Which Playmakers Actually Make Their Teammates Better? (The 2018-19 Update)

courtesy of SNS Group

courtesy of SNS Group

written by: @TheGersReport

Back in February, I presented an experiment with data. The goal was to look at the numbers to see which playmakers made their teammates better...by putting the shooter in the best situation to succeed. How could we use analytics to identify the players who have the vision, timing, & technique that literally sets up optimal goal scoring opportunities for their teammates.

It was an experiment. Rather than rely on stats that are my typical starting points for playmakers (Expected Assists & Scoring Chance Key Passes) I wanted to see if we could determine which players gave teammates the space, time & appropriate delivery of the pass to create the best conditions to get a shot on target. I went on to explain why narrowing the study to assists would be foolish & you can read the rant here if you so desire. Let’s just say if you use assists as the barometer…Richard Foster is just as good of a playmaker as Kieran Tierney & Ryan Kent.

So…I looked at players who had the most Key Passes in the league & used the results of the shots they set-up to make some assumptions. If a passer consistently created the conditions in which his teammate can take a shot that tested the keeper, does that mean they are giving their teammate a better chance to succeed….hence making them better?

In 2017-18, the playmakers whose teammates had the highest Shot Accuracies from the passes that they set-up were some names that were expected (Tierney, Olivier Ntcham, Jordan Jones, James Tavernier, Scott Allan, Scott Sinclair, James Forrest, etc), along with a few that weren’t (Richard Tait, Glen Kamara, Blair Spittal).

Now that we are in the midst of the summer, I decided to go back & run the same numbers again to look back at results from last season.

courtesy of SNS

courtesy of SNS

Like any set of statistics, most of the learning comes from the outliers in that data set. Now, in order to not rely on super small sample sizes, I narrowed the study to players who had at least 30 Key Passes (passes that set up shots) & sprinkled in a few players who had good per 90 rates in lesser minutes…basically Martin Boyle & Connor McLennan). I also made sure to include players who made the list last season to see how repeatable their results were.

Below you’ll find the Shot Accuracy for the shots that these playmakers set-up. The league average of shots being on target last season was 48%, so that is marked on the visual below. It should be noted that the Shot Accuracy for shots that had a Key Pass back in 2017-18 was…..48% (I know…weird, right).

Note: throughout this post, I will only referring to data on unblocked shots. The original starting point was 30 Key Passes, but that dwindled for some as many of the shots they set-up were blocked. For example, 47% of the shots that Daniel Candeias set-up were blocked.

Screen Shot 2019-07-22 at 12.01.19 AM.png

Before we go on, we must breakdown Scott Arfield’s results as a playmaker that were kind of beyond sensational. He had 21 Key Passes in open play & 18 of them led to a shot on target for a Shot Accuracy of 86%. That’s not an accident. Whenever Arfield set-up a teammate, they had the time, space & angle to force the keeper into a save. That led to a Conversion Rate of 28.6% on shots that Arfield set-up (typically the rate from open play is 13.5%).

Odsonne Edouard’s results were similar & so were Martin Boyle’s…whose season was cut short in 2018-19 by an knee injury while playing for Australia’s national team.


At this point I want to compare these results to a new number I’ve been playing around with. The stat is based on another passing rating I’ve published in the past.

The purpose of this new Key Pass Impact Rating is to estimate how much more (or less) likely a playmaker’s Key Passes are to set-up goals. For example, based on the kinds of shots that Martin Boyle set-up via his Key Passes…his teammates were 9% more likely to score a goal (which helps explain his high assist rate…he had four assists from only 18 Key Passes).

I’ll explain how this rating is determined at the end of the post…for now, here’s the Key Pass Impact Ratings for the playmakers we’ve focused on.

Screen Shot 2019-07-22 at 8.50.26 AM.png

Along with Boyle, former Rangers winger Daniel Candeias & Celtic’s Ryan Christie are your main outliers. When Candieas set-up a shot last season it was 5% more likely to score than the average shot, while Christie’s rating was +0.03.

On the other end you have players like Richard Tait, Stevie Mallan, Scott Pittman & Ryan Kent. The way to look at this end of the table is that when Stevie Mallan set-up a shot, teammates were 4% less likely to score then they normally would be.

I plan on writing another blog post looking at this stat more but for now…let’s see how it correlates to the Key Pass Shot Accuracy stats we started with.

note: some player’s names removed to make this more readable

note: some player’s names removed to make this more readable

There are three distinct clusters of playmakers here. The first group of players have positive Key Pass Impact Ratings & above average Key Pass Shot Accuracies. This includes Boyle, Candeias, Christie, Horgan, Tavernier, James Forrest & Connor McLennan. This group consistently set-up teammates with dangerous chances & the Shot Accuracy rates suggest that teammates were in good situations to test the keeper.

This group set up 46 goals for a Conversion Rate of 22.5%.

Note from the graph above that none of the players who had a positive Key Pass Impact Rating saw the Shot Accuracy from those passes fall below the league average rate of 48%.


The second group of playmakers had above average Key Pass Shot Accuracy rates despite negative Key Pass Impact Ratings. This suggests that despite setting up lower percentage shots, the teammates they passed to tended to have time & space to put their shots on target. This group includes Tierney, McGregor, Edouard, Shinnie, Ntcham, David Wotherspoon, Ryan Kent & Richard Tait.

This group set up 27 goals for a Conversion Rate of 13.7%.


Our last group of playmakers had negative Key Pass Impact Ratings & below average Key Pass Shot Accuracy percentages. Teammates were less likely to score when their shot was set-up by these players & complicating this was the fact that the numbers suggest they may not have the time & space to test the keeper with their chance. This group includes Niall McGinn, Matthew Kennedy, Greg Stewart, David Turnbull, Olly Lee, Scott Pittman, & Stevie Mallan.

This group set up 34 goals for a Conversion Rate of 14.1%.

Yes, somehow…someway the results for Group Three were noticeably better than Group Two.

zo4EMKC.gif

When the results of that third group came together, I almost hit delete on this whole post. It made no sense that players should have a similar Conversion Rate to the second group if my theory that the Key Pass Shot Accuracy rates were showing us which playmakers gave teammates a better chance of testing the keeper.

But then I noticed that the third group included set-piece specialists like McGinn, Stewart, Turnbull, Lee & Mallan.

courtesy of Scott Baxter

courtesy of Scott Baxter

While I do factor in set-pieces into the Key Pass Impact Rating, I wondered if the results from that last group truly was skewed because of the number of set-piece specialists?

Below you’ll see the average number of Set Piece Key Passes per player in each of the three groups of playmakers we’ve identified.

Screen Shot 2019-07-22 at 9.53.14 AM.png

Ummmm….yeah. We have discovered a new entry point.

tenor.gif

Let’s look at the detailed breakdown of each group’s numbers, starting with Group One. Again, only Key Passes that set-up unblocked shots are included. Key Pass Impact Ratings include all shots that the player sets-up.

Screen Shot 2019-07-22 at 3.50.46 PM.png

You can see that this group was clearly driven by their playmaking in open play. The league average Conversion Rate on shots set-up in open play is 13.5%, for this collection of players it was 23.8%. The fact that all of their Key Pass Impact Ratings are positive, coupled with above average Key Passing Shot Accuracy rates suggests that this group truly made their teammates better.

The ‘Tweeners

181822729.jpg.gallery.jpg
Screen Shot 2019-07-22 at 4.04.30 PM.png

The Key Pass Impact Ratings for both Scott Arfield & Alfredo Morelos suggest that teammates should be converting at or around the league average Conversion Rate of 13.5% of the shots that they set-up. Instead, Rangers teammates scored on 23.8% of the shots set-up by the duo. Is this attributed to random luck, or the fact that teammates were put in a better position to test the keeper (remember Arfield was 1st in Key Pass Shot Accuracy & Morelos was 6th)?

Group Two

Screen Shot 2019-07-22 at 4.24.49 PM.png

Remember this group is one that featured players who all had a negative Key Pass Impact Rating while also having above average Key Pass Shot Accuracy rates. Like the first group, most of their playmaking comes from open play. The combined Conversion Rate on shots that they set-up in open play was 13.6%…so basically aligned with the league average rate of 13.5%. You could possibly argue that the fact they set teammates up with chances in which they could better test the keeper helped drag up the scoring rate, despite the low Key Pass Impact Ratings.

The Other ‘Tweener

courtesy of RFC

courtesy of RFC

Screen Shot 2019-07-22 at 4.48.56 PM.png

Borna Barišić had an overall Key Pass Impact Rating of +0.003, which suggests that chances he created should have been been pretty close to that league average Conversion Rate of 13.5%. The actual Conversion Rate via Barišić Key Passes was 12.5%. So yeah…

One observation, both of his Key Pass Impact Ratings for set pieces were in the negative. This is based on a pretty small sample size, but I’m wondering if this more to-do with where Barišić is being asked to place the ball on his crosses? The eye-test suggest he’s one of the best crossers of the ball in the league…so, is this a byproduct of Rangers approach to set pieces?

How about Rangers other main set-piece takers? On corners James Tavernier’s rating was -0.02, but on free kicks it was +0.01. Andy Halliday actually had the best Key Pass Impact Rating on free kicks for Rangers with a +0.02. A quick video review by one of the club’s performance analysts could see if this is random or if Halliday was doing something different with where he placed the ball from his corner deliveries. <insert thinking emoji>

Now, for the group that started this whole cluster fuck of a tangent…

Group Three

Screen Shot 2019-07-22 at 5.14.10 PM.png

This group was perplexing due to the fact that they had negative Key Pass Impact Ratings & the shots they set-up were on target at a below average rate…but still had a slightly better than average Conversion Rate on the chances they created. The theory was that set pieces may have played a role in driving up the goal rate.

The league average Conversion Rate for shots created by corners is only 10.8% but this group has a collective rate of 14.7%. There’s a similar bump in the success rate from free kick Key Passes. The league average rate is 14.6%, while this group’s Conversion Rate is 18.5%. In both cases, the success rates are driven up by successful set-piece deliveries from the likes of Steve Mallan, Olly Lee, & to a lesser extent Niall McGinn & David Turnbull.

In open play, their Conversion Rate of 13.0% is slightly below average.

So this is a case in which, even though the overall numbers don’t suggest that they are making teammates better, when you peel back some of the layers you can see the impact set pieces can have on these numbers.

Below you’ll find the league results on the different kinds of shots set-up by Key Passes.

Screen Shot 2019-07-23 at 8.53.21 AM.png

The numbers suggest that set-plays from free kicks can give you a bit of an advantage (as seen by the highest Conversion Rate & a noticeably better Shot Accuracy than from corners). The next step to creating a competitive advantage is to find the playmakers who can create higher percentage shots from free kicks.

Below you’ll find each player who had a Key Pass from a free kick alongside their Key Pass Impact Ratings for free kicks.

Screen Shot 2019-07-23 at 9.10.19 AM.png

There two clear outliers here: James Tavernier & Olly Lee. This explains why they combined to set-up 25% of all the goals scored via Key Passes from free kicks.

When it comes to corner kicks, this yet another objective reminder that corners have a relatively low success rate. Last season in the Scottish Premiership, there were 2,204 corner kicks & only 10.5% of them led to an unblocked shot on goal.

Of those 2,204 corners, only 25 led to goals in which the corner kick taker directly assisted on the goal. That translates to a success rate of 1.1%.

Given how rare goals can be from corner kicks, let’s see if any of the regular corner kick takers last year gave their teams an advantage.

Screen Shot 2019-07-23 at 9.41.51 AM.png

The data suggests that there really isn’t any true outliers whose corner kick delivery gives their team a real advantage. The closest would probably be Kilmarnock’s Chris Burke, who actually saw a 16.6 Conversion Rate on shots he set-up via corners.

Remember that only 10.5% corner kicks directly led to an unblocked shot on goal. Let’s see how the playmakers mentioned in this post faired.

Screen Shot 2019-07-23 at 9.55.12 AM.png

The good news is that all of the players (except for McGregor & Turnbull) increased their teams odds of generating an unblocked shot directly from a corner kick.

Now let’s bring this (really long) post full circle. Only 35.8% of these corner kick shots are actually on target. Which of these players set-up shots with a better Shot Accuracy? Olly Lee (47.6%) & Borna Barišić (44.4%)

That’s it….well actually Andy Halliday as well - but he was only included as a comparable to the other Rangers corner kick takers.

Let’s recap:

  • The numbers suggest that the playmakers who truly made their teammates better with their vision, timing, & delivery from open play last season were Daniel Candeias, Ryan Christie, Daryl Horgan, James Tavernier, Martin Boyle, Connor McLennan, James Forrest & to a lesser extent: Scott Arfield & Alfredo Morelos.

  • Others helped teammates have a better chance to test keepers on low percentage shots, but the ones above consistently set-up those conditions on high percentage chances.

  • The best playmaking set-piece specialists in the league last season were Olly Lee, James Tavernier & Stevie Mallan.

Some notes…

  • Stats are courtesy of data provided by Ortec Sports.

  • For those of you who stuck with this blog post & all its tangents…I thank you. I’ve try to reign in those tangents in my writing but couldn’t help myself this time.

  • I promised to breakdown how I determine the Key Pass Impact Rating. So here you go…the idea was inspired by a Chase McCallum blog post from the now defunct site, Hockey Talk Blog. The concept was to use a player’s Expected Assist totals & compare it to the league average Conversion Rate on all shots : Expected Assists - (Key Passes x the league average Conversion Rate on all shots) / total Key Passes. Given that I now had access to data on shots set-up in open play vs. corners & free kicks, I separated out those three data points.

    • Let’s look at James Tavernier for an example.

      • Open Play: 8.57 xA - (58 Key Passes x 0.135…the league average Conversion Rate on all Open Play shots set-up by a Key Pass) = +0.740

      • Corners: 1.26 xA - (14 Key Passes x 0.108) = -0.252

      • Free Kicks: 1.67 xA - (11 Key Passes x 0.146) = +0.060

    • Tavernier’s overall MF Passing Rating after combining these results was +0.555 (which was the 14th best rate in the league).

    • His Key Pass Impact Rating was +0.01, which again translates to the idea that shots he set-up had a 1% better chance of scoring a goal.

  • There were nine playmakers who featured in both posts looking at Key Pass Shot Accuracy.

Screen Shot 2019-07-23 at 10.36.08 AM.png
  • There was very little variance in the results for Tierney, Ntcham, Tavernier, Forrest, & Candeias. The difference in rates for Ryan Christie is probably attributed to playing for two very different clubs. For the other three (Tait, Morelos, & McGregor) it would be interesting to do some video analysis of the shots they set-up over the past two seasons to see what happened.

  • This was written under the influence of Peter Tosh, Joy Division, the Who & the Rollin’ Stones.