Does Finishing Skill matter in MLS?
/By Sean Steffen (@seansteffen)
If you’ve ever played FIFA, you’ve probably noted the importance of a forward’s “finishing” rating to how often they finish their chances. That’s how it works in the video game, but is “finishing” a real life skill significant enough to make an impact in a forward’s goal scoring tally?
While I have yet to meet a data analyst who thinks that “finishing skill” is as relevant to goal scoring as most soccer fans tend to believe, there doesn’t seem to be a consensus in terms of whether “finishing” is a repeatable skill. In other words, can forwards depend on a superior ability to convert chances year to year?
With forwards like Gyasi Zardes (16 goals in 2014) and Cyle Larin (17 goals in 2015) bursting onto the scene by converting a high percentage of their chances on goal, the question within MLS is as important as ever. Are these players scoring so many goals because of some underlying finishing skill, or are their unusually finishing rates something closer to statistical noise?
Is finishing a skill of any importance within MLS?
One important tool we can use for answering such a question is to study discrepancies in expected goals (xG) data. Since the expected goals model is built around league averages of conversion, if finishing were a skill of any statistical note we would see a consistent out-performance of the model by certain shooters who are highly skilled finishers. But before we get into repeatability for individuals, I’d like to use goals minus expected goals (G-xG) data to look at the question in much broader strokes.
Evidence of finishing in the aggregate
One of the better articles on the subject of finishing was a study conducted by Michael Caley of Cartilage Free Captain which can be read here. In said article, Caley examined forwards in the EPL over multiple seasons and discovered that “almost all of the over-performance of expected goals among forwards comes from the performance of a relatively small number of high-volume shooters.”
Specifically, he found that among forwards, high volume shooters (which he defined as forwards with over 3.5 shots per 90) score about 15% more goals than expected, while medium and low volume shooters combined to outscore their xG by a mere 2.5%. “Shot volume predicts goal conversion in the aggregate,” Caley concluded, further explaining “better finishers get more chances.”
It’s truly excellent work by Caley and it’s an experiment which I have repeated within our data set to see if similar evidence of finishing can be found in MLS.
The parameters I used were quite simple, but slightly different than Caley’s. I aggregated four years of MLS data (2011-2014) and looked only at seasons where a forward had 1000 minutes or more, whereas, Caley put no controls for minutes. This came out to 254 individual forward seasons with a combined total of 13,621 shots, which compares favorably to the 287 individual season performances looked at by Caley which combined for 12,547 shots.
I then divided the group into the three categories specified in Caley’s article: Low Volume shooters, Medium Volume Shooters and High Volume shooters.
“Low Volume shooters” were defined as forwards taking 0-1.99 shots per game, which came out to 56, or 22% of our data set. By comparison, this group comprised 32% of Caley’s data set. The findings on this end of the spectrum happen to be the least important to this article; however, in the interest of putting all data on the table, low volume shooters in MLS fared better than their EPL counterparts, outperforming their xG by 6.8% compared to their 2%. The big caveat here is that this 6.8% is heavily skewed by our minute limiting in a way the other two groups are not.
The group actually under performs by 4.6% if you add the players with under 1000 minutes into the mix. Of course, this creates a bigger sample size than Caley had, and, given the skewing that occurs with so many sub 1000 minute players who get 4 or 5 shots, accruing xG but not playing enough minutes to let it stabilize, data gets skewed in the other direction. Given this group is unimportant to this article and the findings gleaned from the medium and high volume shooters, which I am about to go into, remain the same either way, I decided to keep a control on minutes in the interest of keeping the sample sizes similar.
“Medium volume shooters” were defined as forwards who took 2-3.49 shots per 90 within a season. We had 172 instances of this which makes up 68% of the data set. By comparison, this group made up 49%of Caley’s data set. Medium volume shooters in MLS fare about as well as their EPL counterparts, with MLS medium volume shooters outperforming their xG by an average of 7.7%, while this same group in the EPL outperformed by 7.3%.
Our final group is high volume shooters which, if you recall, is where Michael Caley found the bulk of the G-xG over-performance and strong evidence for the existence of finishing.
“High volume shooters” were defined as forwards who took 3.5 shots per 90 or more within a season. We had 26 instances of this making up 10% of the data. This group in MLS was significantly lower than in the EPL where Caley had 56 data points which comprised 18.5% of his sample. But this wasn’t the only discrepancy. One of the most significant findings in Caley’s study was that forwards in this group tended to outperform their xG by a whopping 14.8% . In our MLS data, however, the over-performance was a mere 7.2%, showing no significant differences with the medium volume group.
Does this mean that the top forwards in the EPL possess a skill that the top forwards in MLS simply do not? This is certainly possible, as there is an undeniable talent gap between these two groups. Of course, we could also be seeing an artifact of a small sample size since we only have 26 data points in the high volume category whereas Caley had 56. This sample size issue remains a problem even if we expand our data to players with fewer than 1000 minutes, while the congruity with the expanded medium volume group remains the same.
Of course, the fact that MLS has so few high volume shooters is itself a significant finding because it demonstrates that MLS does not show the same level of stratification in shot volume data. MLS has more medium volume shooters than the EPL and fewer low and high volume shooters.
Perhaps another explanation for this beyond talent is the leveling effect of MLS’s built-in parity. With all teams working under the same cap, the shot deficit between the “haves and have nots” is more level, allowing for a more standardized bell curve.
With regards to finishing, we see, as in Caley’s data, that there is a basic difference in conversion between defenders (who tend to slightly under-perform xG) and forwards (who tend to slightly over-perform xG). However, it appears to be more of a qualifying skill than anything. In other words, while we can say that consistently beating xG is something which most forwards can do that most defenders cannot, there doesn’t appear to be any stratification of this skill among forwards. Between strikers, G-xG and thus “finishing skill” seems to be fairly standardized.
To study this further, let’s take this study down to a more micro level and look for evidence of repeatability of above average finishing performance in a season.
Repeatability of Conversion
The general soccer watching public as a whole tends to put a lot of weight on a forwards “finishing ability,” which is generally described as acumen for putting the ball in the back of the net. In other words, it’s a matter of shot placement. A good finisher is able to more consistently put the ball where the keeper can’t get it than an average finisher, or so the theory goes.
If this were true, and finishing was a statistically significant skill when it comes to goal scoring, forwards who posses said acumen would consistently put balls past the keeper at a high rate, across multiple seasons. To examine this, I looked at shots on target conversion across the league going all the way back to 1996.
The average shot on target conversion rate for forwards within MLS is 37% without adjusting for penalties. If you adjust for penalties it’s around 35%. Earlier in the article I brought up Cyle Larin and Gyasi Zardes in regard to the signal vs. noise debate of finishing. In 2014, Zardes converted 54% of his shots on target, scoring 16 goals. In 2015 Larin converted 58.6% of his shots on target, scoring 17 goals. Both forwards were heavily hyped for their goal scoring totals that year, earned national team call ups and were tapped by many to repeat their performances.
But are such high conversion rates historically repeatable in MLS?
Since 1996, there have been 53 instances where a player with 15 shots or more on target has converted over 45% of those shots (stripping penalties). That comes out to about 2.7 players a year. Repeatability of this feat, however, is rare. Of those 53, there are only 8 instances where a forward did it in two or more seasons: Jozy Altidore (2007, 2015), Brian Ching (2004, 2009), Nate Jaqua (2006, 2009), Robbie Keane (2013, 2015) Obafemi Martins (2013, 2014, 2015), Melo Welton (97, 98) and Chris Wondolowski (2010 and 2012) are the only repeaters. That means 68% of the guys who pulled off this feat were only able to do it once, which isn’t surprising given the overall statistical noise we see in goal data in general.
There does, however, appear to be a bit of signal in that remaining 32%. For those who have followed the league for a long time, you’ll probably notice some commonalities in the above list of guys who have been able to repeat. Guys like Ching and Jaqua are classic “box forwards” which no doubt boosts the percentage of the shots they put on target. Guys like Melo and Oba, on the other hand, specialized in using their speed to get behind defenses, which no doubt helps conversion rates since keepers are more vulnerable.
It’s interesting to note that fewer still were able to do it in years where they scored 15 non penalty goals or more. Only Welton, Keane, Wondolowski and Martins were able to do it, and only Wondolowski and Martins were able to do it more than once. Here are all the instances where a player with 15 shots on target or more scored 15 non penalty goals or more in a season they converted over 45% of their non-penalty shots on target.
| First | Last | Team | Season | Min | G | NGP | SoT Conv | PK Adj SoT Conv |
|---|---|---|---|---|---|---|---|---|
| Stern | John | Columbus | 1998 | 2170 | 26 | 25 | 55.3% | 54.3% |
| Roy | Lassiter | Tampa Bay | 1996 | 2580 | 27 | 24 | 55.1% | 52.2% |
| Chris | Wondolowski | San Jose | 2012 | 2813 | 27 | 22 | 49.1% | 45.8% |
| Cobi | Jones | Los Angeles | 1998 | 2136 | 19 | 19 | 46.3% | 47.5% |
| Alex | Pineda Chacon | Miami | 2001 | 2055 | 19 | 19 | 52.8% | 52.8% |
| Cyle | Larin | Orlando City | 2015 | 1891 | 17 | 17 | 58.6% | 58.6% |
| Melo | Welton | Los Angeles | 1998 | 2435 | 17 | 17 | 45.9% | 45.9% |
| Robbie | Keane | Los Angeles | 2015 | 2060 | 20 | 16 | 51.3% | 45.7% |
| Obafemi | Martins | Seattle | 2014 | 2597 | 17 | 16 | 47.2% | 45.7% |
| Gyasi | Zardes | Los Angeles | 2014 | 2394 | 16 | 16 | 53.3% | 53.3% |
| Obafemi | Martins | Seattle | 2015 | 1805 | 15 | 15 | 53.6% | 55.6% |
| Chris | Wondolowski | San Jose | 2010 | 2308 | 18 | 15 | 50.0% | 45.5% |
With so many forward performances in MLS since 1996, this could just as easily be expected noise than evidence of a repeatable skill. With so much data, it’s simply a numbers game and a matter of time before one of them get 15 or more NPG from >45% conversion of SOT, and from there, only a matter of time till one repeats.
If we loosen the parameters a bit and consider anything above 40% to be a sign of finishing skill, we get similarly low repeatability. Of the 85 players who have achieved it, only 21% of them have done it more than once. We do see something, however, if we take this group and look only at those in it who score 10 or more goals in a season.
Forty-one players have scored 10 goals or more with a conversion rate above 40% and 56% of these names appear more than once. They are Wondolowski, Welton, Martins, Keane, Ching, Stern John, Thiery Henry, Ethan Finlay, Landon Donovan, Jeff Cunningham and Conor Casey. As you can see, we once again have a list dominated by box forwards and speedsters.
The above leaves us with two possible conclusions. Perhaps “good finishing” is the difference between a 35%-38% conversion guy and being at 40%-43%, with anything above 45% being viewed as anomalous. Another conclusion is that we aren’t really seeing finishing skill, but rather, skilled forwards in rolls that tend to convert at higher percentages such as box and speed forwards.
To explore this further, there is another way to look at this. One of the problems with shots on target conversion rate is that it will skew to forwards who shoot from closer range. To truly see if these forwards are beating average conversion rates relative to their shot locations, we must looks at expected goals.
Unfortunately, unlike SOT data, we only have xG data going back to 2011. Since xG is also influenced by time on the field, I decided to investigate a per 90 minute adjusted G-xG metric.
I specifically looked at forwards with 1000 minutes or more in a season from 2011 to 2015. Of that group, only five had multiple seasons where they averaged more than +0.1 G-xG per 90: Keane, Donovan, Martins, Henry and Camilo. Of those five, only Keane and Henry managed multiple seasons of 0.15 or more G-xG per 90. For reference, Larin in 2015 was at 0.30 while Zardes in 2014 was at 0.17.
When cross checked with the SOT data, it certainly appears that some of what is going on in terms of the outperforming SOT data is less finishing and more about shot location, as the box forwards are now gone. That being said, we also see Henry, Martins, Keane and Donovan showing up on both lists, and I think these are players that most fans would classify as “good finishers”.
This is not necessarily proof an underlying skill above the rest of the league when it comes to conversion, however, as players like Donovan, Keane, Camillo and Henry have xG numbers ballooned by penalty kicks. Martins, Donovan and Keane have also had seasons where an unusual number of their shots have come from counter attacks, which, as has been written about, is an attacking style that tends to outperform the model as a whole since the model cannot take defender positioning into account.
In other words, we still can’t definitively point to an underlying finishing skill as the cause, and even if we could, our conclusions would be limited to six players in league history.
As it stands, solid evidence of finishing skill in MLS cannot be found, at least not to a level of importance that is statistically significant to goal scoring. While a very small handful of forwards have managed to show some level of repeatability in certain metrics, expected goals data has managed to reveal how guys like Chris Wondolowski (no doubt the most elite box forward in MLS history) have been able to pull off such feats via their shot location. The consistent presence of counter attacking forwards in our lists further suggests that the circumstances of a shot (in this case, being one on one with the keeper) are having more of an effect than any underlying ability to put it past the keeper.
I should be clear that this is not a knock on these forwards by any means. These forwards are good precisely because they are able to get shots from high leverage positions so often, and, when you look at the top goal scorers in league history, you’ll see that none of them are particularly impressive in the conversion department. It was their longevity and ability to get so many good shots which made them such good forwards in this league.
The best forwards this league has seen have not been ones that have excelled at putting the ball in the back of the net. As we have seen, evidence of such a skill of any significance within MLS is marginal at best. Instead, good forwards rely on getting good shots, and the best forwards in the league get a lot of them.
At the heart of shooting is a law of averages, and, a lot of what most soccer fans perceive as “good finishing” is simply data noise in a small sample size of a season. A forward may outperform his expected goals by 10 one year, but attributing such a performance to finishing is folly, considering that that same forward may be end up finishing three or four goals under his xG the following year. While fans may find it hard to accept, finishing skill, at least in MLS, is largely irrelevant.