifiable and countable format. For example, passing is an important part of the game as 'no passes' obviously equals 'no goals' but soccermetrics applies a little more focus, brings a little colour if you like. The explosion of interest around statistics in football has roots in the past and also with baseball but I'll not delve too much into its development. What I will do is focus on just two main areas of interest that are in use at the moment and that help to illustrate two different areas of football; these being what are termed TSR and PDO.

TSR or 'Total Shots Ratio' is basically a posh term for a very simple metric (measurement) in football. What it does is that it looks at the number of shots a team takes as a ratio of the total shots taken; here's the Maths bit if you're interested (TSR = team shots/team shots + opponent shots). Proponents of this metric point out a few things about it that are useful to put forward here, but in brief. Teams with a high TSR tend to be teams that traditionally do well at the end of a season as they are fashioning a high number of shooting opportunities per game (p/game) whilst conversely denying opponents shooting opportunities. As such, TSR is often seen as a metric based on skill; the skill of a team to create for themselves as well as deny shooting opportunities for their opponents.

PDO is a metric borrowed from Ice Hockey and is a way of introducing an element of luck into proceedings. As the great Bill Shankly says at the head of this article, Roger Hunt misses goals (bad luck) but did get into a position to take these shots (good skill). PDO looks at the combined elements of 'score percentage' (goals scored/shots on target) and 'save percentage' (goals conceded/opponent shots on target. The luck element works very easily in this simple metric. Not every shot a team has on target leads to a goal, there have to be some elements of luck in place such as bad shot placement too close to the keeper and a parry or defenders making goal line clearances. Equally, not every shot conceded leads to a goal as a keeper may flick out a despairing hand and deflect the ball onto a post and the rebound cannons off for a corner. When added together these two form a percentage where the mean PDO rating is 1.000 or 100%. Teams playing above the 'luck mean' have higher than 1.000/100% and those playing below the 'luck mean' have lower than 1.000/100%. The thing is luck, due to its very nature, cannot last forever and over-performing teams regress/lower closer to the mean and those under-performing gradually break that string of bad luck and climb closer to the 1.000/100% level.


The case at Leeds, why are we where we are?

TSR - Total Shots Ratio

TSR is usually higher in successful teams, often as a result of their superior skill in attacking and fashioning shooting opportunities and superior, more organised defence limiting the opposition. Less successful teams, conversely suffer in two ways: they have either a much lower TSR rating or a higher TSR but with low goal production (something I will touch on later). After roughly halfway in the Championship season, how do the TSR ratings stack up dependent on position in the Championship table?

What this table does is illustrate general trends such as teams with a high TSR rating (if over .500 they have more than 50% of match total shots) tend to be, although not always, placed higher in the table. To suggest it is all about TSR, to turn TSR into some kind of StatGod is too simplistic. For example, Leeds' Boxing Day opponents Wigan have a TSR rating of .522 yet are in 23rd place in the Championship; Brighton (22nd) and Millwall (20th) similarly have TSRs of .539 and .522 respectively. All three of these teams have higher TSRs than 2nd-placed Ipswich (.497) and 3rd-placed Brentford (.469) but other factor must be coming in to play for higher TSR teams to be placed lowly in the league table. A quick glance at each of these team's 'goals for' rating gives one insight; all of them are low scoring teams (Wigan 22 goals, Brighton 22 goals and Millwall 23 goals) and all have sizeable, negative 'goal differences' (Wigan -8, Brighton -8 and Millwall -9).

Leeds are an enigma, packaged in a mystery and wrapped up in an enigma; low TSR rating (.426; worse than every team below them bar Blackpool). The three relegated teams from last season Doncaster (.420), Barnsley (.460) and Yeovil (.410) were relegated with TSR ratings around Leeds' current TSR of .426 - suddenly things don't look too rosey. Things begin to look even less rosier if you look at 'rolling TSR' (basically cumulative TSR week-on-week). If Leeds continue with the same output level as they are now and let opponents shoot with the same frequency as they do then Leeds projected end-of-season TSR is likely to be .406; that's relegation TSR in anyone's language.

Leeds vs Opponent ROlling TSR

Projected TSR based on current performance - totals based on 22 games

Quite simply what can Leeds do? It's no use leaving it in the hands of those below us, the fact that they are worse than us. What happens if a couple of those teams put together a run of games with positive results and high game TSRs? It's pretty simple from a TSR point what Leeds have to do; they have to dominate the shot count more, work players into positions whereby they can shoot. Look, it's not a complicated formula but: accurate passes lead to increased possession which generally leads to better opportunities to establish good field position which in turn tend to generate more shooting opportunities which usually allows more shots on target where you will likely have a better chance of scoring more goals than the opposition...usually. There will always be that game were a team dominates for the whole match with 20 shots at goal and then the opponent snakes on in in the 92nd minute via a bobble shot hitting a divot in the ground. That's basically the scenario against Bolton in game 5 of the season: Leeds 8 shots, 2 on target, 1 goal; Bolton 22 shots, 8 on target, 0 goals. Marco Silvestri made SEVEN saves that game but he's never likely to have too many of those games in a season.

PDO - when skill relies on luck

At the simplest level possible PDO is simply a measure of a team's shooting ability and save capacity. Since every shot on target must result in a save or a goal then the mean is 1 (1.000). Over the course of a game a team can score one goal from one 'on-target' shot and be at a shooting ratio of 1.000 (no save made) and have four goals scored against it but make two saves of 'on target' shots for a save ratio of 0.667 ( 4 goals from 6 shots on target) and end up with a PDO of 1.667 which means that, basically, in that game despite getting hammered 4-1, they did have some luck: they scored off their only shot on target and saved 2 shots. That was basically our game against Watford; we were dire as a performance but we were lucky in some aspects.

Now the thing is that such outstanding luck is simply not sustainable, even the mighty stumble after a while and over a period of games the PDO ('luck') metric slowly comes down or regresses to closer to the magic 1.000 over time. Just as Marco Silvestri cannot maintain the 7 saves lucky streak against Bolton for a whole season. In fact, if you look at the 8 games starting with the Bolton match and ending with the Rotherham match (3 wins, 3 draws, 2 losses) then Silvestri made an outstanding 36 saves at a per match average of 4.50 saves. Take the next block of 8 games from the Norwich match to the Ipswich match and that luck hasn't as much deserted him as ran across the desert with no clothes on. In this series of games (2 wins, 2 draws, 4 losses), Marco Silvestri's 'luck' is only good enough for 12 saves at a per game average of a measly 1.50.

Just as one individual cannot cannot maintain his lucky streak over a 46 game season; neither can a team. A team performing badly, such as Blackpool, will likely catch a break at some point and put together better performances. If this coincides with bad luck, typified by PDO, for those teams around them, well that's when that 'pundit-friendly' term "sucked into a relegation battle" gets brought out.

Leeds are on fringes of that at this moment and only 5 points above Brighton who are in 22nd place in the League table. A good run of luck  from them, coupled with a similar burst from Rotherham and Huddersfield  (same points as us with 24) and Millwall only 1 point further back on 23 and it could be squeaky keyhole time up around LS11.

I'm not going to jump on the bandwagon demanding "strong centreback A from club B" or "tricky winger Y from club Z" like the clamouring masses on Facebook seem to be demanding; neither am I going to say "it takes time to bed in a team blah blah blah". We needs to be more consistent in shooting, more consistent in getting into shooting positions and more consistent in stopping the opposition doing the same.

A s the great BIll Shankley said at the head of this article; you need to get players in positions in positions (TSR skill) and allow them to shoot. Yes they miss some (bad PDO/luck) but they also score some (good PDO/luck); remember the keeper can't keep 'em all out.

Special thank you to Owain Thomas at 'The Only Statistic That Matters' blog for the use of his Championship 2013/14 statistics