"Are our heroic goal-scorers merely products of exceptional runs of luck?"

courtesy of SNS

courtesy of SNS

written by: @TheGersReport

Back in 2014, Devin Pleuler, now the head of analytics for defending MLS Champions' Toronto FC, wrote a post that was a reality check that breaks the hearts of both the young & old.  He wrote:

The latest target of the soccer analytics community is the repeatability of goal scoring. Suspicion has arose that a player’s shot conversion rate does not correlate year-on-year. This suggests that players who are exceptional goal-scorers one year are no more likely to be exceptional the following year.
If true, this shocking discovery would shake our understanding of strikers to the core. Are our heroic goal-scorers merely products of exceptional runs of luck? The implications of this research are transcending; cascading from the youngest fans to the grizzliest veterans embedded deep in club offices.
— Devin Pleuler, Opta Pro (2014)

The fact that "great finishers" may not actually be exceptional at the skill of finishing & might actually be at the mercy of 'exceptional runs of luck' could be blasphemy & probably only partly true.  Whenever you are dealing with professional athletes, there will be players who are elite at certain skill-sets & finishing has to be one of those skill-sets...right?

The problem is that people put labels on players, like "poor finisher", "great finisher", "absolutely excellent finisher", without basing it on much beyond their most recent (or famous) memory of the player.  For example, do a Twitter search of "Morelos finisher" or "Lafferty finisher" & you get some real gems.

Last year, he scored on 21% of his unblocked shots & 47% of his Shots on Target - both of those would put him among the top finishers in the league over the past four seasons....so definitley not the "worst finisher"?

Last year, he scored on 21% of his unblocked shots & 47% of his Shots on Target - both of those would put him among the top finishers in the league over the past four seasons....so definitley not the "worst finisher"?

Since 2014, 33% of his shots have actually been on target & he beats the keeper 13% of the time. That's basically equal to Stevie May's output from last year (who averaged 0.22 goals per 90 last season...which is bad).

Since 2014, 33% of his shots have actually been on target & he beats the keeper 13% of the time. That's basically equal to Stevie May's output from last year (who averaged 0.22 goals per 90 last season...which is bad).

Now...I do believe that some players are better finishers than others & football is not just a crapshoot of luck.  It's just that I'm not sure how many "great finishers" actually play in the Scottish Premiership.  

Since the beginning of the 2015-16 season, only six forwards have had a Goals Above Average (GAA) of +0.12 in multiple seasons.  What's Goals Above Average?  For a more in-depth description, check out this post from last summer in which I used the stat to highlight why Alfredo Morelos was a bomb-ass signing by Rangers.

Basically, Goals Above Average measures how much better a forward's goal scoring output has been than the average Scottish Premiership forward.  Since 2014-15, the average forward who has played at least 400 minutes & scored at least one goal averages 0.325 goals per 90 minutes.  That's the baseline for average.  Teams competing for a top three spot in the league probably don't want just an "average" forward, while teams in the bottom three would be pretty damn happy to have an average goal scorer in their squad.  Last season, Kyle Lafferty's Goals Above Average was +0.04.  He was extremely average, while Alfredo Morelos' GAA was +0.17. 

How good is that? 

Below you'll find a visual showing the volume of players at each of the GAA rates since 2014-15.

Screen Shot 2018-08-21 at 7.03.01 PM.png

You can see that Lafferty's output last season put him right there in the biggest group of forwards (hence the label average), while Morelos was in that pretty damn good, but not necessarily elite group.

But, you can see why I used a Goals Above Average of +0.12 as a benchmark.  If you are scoring at that rate or higher, you've entered the realm of being a very good goal scorer (errr...finisher) at this level.

Again, in the past three seasons only six forwards have hit that mark in multiple seasons.  Does that mean the league has only seen six good, consistent finishers in the past three seasons?  

Before I answer that...it's time for the big reveal.  Who are these mystery six forwards who have actually earned the title of being "absolutely excellent finishers?"

One of the six, Louis Moult

One of the six, Louis Moult

Of those six, there are some names you'd expect: Leigh Griffiths, Louis Moult, Moussa Dembélé, & Kris Boyd....along with a blast from the past in Liam Boyce....and......

Kris Doolan.


That's it...they are the only six forwards who were "good enough finishers" to score at a level which you'd consider 'very good' at least two of the last three years.  

Screen Shot 2018-08-21 at 8.24.48 PM.png

To add more context, I included their output going back to the 2013-14 season in the above visual.  These are the six great finishers who have graced the Scottish Premiership with their talents.

Let's follow each of their wavy lines...look there's Kris Boyd who was a great finisher in 2013-14 & then again last season....but was actually a BELOW AVERAGE goal scorer for three straight seasons (including one season being a very below average Championship forward).  

Screen Shot 2018-08-23 at 11.49.51 PM.png

Dembélé had an elite season in 2016-17 & then in a down season was still actually very good with a GAA of +0.16.  Liam Boyce also was scoring at an elite level in two of his three seasons in the Premiership & flipped that into a £500,000 transfer down to the English Championship.  The move didn't work out in year one, as his Goals Above Average at that level was -0.10.  More on that later....

Kris Doolan was probably the most consistent finisher in the league (along with Griffiths) in recent memory. 

Who would have thought? 

It makes you wonder if age is catching up to the 31-year old Doolan as you see his form take a drastic dip last season as his GAA was a -0.12.

Louis Moult is the only forward in this group whose goal scoring record consistently improved, which obviously triggered his move down south.  Moult didn't get much first team action for Preston North End (less than 400 minutes) so it's hard to make say much about the fact his Goal Above Average for the English Championship was +0.27.  That rate would be beyond elite for that level BUT, again it's less than 400 minutes...so who knows?

Leigh Griffiths clearly is the standard for an elite goal scorer in Scotland & nobody has really come close to his level of consistent output.  It'll be interesting to see what his underlying stats show us about his finishing ability.

So, how do we define finishing?  It obviously comes down to scoring goals, duh.  But which player is a better finisher - the guy who scores 10 goals on 80 shots, or the one who scored eight goals on 40 shots?  What if you're someone like Kyle Lafferty whose shots have been off target 67% of the time since 2014?  Can he be labeled a great finisher if the majority of his shots are saved by the supporters in the stand?

Let’s start with looking at the Shot Accuracy for the forwards who have proven to be consistent finishers in the Premiership, our aforementioned six.  Shot Accuracy is simply the rate of total shots taken that are on target.

Screen Shot 2018-08-21 at 9.15.32 PM.png

That line running across the chart is not a glitch.  I went back & took the top 15 goals scorers (based on goals per 90) over each of the past three seasons to see what 'good finishing' looks like.  This collection of 45 players combined for a total of nearly 85,000 minutes & 2,500 shots.  These were the best finishers at this level & averaged 0.51 goals per 90 minutes as a group.

For the purposes of looking at Shot Accuracy, 48% is the benchmark.  It's hard to label someone a "great finisher" if they can't find the target at this kind of rate, right?

Ummm....Kris Boyd?  Leigh Griffiths???

Doolan & Boyce were the only players who outpaced the 48% mark in multiple seasons, while Moult & Dembele consistently put shots on target near that 48% rate.

Honestly, getting the ball on target is only the first part...actually beating the keeper is the key to determining who a great finisher is.  Below is a visualization of Shooting Percentages (Sh%) from the past three season (goals divided by shots on target).

Screen Shot 2018-08-21 at 9.25.15 PM.png

Even this collection of the league's most consistent goal scorers....weren't actually  consistent finishers.  The benchmark Sh% was 0.412.  All summer I've been preaching that if a player's Sh% is 40% or above, expect a regression in finishing next season.  

You can see that Kris Boyd & Louis Moult are the only players in this group that managed to put together back-to-back 40% seasons (Moult actually hit that mark in three straight seasons). Everyone else hit that threshold at least once & than saw the expected  regression.  

What about Liam Boyce?  He had an ungodly Sh% of 0.636 in 2016-17, guess what happened next?  Of his shots on target for Burton Albion, only 33% beat the keeper.  That's the same rate he scored on back in 2015-16.  Consequently, his goals per 90 dropped down to 0.25.

Ultimately, when you combine Shot Accuracy & Shooting Percentage - you want to see which players score on the largest rate of their overall shots.  If a player scores on 40% of their shots on target, but only get the ball on target 35% of the time...are they really that great of a finisher?

Below are the Conversion Rates for this group of forwards (goals divided by shots).

Screen Shot 2018-08-21 at 9.34.12 PM.png

The benchmark Conversion Rate is 20%.  There's a whole lot of interesting here.  

  • You wonder if anyone at Burton Albion understood the concept of regression when signing Boyce? I mean, his Conversion Rate was 37%...nearly double the standard of a very good goal scorer in the Scottish Premiership. Well...guess what happened once he moved to England. His Conversion Rate somehow plummeted to 10%. Shocking.

  • Leigh Griffiths' goals per 90 & his Conversion Rates have been in a steady decline for three seasons now. Despite that his goal rates are still better than anyone on this list, despite kind of meh Conversion Rates. Why is that? Shot volume...if there's one common theme of being a Celtic forward...it's that you get a shit ton of shots.

Screen Shot 2018-08-25 at 11.11.19 PM.png

I think there's a little case of "bar envy" for the non-Celtic forwards here.  When you looked at the previous visuals, there really wasn't much consistency from Griffiths outside of the fact he always scores at a better rate than literally everyone else.  His accuracy fluctuates, his finishing goes up & down, & his actual Conversion Rates have been in a steady decline...but he always gets his shots.  He more than doubles the number of shots per 90 than his peers do for the most part.  

So, is Griffiths just a byproduct of a dominant system?  Moussa Dembele & Odsonne Edouard have certainly thrived playing for Celtic (Edouard led the league in goals per 90 last season).  Griffiths is the best of the trio at generating shots, but both Dembele & now Edouard are getting 3+ shots per 90 minutes.  Are they great finishers??

Dembele's underlying numbers from two seasons ago showcase a very good finisher, but then last season...not so much.  But, his Goals Above Average was still +0.16 which is nearly elite.

Now, we all know that all shots are not created equal...so let's slow our roll a little bit.  Below is the breakdown of Expected Goals per shot for each of the forwards in this group.  Basically, who is consistently better at getting on the end of quality chances.

Screen Shot 2018-08-22 at 7.25.00 AM.png

Go back to Griffiths.  As his goal rate has decreased since 2015-16, his ability to get better chances actually peaked last season.  But, his Conversion Rate was the worst it's been in that same time period.  He's an elite goal scorer...but is he an elite finisher?  Or, does he feed off the fact he plays on a team in which he can average nearly five shots per game?

courtesy of SNS

courtesy of SNS

Of the six forwards in this group, the only one whose xG per shot was above that threshold line of being really good each season  was Kris Doolan.  If you scroll back to the actual Conversion Rates visual, you'll see again it's just Doolan above that line of being awesome each season.

I stumbled upon an article by Barney Ronay from back in 2012 during some research for this post.  In it he was bemoaning how the English forward seemingly had been left behind in the evolution of the modern striker.  Some of his prose seemed to be fitting when thinking of Kris Doolan.

The finisher has always been there in English football. A master of the dink, the poke, the clip, he has often been something of an anti-athlete, a languid roué prowling his kill zone. Beyond this the finisher’s art has simply suited the style of the game here, a place where football is often a scuffling, formless entity given shape by the finisher’s ability to pluck a moment of clarity out of bobbling confusion.

At which point I started to wonder if finishing really exists at all as a separate thing, or if football’s hunger for cod-science has perhaps gilded these virtues with an unwarranted mystique. Certainly, the finishing school has yet to materialise, its travelling caravans yet to roll in through streets thronged with tearful children, all previously stuck on the bit about not kicking it in the middle of the goal. The game has changed fundamentally and finishing seems just a little lost as an independent discipline. Attempting to carve out a career as a finisher would be like turning up at a job centre interview and describing yourself on the form as a bounty hunter, or a high plains drifter, or a man with a past. It just isn’t going to fly.
— Barney Ronay, The Guardian (March 2012)

High-plans drifter:  Kris Doolan. 

He's a poacher & you know what...it's worked.  The problem becomes...being a poacher on a team that doesn't give you anything to catch ahold of.

Ironically, I brought up this dilemma in March of last season.

Screen Shot 2018-08-25 at 11.48.36 PM.png

Playing for relegation bound Partick Thistle last season, a team who truly struggled to create chances, meant that the volume of shots just wasn't there for Doolan.  His finishing stats didn't bottom out like his -0.12 Goals Above Average would suggest.  His Shot Accuracy had declined but it was still nearly at the same level of the best goal scorers in recent seasons & his Shooting Percentage was actually the highest it's been in years.  But, if you scroll back to that visual of Shots per 90 you see the sad figure that was Kris Doolan's shot rate last season.  By the end of the season, he was only averaging 0.90 shots per 90 minutes.  They were good shots...& his Conversion Rate was still pretty spectacular (25%) but the rest of his team...kind of sucked. 

So...are any of these six forwards truly "great finishers"?  Rather than simply rely on our gut feeling:

Actually...Morelos' Conversion Rate: 16% Lafferty's: 14% Not really sure what they keep in their lockers?

Actually...Morelos' Conversion Rate: 16% Lafferty's: 14% Not really sure what they keep in their lockers?

Let's rely on facts:

  • Kris Boyd: everything came together for him last season but in the three previous seasons he was actually a kinda bad finisher. So, was last season lucky? Or did he spend three seasons simply pretending to be a bad finisher?

  • Kris Doolan: a pretty damn good finisher in "the bounty hunter, high plans drifter" sense of the word (That was some serious prose form Barney Ronay).

Screen Shot 2018-08-26 at 12.16.03 AM.png
  • Leigh Griffiths: this may be controversial --- but....not that great of a finisher. Gets tons of shots, playing for a team that lets you get tons of shots. If you look at the numbers, those shots turn into goals not because of his clinical finishing rather because he gets.....lots of chances to score.

  • Louis Moult: Of the six, he & Dembele would be the most appealing target for a team looking to improve their quality up front. His goal production was steadily increasing before he left for Preston North End. His Shot Accuracy was remarkably consistent & his Shooting Percentage always sustained near that 40% mark - which is even more impressive.

  • Given we only have two years of shot data for Boyce & Dembele in the Scottish Premiership, it's tougher to make conclusions about their finishing ability. Boyce was riding unsustainable rates, as evidenced by the bottom falling out once he went south of the border where he produced as a below than average English Championship forward (GAA was -0.05).

  • Dembele is primed for a bounce-back season. Even though he made this group, the perception is that he had a down season. A big part of that was that he was cursed by the inevitable decline in Shooting Percentage once he passed that 40% mark the season before. Last season, it dropped to 30%. However, his Projected Goals of 0.54 suggests he is poised to return to elite scoring levels again.

Now this post wasn't about evaluating six forwards & their finishing ability, it was written to show that even the very best goal scorers in recent seasons aren't consistent finishers.  Louis Moult & Kris Doolan are probably the only ones you factually suggest can claim that label.

How about other forwards who have experienced success at this level.  This next group of five forwards have played multiple seasons in the Scottish Premiership & have crossed that threshold of a +0.12 Goals Above Average once in the past three seasons.  

Let's look at their group of graphs...but with minimal commentary.  Let's just look to see when & where we see consistent results.

*Simon Murray's numbers from 2016-17 are actually from the Scottish Championship.

Screen Shot 2018-08-21 at 8.45.15 PM.png

There is less variance in this group of forwards, outside of Steve MacLean's woeful 2014-15 season.  For the most part they have hovered around being an average forward for this level (Goals Above Average hovering around 0.00).  Rooney, Murray, Curran & MacLean all peaked three seasons ago & struggled to produce beyond that of an average striker since.

Screen Shot 2018-08-22 at 8.51.14 PM.png
Screen Shot 2018-08-26 at 10.22.59 AM.png
Screen Shot 2018-08-22 at 8.58.26 PM.png

Kris Doolan...say hello to Steven MacLean.

Screen Shot 2018-08-22 at 9.07.14 PM.png

Shot volume rears its ugly head guiding even the decent finishers to mediocrity.

Screen Shot 2018-08-22 at 9.02.48 PM.png

These two groups represent 11 of the 16 forwards who have bypassed the +0.12 Goals Above Average mark, & the most noticeable trend when you look at some of the underlying numbers is that most of them aren't consistent finishers.  

Two of the most consistent are Steven MacLean & Kris Doolan - whom you could probably classify as poachers.  Of their six combined seasons, half of them have seen their goal scoring rates be very good.  Part of that, as I mentioned with Doolan, is that the one-dimensional play of being a poacher.  You rely much more on the quality of your teammates to create those situations in which poachers thrive.  They both play for below average teams & you can see that in their mediocre shots per 90 rates.

The lack of consistency in Shooting Percentages & Conversion Rates for all of these stats brings up the idea of looking for repeatability when analyzing stats.  What is the correlation between shooting at a certain rate in one season to getting similar results the next season. 

I took all of these categories & applied them to each & every player who had 20+ shots in both the 2015-16 season & the 2016-17 season.  The correlation between one season to another is measured by R squared.  I'm not going to get into much detail about this measure, but let's just say the closer to 100% - the more likely an output in one season, was to be repeated by a player in the subsequent season.

Only non-penalty goals are included.

Correlation is measured using R Squared

Correlation is measured using R Squared

Note:  when the same process was run for every player who played in both seasons the goals per 90 correlation was 0.47 & the xG per 90 stayed the same (0.63).

  • There is literally no correlation between a player's Shooting Percentage from year-to-year. That's why when a player has a really good Sh% (like 40% & above)...beware of regression. It's a comin'

  • Couple that with very little correlation between a player's Shot Accuracy from season-to-season & you can see why a player's Conversion Rates are also going to be inconsistent.

  • Maybe "good finishing" really does come down to "exceptional runs of luck"

  • So what drives consistent goal scoring? It's not so much finishing...it's volume & shot quality. Players shots per 90 & xG rates tend to be pretty consistent from season to season. We saw that, to an extent, in the groups of forwards profiled.

There are five other forwards who have bypassed we haven't mentioned yet.  They all passed the GAA benchmark of +0.12 in their first seasons of at least 400 minutes in the Scottish Premiership.  Here is how they ranked amongst other forwards in Shots per 90 & xG per 90.

I also added Kyle Lafferty, to stay topical.  His Goals Above Average last season was only +0.04...so very average.

Screen Shot 2018-08-26 at 4.59.00 PM.png

You can see that the five on this list who had that high GAA all had a pretty good mix of shot volume & shot quality.  Any real analysis to make projections about future production would obviously have to dig deeper than this...but it's a starting point.

You can also see why Lafferty's goal rate was very average last season.  He takes a good amount of shots, but for the vast majority of them are low percentage chances.  There's a pretty big difference in shot quality between him & the others on this list.  There was also a pretty noticeable difference in actual results.

Ultimately, we have seen that there are few rare examples of consistent finishing at this level & any recruitment of forwards from the Scottish Premiership really needs to begin with shot volume, followed by shot quality.  Obviously you would be looking at the underlying stats that measure finishing ability, but it's with the lens that they are very much subject to change (hello, Liam Boyce).

It wasn't that long ago, that "the best finisher" in the league was this guy...

Some notes...

  • Obviously, Kyle Lafferty has hit the ground running with Rangers. I still believe the data will come true & his production will be that of an average-to below average forward. This is also a chance to remind folks that Michael O'Halloran was well on his way to winning Player of the Year at this time last season.

  • The supporter in me is happy to see Lafferty back. The rational side of me suggests that this was lazy recruitment....especially when compared to the fact Hibs spent £100,000 to bring in Florian Kamberi - a player who is EIGHT years younger.

  • Ironically, last summer I wrote about how Hearts signing of Lafferty was a wise investment. What's changed for me...last season's underlying numbers are a giant red flag.

  • Stats from 2015-16 & 2016-17 are courtesy of the great @TheBackpassRule

  • This was written under the influence of Spacehorse, The Creation, Traffic & Drahla.