Lowered Expectations: What the Whitecaps are doing wrong and Vako is doing right / by Harrison Crow

By Harrison Crow (@harrison_crow)

The premise of this column, for anyone new, is that we take the five highest valued shots from open play according to our expected goal model, and dissect the plays while talking about the players, the teams, and how our model works (or sometimes doesn’t). Sometimes I just yell into the wind about crossing.

 Let’s see what we’ve got.

#5 - Quincy Amarikwa, DC United, 0.476 xG (74.6)
Assisted By: Paul Arriola
Keeper: Stefan Frei
Result: Blocked

Amr.gif

First up it’s Mr. “Rent Free”, Quincy Amarikwa.

His first touch on this I think was purposefully rough. If you see how Brad Smith is coming in like missle, Amarikwa is trying to set himself up for a shot while creating some space. Unfortunately the ball gets a bit further away than intended and it makes for a poor shot with both Stefan Frei and Smith able to make a play on the ball. It was a good thought, and a less than great execution.

The build up to this is really the exciting part. I wouldn’t have guessed 14 months ago Wayne Rooney would have so many key moments winning possession in the defensive third or times feeding passes to attackers. Maybe I was naivete, but it’s still amazing every single time I see it.

The lead pass from Rooney to Paul Arriola into space after winning possession back is simple yet gorgeous. Arriola follows it up with his own great pass splitting the defenders for Amarikwa, after two dynamite sprinting touches helped to create his pass.

This is exactly what counter attacking teams are trying to do. Moments like these sometimes get lost when you remember they happen fewer than six times a game, and rarely go as smoothly as this transition. Most people don’t look at this chance and think “that was an incredible attack and chance created” but it was and it is.

Next up on our list is a great example of that...

#4 - Tosaint Ricketts, Vancouver Whitecaps, 0.481 xG
Assisted By: Ali Adnan
Keeper: Eloy Room
Result: Blocked

This exciting moment was one of maybe two or three in the second half of a game that agonizingly dull at times. Despite being at home, the Whitecaps looked utterly disinterested and hapless. This shot even happened so quickly that I don’t remember it even though I watched the game.

Here is what the Whitecaps second half pass map looked like in the attacking third:

It’s not great, but it’s actually not as bad as you might think. It lacks execution, but the thing which stands out most is the lack of volume. If you need some comparison, here is Atlanta’s pass map against San Jose in the second half at home.

Both teams had issues with “execution,” but due to Atlanta’s volume of chances it worked out for them. Vancouver, at home and against a team that has shown itself to be average defensively during the Caleb Porter era, needed to create more chances.

 I realize that it’s not that simple. You can’t just say “hey just take more shots” and poof more shots are created. Vancouver has a larger, more deeply rooted problem in the attack and I suspect it’s partially due to their ball progression.

 Our expected passing model (think of it as expected goals but for passing) shows that they make a ton of safe passes. They sit in the top 10 in safest passes out of the defensive third. What really stands out however is that they’re incredibly bad at it. In other words, they’re making passes that are neither difficult nor dangerous, but also not completing they as much as they should. Only Columbus has been anywhere near as poor in their own defensive third.

Team Passes per game PassPct xPassPct Score/g ScorePer100
FCD 177.8 84.0% 83.2% 1.5 0.8
SKC 149.4 83.1% 82.1% 1.6 1.1
LAFC 132.7 83.0% 82.0% 1.4 1.0
SJE 158 82.6% 81.7% 1.4 0.9
TOR 151.1 83.3% 81.5% 2.7 1.8
NYC 178.5 81.2% 80.5% 1.2 0.7
ATL 137.1 80.8% 80.4% 0.6 0.4
VAN 166.6 78.8% 79.2% -0.7 -0.4
SEA 138.6 79.3% 79.1% 0.3 0.2
CLB 147 78.3% 78.7% -0.7 -0.4
Score: Total additional passes completed over expected: (Pct - xPct)*Pass.

This is where they need to start to fix the problem. It’s not a Fredy Montero or Theo Bair problem (I mean, I’m sure they have their respective issues--just about everyone on this team has something they can do better on), rather it’s more of an issue with whose responsibility lies with getting them the ball in advanced attacking positions.

Here are their worst passes, as defined by pass percentage minus expected pass percentage:

Player Min Pos Passes PassPct xPassPct Score Per100
Andy Rose 1829 CB 787 81.2% 82.2% -8.3 -1.1
Zac MacMath 783 GK 311 71.4% 74.0% -8.2 -2.6
Maxime Crepeau 2330 GK 880 72.0% 73.0% -8.1 -0.9
Victor Giro 833 FB/WB 350 80.3% 82.5% -7.8 -2.2
Ali Adnan 2505 FB/WB 1431 74.8% 75.2% -4.5 -0.3

Ali Adnan being on this list while also getting a rather big contract from the team isn’t a great sign. His build up out of the back has been somewhat culpable for their issues with ball progression.

#3 - Andres Rios, San Jose Earthquakes, 0.567 and 0.507 xG
Assisted By: N/A
Keeper: Brad Guzan
Result: Saved/Blocked

This whole chain of events is rather crazy. Andres Rios has not one but two shots awarded on this play, which are considered to be part of the same chain and (technically) earned him the 4th and 3rd most valuable missed shots this week.

Maybe you wonder about how each of Rios’ chances are worth more than half an expected goal (0.567 and 0.507 respectively).

“HOW DOES THAT EVEN WORK, HARRISON?!?! THAT MAKES NO SENSE! HOW CAN A PLAYER PRODUCE MORE THAN ONE xG AND IT NOT BE A GOAL?”

Well--simple, we have multiple models. They don’t equate to one xG for our team model, just the player model. Combined these chances actually equate for 0.78 xG for the team. Let me say that once more FOR THE TEAM.

From our chief analytical mind, Matthias Kullowatz on the difference between the player and team models:

“There are really just two sources of difference: sequential shots and penalties. The hope is that this version of xG is more stable (because penalties and multiple shots in sequence aren't really "skills" per se)”

While the player model gives Rios the individual credit for the respective opportunity, the possession sequence itself is capped at one possible expected goal across the entire event chain. This gives us a better idea how the player’s capabilities while also being able to prognosticate the team and their long term capabilities.

#2 - Rashawn Dally, FC Cincinnati, 0.643 xG 
Assisted By: N/A
Keeper: Brad Guzan
Result: Saved

Rashawn Dally makes a great run and, if not for the way the ball comes in, he’s looking at an incredible opportunity. The pass is a solid one from Caleb Stanko, but it’s also a bit tricky to hanlde. Props on Dally for getting a foot on the ball while it’s coming across the face of goal. It’s a bit difficult to make contact with, but he’s able to direct it towards the frame.

It’s good to see Cincinnati try and be interesting. They seem intent on playing a possession style and an aesthetically pleasing brand of soccer. Unfortunately I think they’ve done so at the expense of the collection of their talent.

They were probably going to miss the playoffs regardless of the style they played, but they now seem to have something to build on for next year. While that’s an entire year too late and with their expansion resources drained, they still have some semblance of a plan to get better and plan to build with a specific style in mind.

I maintain they have the talent to have played a more pragmatic style that would have caused teams problems in breaking them down and created more on the break than in their possession attacking style.

#1 - Danny Hoesen, San Jose Earthquakes, 0.672 xG
Assisted By: Valeri Qazaishvili
Keeper: Brad Guzan
Result: Saved

Danny Hoesen’s shot happens so quickly that you almost miss what happens. He almost looks to have fallen down rather than sliding towards the ball to scoot it past Brad Guzan. The real takeaway is Vako and his slick moves inside the 18 to drive deeper towards the box and create a shot opportunity.

Vako came into the league in 2017 with a bit of publicity, but slid off the radar after an unremarkable season. He had a further dip in his stock last year, despite producing 10 goals and five assists, after he served on a team that struggled mightily.

That said, his underlying numbers showed great potential. He produced 9.5 xG in 2018 from open play, and that fell right in line with his goals scored. He even ranked 22nd overall in the league in xG+xA from open play, despite not getting a lot of attention.

qaz.png

This year however... he’s really started to bloom and has taken a step forward.

qaz2.png

What he excelled at last year he continues to excel at, plus he’s integrated into Matias Almedya’s Samurai Soccer program with a more defined role.

What’s really interesting is how Vako has a bit of Giovinco flavor to him. Below you can see each of their stats per 96 minutes in their first three seasons in MLS.

Vako Open Play Stats
Season Shots G xG xG pSH KeyP A xA xG+xA
2017 3.97 0.5 0.46 0.116 1.4 0 0.1 0.59
2018 3.29 0.3 0.32 0.097 1.37 0.1 0.1 0.43
2019 4.29 0.2 0.31 0.072 1.56 0.2 0.2 0.47
Giovinco Open Play Stats
Season Shots G xG xG pSH KeyP A xA xG+xA
2015 3.88 0.4 0.39 0.101 1.36 0.4 0.2 0.62
2016 4.76 0.4 0.4 0.084 1.25 0.3 0.2 0.6
2017 3.6 0.4 0.33 0.092 1.2 0.3 0.1 0.47

Obviously Giovinco was a great player and Vako probably doesn’t fit in that same tier. That doesn’t mean they don’t have similarities.

I want to focus on the fact both are individuals who love to take shots, both frequently and from any location on the field. The byproduct of that for both is they’ve created an attacking gravity which pulls defenders to them. Each also have the creative and technical abilities to find ways to exploit those 1v1 opportunities to find space and create high leverage shots.

Vako might not be some wizard creator with assists, but his presence on the field still means a lot to his teammates much in the same that Giovinco meant for his teammates in Toronto.

Lofty Expectations: Justin Meram, Atlanta United, 0.033 xG
Assisted By: N/A
Keeper: Daniel Vega
Result: Goal

Meram.gif

This is both a great example of why we call this a low leverage shot and also why it’s just important to sometimes get a shot off. Justin Meram could have laid the ball off and allowed Josef Martinez to take possession. But Meram had a shot and he correctly took it. It’s not the result that makes it the correct choice, but rather the likelihood of a better shot taking place after the recycling of possession. A better shot could very well have happened. But the time would it take to create that shot and the odds it would it actually be better are all things you have to think about.

One thing to add here is how incredibly lucky this result ends up being for Atlanta, who seem to swim in luck like a leprechaun selling breakfast cereal. That said, they made some of that luck happen just by their movement, regularly getting the ball into dangerous places and taking chances.

xG minutiae and other office clean-up

Nicholson1.gif

As pointed out on twitter, this was a crazy low xG goal and it would have supplanted our “Lofty Expectations, goal of the week” last week had it been from open play. This specific event was credited as being part of a free kick possession chain which is why it wasn’t included.

 Let’s talk about run of play versus dead ball scenarios in the expected goal model and some thinking with them. 

 Obviously, when a foul is committed it gives time for both teams to change where their players are located are on the field. This isn’t different than open play, but it is a slight deviation in the fact that the ball doesn’t progress with them, which changes a few things in the approach. We see this as a specifically popular topic for where analytics might be able to help moving into the future as data progresses.

 One specific thing stands out in the GIF above, which is that the defense is allowed to set itself. We see the majority of Toronto’s defense dropping back, finding their line, and spacing and preparing themselves for the eventual transition of the attack. This is a big advantage and it’s one of the reasons why we want to specifically tag that change in phase of play.