San Jose Earthquakes

Player Stats Analysis: End-of-Season Edition

Colin Etnire

— 15 min read
Share
Player Stats Analysis: End-of-Season Edition
Brain Meme
Earlier this year I looked through some statistics (both the advanced sort and the simplistic "counting" sort) to get an idea of whether or not my subjective impressions were on point. I take no pleasure in reporting that my doubts about Mikael Stahre, written this May after just his 9th game, turned out to be entirely justified.But now is anything but the time for gloating. What it's really about is evaluating the personnel for who should be back next year, and where improvements need to be made.Note: all statistics (where applicable) are based on a "per full game" basis defined by minutes, not appearances. Due to differences in how they collect data, statistics from WhoScored are on a per-90-minutes basis, whereas American Soccer Analysis data is on a per-96-minutes basis. Since those account for a "full" game in each system, I don't expect any discrepancies to be large. Also, it doesn't include the Houston game since those numbers weren't all ready.Fullbacks: Disaster Area, Bar OneLet's start here, because it's simple: Nick Lima was excellent in all phases all year long, whereas Shea Salinas simply never put it together on the defensive side, which is predictable enough from a right-footed career winger. Kevin Partida and Joel Qwiberg both got a handful of games at left back, but their stats were even worse than Salinas's. Oh and keep in mind? There was no backup right back on this team either, so had Lima not been remarkably durable this year, things could've been even worse.Here are the comparisons in the relevant areas, then, not so much to make the point that Lima is better, but more to give an idea of quite how much the Quakes were missing out on by having an inadequate left back:
Nick Lima Shea Salinas
xGoal Chain (Buildup contribution) 0.623 0.527
Passing % 74.8% 76.7%
+/- Expected Passing -0.5% +1.7%
Passing Verticality (yards advanced per average pass) -0.72 +3.26
Key Passes 0.77 0.99
Dribbles (offensive) 1.51 1.66
Fouled 1.51 0.64
Aerials Won 1.41 0.89
Tackles 1.61 1.91
Interceptions 1.81 1.28
Fouls 0.60 1.15
Clearances 2.91 1.91
Dribbled (defensive) 0.60 1.02
Not much to analyze here, other than to highlight the massive gulf in key defensive categories like fouls conceded, aerial duels won, and occasions dribbled. It's slightly interesting to me that Salinas comes out ahead as a passer (more vertical progress per pass, higher completion percentage despite more difficult passing, and more key passes) but I suppose that's to be expected for a winger-by-trade, and they're relatively even in most other offensive categories.Next steps are so obvious that I hardly need to say it: acquire a starting-quality LB. You might want to have a backup RB on the roster, too, but I don't want to get greedy.Attacking Midfielders: Why Do People Hate Magnus?Magnus Eriksson has become a whipping boy for a segment of the Quakes Twitterati, and it baffles me. The guy leads the team in all sorts of statistical categories essential to offense creation (supposedly the thing that fans cared most about improving) while tracking back and being consistent as a defender. My best guess is that there's a psychological tendency, especially in a sport like soccer, to overvalue guys who rarely if ever make visible mistakes but don't necessarily move the needle much, and undervalue guys who make visible mistakes yet also make major contributions to chance-creation, particularly if those contributions are one or more actions away from goalscoring.So with that, here's the table, with the three players who dominantly played attacking midfielder roles (Tommy Thompson and Christopher Wehan, although frequently in these roles, were excluded due to small sample size):
Magnus Eriksson Jahmir Hyka Valeri Qazaishvili
Shots 1.98 1.77 3.84
Shots on Target 0.71 0.62 1.21
% shots on target 35.8% 35.0% 31.5%
Shots per xG 8.6 9.3 10.7
Key Passes 2.30 2.08 1.85
xGoals (xG) 0.23 0.19 0.36
xAssists (xA) 0.25 0.13 0.15
xBuildup (xB) 0.57 0.61 0.61
xG Chains 0.90 0.73 0.97
Final 3rd Passing % 62.6% 82.8% 75.1%
Final 3rd +/- expected passing +0.8% +6.9% +3.3%
Final 3rd passing Verticality +1.44 -4.06 +0.02
Dribbles (offensive) 0.88 0.73 2.74
Fouled 2.96 0.73 0.95
Dispossessed 1.42 2.02 2.63
Poor Touch 1.64 1.28 2.00
All three attacking mids actually have remarkably similar statistics across a lot of categories, so instead I'll focus on the outliers. Vako's outliers are predictable: he generates the most expected goals (mostly by being the most prolific shooter) and he's miles ahead in terms of successful offensive dribbles. He's also, unsurprisingly, at the head of the chart in dispossessions and unsuccessful touches, given how much time he spends with the ball at his feet.Hyka is solidly middle-of-the-road in basically every category, if a bit of a laggard in the dribbles and fouls won categories (ordinarily essential for a winger's offense), but does stand out with the highest passing percentage.That leads me, however, to precisely my bone of contention with Eriksson: Hyka's passing percentage relies on a remarkably negative passing approach. He averages a staggering -4.06 yards advanced per pass in the final third. That means every time he releases a pass, the team loses 12 feet of territory in attack. Eriksson, in contrast, averages positive 1.44 yards advanced per pass in the final third, indicating that his passing is attempting to move the ball into dangerous areas. For me, coaches should prefer the latter, because risk-taking in the final third is necessary to actually generate goals, and the consequences of losing the ball near the opposing goal are fairly low.The advanced stats back that up the idea he's the best creator in the group: Magnus also leads the group in Key Passes and Expected Assists (xA) per 96 minutes. Magnus also is massively ahead in fouls won, which is particularly interesting to me since he is a fairly pedestrian dribbler without much pace. Those fouls won are potentially invaluable, since they lead to set piece scenarios that are significantly more dangerous than open play, particularly for a team like San Jose. Finally, he's the most "efficient" with his shooting, generating more xG per shot than the trigger-happy Vako and the trigger-shy Hyka.My conclusion about this group is simple: Vako and Eriksson, as individual players, are more than strong enough to be starting attacking mids in the league. Hyka is an excellent rotation option. If San Jose stands pat with that, it'll take a coach who knows how to utilize them, and how to design a system that compensates for any weaknesses they have. I wouldn't pull my hair out if that's the group on opening day 2019.However, there are scenarios where the Quakes could do better than standing pat. Elite attacking midfielders are among the best uses of DP slots and TAM funding, although fans should not underrate how difficult it is to find the Johnny Russells and Miguel Almirons et al on the international transfer market at anything close to a reasonable price. If the right fit becomes available, you pull the trigger. I also think there's a reasonable case to be made that Hyka's age and talent doesn't justify his annual salary. Tommy Thompson remains a perfectly adequate second stringer at these positions, but more depth would be welcome (particularly depth with pace, which is not a strong suit of the existing group). Finally, although each player can play well enough in his preferred role, there's so little depth or versatility that there are very few tactical looks on the table. I'd feel more comfortable with some different options depending on match-up and player availability. CBs: Not Great, But Outlook ImprovingFlorian Jungwirth is a difficult one to analyze statistically because he's played about half of his games at center mid and half at center back, and I couldn't find a way to separate his games in each position in either dataset. However, I think we have a pretty good idea of who he is regardless: as a CM, he's productive but tactically undisciplined; as a CB, he's rock-solid and starting quality, with skill on the ball. So I'll be excluding him from this list.The rest of the CBs are an interesting comparison on their own. Five different ones got significant playing time, because the defense was wretched from the moment the season kicked off, and Stahre appeared to be trying as many permutations as possible to see if it was a personnel issue (Arrested Development Voice: "It wasn't").However, when the dust settled, it was obvious which two of the true center-backs were starting-quality, and which of the backups is worth keeping around:
Harold Cummings Guram Kashia Yeferson Quintana Francois Affolter Jimmy Ockford
Yellows 3 2 3 2 3
Reds 1 0 0 0 0
Aerials Won 2.42 2.50 2.59 1.39 3.36
Tackles 2.22 2.00 1.09 2.03 1.81
Int 2.62 2.90 2.45 2.54 3.49
Fouls 0.71 0.50 1.63 1.27 1.55
Dribbled 0.81 0.90 1.77 1.27 0.65
Clearances 3.63 3.90 5.58 3.93 4.78
Long Balls 3.94 4.90 3.95 1.77 2.84
Passing % 86.0% 81.6% 76.5% 79.9% 79.1%
Passes 44.3 51.7 41.6 37.3 39.8
+/- Expected Passing +4.00% +4.00% -0.50% -2.40% +1.60%
Verticality +6.77 +8.55 +7.35 +5.56 +7.05
Defensively, the key stats to notice are not tackles and interceptions (since you don't want your CBs to be flying around the field winning the ball, necessarily) but rather fouls conceded and occasions dribbled, which speak to times the defender has been beaten. In those categories, Kashia and Cummings come out on top. They also have strong tackle/interception numbers too. The entire group, it should be said, has an excellent overall disciplinary record in terms of bookings. Although "clearances" could be interpreted as a positive stat, I tend to see it as a negative stat for center backs: it speaks to an occasion where the defense is scrambling and distribution fails to occur. Unsurprisingly, the backups register a much greater number of clearances.One thing worth noticing is on the distribution side: Kashia and Cummings are playing a higher volume of passes, and completing a higher percentage of them than their degree of difficulty would indicate, than their respective backups. Kashia in particular appears comfortable playing long balls from the back. That matches my subjective observations, and could serve as the basis of a more possession-oriented style, although it's interesting in the sense that Cummings in particular has a reputation as an enforcer rather than a ball-playing defender. Again, this is a position where the "next steps" are blindingly obvious to me: Quintana's loan should be allowed to expire, freeing up his large salary and international slot. Ockford should be retained as excellent cheap, domestic depth. Affolter has another guaranteed year on his contract, and has a green card, but should be shipped out if at all possible. Flo should stay in San Jose if he is willing to be a team player and serve primarily as a CB. Cummings and Kashia represent a perfectly adequate starting CB pairing and should be allowed to stay. No new major acquisitions are needed here: I'd rather see the numbers cut back from 6 to 4. Of course, as always, be opportunistic in the draft and other mechanisms in case good value becomes available. I truly believe scheme is the primary impediment to these guys succeeding.CMs: Godoy Returns to FormIn that early-season analysis, I was concerned about Anibal Godoy not playing up to his own high standards. After the World Cup, however, he found his groove. As referenced above, Flo won't be included in this group, but you can see his predominantly CM-based stats in that early season analysis. The interesting development here was Luis Felipe's promotion from Reno and subsequent starting-caliber play in the second half of the season.
Aníbal Godoy Jackson Yueill Luis Felipe
Yellows 8 0 4
Aerials Won 1.14 0.48 1.11
Tackles 2.60 1.77 3.87
Interceptions 2.08 0.48 2.10
Fouls 2.39 0.32 1.44
Dribbled (defense) 1.14 1.77 1.77
Dribbles (offense) 1.56 0.64 0.33
Fouled 2.70 0.96 0.55
Dispossessed 1.46 1.29 0.66
Poor Touch 0.94 1.61 0.77
Passes 57.0 57.6 49.8
Passing % 90.40% 85.30% 87.00%
Expected Pass % 86.30% 83.20% 85.90%
+/- Expected Passing +4.20% +2.10% +1.10%
Verticality +3.54 +2.63 +2.06
Long Balls 6.03 3.54 2.43
Key Passes 0.65 1.14 0.91
Shots 0.52 1.13 0.55
xGoals 0.03 0.08 0.04
xAssists 0.06 0.08 0.05
xBuildup 0.58 0.66 0.44
xGoal Chains 0.64 0.73 0.58
Godoy remains MLS starting quality, and the best central midfielder on the team. He also remains someone at his best when paired with a true defensive midfielder beside him to break up play and allow him to roam. His passing numbers are simply stellar (incredibly high completion percentage, best ball-progression of any of the midfielders, and almost the same buildup stats as Yueill). His offensive arsenal also includes dribbling stats second only to Vako, and foul-winning second only to Eriksson. Defensively, he's extremely productive, just a shade behind Luis Felipe in the primary categories. Of course, unsurprisingly, he concedes by far the most fouls of the group.Yueill's deficiencies become very obvious here when looking at the counting stats: he's simply not athletic or rugged enough to be productive as a defender. Nor does he have enough burst to dribble past his mark. He's truly a passing specialist. That can be a beautiful, invaluable thing, but he's only had modest success imposing himself in that manner in his first two years as a Quake. There are some promising signs: he has the best "expected buildup" rating and the most key passes. I don't have access to the "packing" statistic in MLS, but I would be willing to bet Yueill's numbers there would be very good. In the last month, after regaining his place in the team, and playing in a role better suited to his strengths, he's looked fairly sharp. However, he's still not influential enough to be a starter on a playoff team. Hopefully that will come with game time and development.Luis Felipe is perfect and should never change. He's the exact right fit for his role.Next steps? For me, either the staff has to commit to Yueill and fit the system around him to take advantage of his skills, or invest in a DP or TAM-level midfield passer (probably a number 10) if you think he can't play at that level. Simply put, the midfield wasn't good enough throughout much of the year. Additionally, the depth after the starting three is a huge question mark, so that needs to be shored up. This is a position where fatigue can be an issue, where rotation players are needed. Some might come from Reno (Eric Calvillo has a lot of quality and could serve in Yueill's role), but more than that needs to be done. I'd personally want to see at least two incoming central midfielders.The only potential wrinkle is if Godoy demands to be sold to Europe (a perfectly fair request after his four years of service to a miserable San Jose team) and you'd want to snag a DP or TAM level player to fill his boots.Strikers: They Are Not The ProblemI'm not including Vako in this category because even when he was playing as a nominal striker (which was the minority of his season anyway), he was playing in such a deep, recessed role that he functioned much more like an attacking midfielder. The only "true" strikers that saw more than a pittance of playing time were Danny Hoesen and Chris Wondolowski.Big caveat before the chart: I think passing statistics can be misleading for strikers, since they're frequently playing passes with a back to goal, and under much greater pressure than any of their teammates positioned deeper on the pitch. Even when comparing strikers to other strikers, the numbers will look extremely different depending on whether the guy plays back to goal or off the shoulder of the back line. So grain of salt, I suppose:
Danny Hoesen Chris Wondolowski
Shots 3.08 2.03
Shots on Target 1.27 0.86
On target % 41.2% 42.4%
Shots per xG 8.3 4.8
Difference between Goals and xG +2.98 -1.27
xGoals 0.37 0.42
xAssists 0.14 0.13
Key passes 1.35 0.72
Passing % 68.40% 72.10%
+/- Expected passing -7.60% -6.60%
Aerial Duels Won 1.34 2.04
Dribbles 0.85 0.41
Fouled 0.73 0.68
Dispossessed 1.58 1.09
Poor Touch 2.68 1.36
To me, what this shows is two solid strikers who play the game a bit differently. Of note is Wondo's unexpected advantage in winning aerial balls despite the fact he...um...can't really jump. Chalk it up to veteran savvy I guess. Also notable is that Wondo is generating more expected goals per game despite taking only about two-thirds of the shots that Hoesen did, which likely speaks primarily to the quality of the scoring positions he receives the ball in, a Wondo trademark.Finally, it's worth noting that Hoesen bagged 3 more goals than the underlying model would suggest, whereas Wondo had 1 fewer. While there are a few exceptional strikers on Earth whose elite finishing skills consistently see them outperform their "expected" numbers, the vast majority tend to regress to the mean. That means we shouldn't expect Danny to necessarily be able to top his total of 12 goals this year unless the team's overall dynamic changes, whereas conversely Wondo is a good shout to threaten double digits yet again. On that, after a bit of a down year in 2017, Wondo's xG/game this season has gotten closer to his historical average of 0.5, which means he really didn't have much of a down season after all. For me, he was just repeatedly misused by Stahre (remember when he played left wing for purely defensive purposes??), and there's no reason to assume his play will drop off a cliff in 2019.For the front office, then, the starting strikers are just fine. However, the MLS salary rules make using a DP slot on a striker extremely attractive bang-for-buck, so Front Offices should basically never stop looking to see if they can upgrade their forward corps. If an elite, true number 9 comes available, you take him, period. That would help solve the much more pressing problem: the team only has two strikers, and likes to play with two up top. Signing another elite striker pushes Hoesen to the bench in two-striker looks, and Wondo as well in one-striker looks. However, even in that case, and especially if an elite striker doesn't fall into San Jose's lap, they'll need more depth than they currently have. Dominic Oduro's contract is expiring and he shouldn't be retained. Brian Brown is a very viable Reno signee and I think he has the skills to be a good backup option. Mohammed Thiaw has a bunch of raw tools and could be worth a recall, but he's still probably a ways away in his development. Regardless, it's relatively easy to find backup strikers around the league or in the draft. Goalkeepers: Surprising no one...I wont go for a chart here, since goalkeeper stats are by far the most speculative that we have (even the advanced variety). The best advanced stat we have available is goals scored above or below the xG number, but even this is subject to so many variables I wouldn't bet my life on it.What it tells us is that Andrew Tarbell was the worst keeper in MLS this season, with 6.37 more goals conceded than the underlying model would've predicted. David Bingham is third-worst, with 4.79 more conceded than expected, so I don't have much time for the argument that retaining Bingham (at three times the cost, and without all the allocation money in return) would have made a meaningful difference in the team's defensive fortunes this year.J.T. Marcinkowski is now in goal, but the sample size is too small to say anything meaningful about him from a statistical perspective. From a qualitative perspective, though, he's worth the hype, based on my scouting of his performances with Reno. Oh and if you were curious about distribution? J.T. is playing the ball with his feet about 50% more than Tarbell, with significantly less vertical profile, and a better figure against expected passing. No one should be surprised.The trickiest thing for the FO to pull off is not any acquisition, since JT is ready to start in MLS in 2019. Rather, it's what to do with Tarbell. Only a year ago, he was a sought-after trade commodity. Now, he's close to worthless as a trade asset. Do you roll into next season with Tarbell as a replacement-level backup on a reasonable contract? Do you send him to Reno to rebuild his confidence and work on his game in order to build his value back up? Do you decide his value is more or less permanently shot and get rid of him? Those decisions, to me, will be of fairly little consequence to the team next year, as long as Marcinkowski is the starter.Team-levelIt took about 4 games to fully reveal how shaky this team was, and it never got better. There was a brief run, for the first 6 games after Luis Felipe and Kashia entered the lineup, where San Jose averaged a modest +0.13 xGD per game, which is just below where they were under 2017 Dominic Kinnear, and ordinarily translates into a fringe playoff team. Other than that period, however? An apocalyptic -0.55xGD/g, which would be the second-worst mark of any team in MLS in any season since Chivas USA folded. Simply put, this team wasn't good all year, and Stahre's final average of -0.35xGD/g isn't even close to being a playoff team. It's also a dramatic step back from last season, even though the personnel got better on paper. Coaching, coaching, coaching...One odd observation that periodically cropped up for me this season was: how friggen lucky was Stahre in terms of player availability? During my 3 years covering Dominic Kinnear, it seemed like there was a non-stop slew of injuries, suspensions, and international commitments that constantly interrupted his ability to put his best eleven players on the field. In 2018, however, Stahre suffered through no major injuries to any key contributors, relatively limited exposure to international breaks, and almost no suspensions. If anything, that just makes his abysmal performance even worse, and Kinnear's performance even better, in retrospect.This team will get better next season very simply through the introduction of a competent coach, so Jesse Fioranelli badly needs to get that right. An excellent coach may even turn the exact same personnel into a playoff team. And finally, a coach may help resolve some of the personality conflicts that have emerged within the team, and between players and management. Certainly there are some holes on the roster to be filled, but there is not a single player-related decision that is on the same plane of importance as who the next coach is.

Read more