Weekend MLS Projections

Philadelphia Union versus New York Red Bulls

Philadelphia has performed much better in crucial times on the road this season than has New York. Though Philadelphia isn't playing on the road, that fact still encourages the model that the Union is the better team. Plus Philadelphia is playing at home. 

PHI: 60%

NYRB: 22%

Draw: 18%

Spread: 0.80 (PHI)

New England Revolution versus Montreal Impact

Past shot data show these two teams to be fairly equal, though the standings don't reflect that. Jermaine Jones throws a big wrench into the model because he hasn't played enough to start affecting the Revs' stats. The following probabilities are likely underselling New England a bit.

NE: 48%

MTL: 20%

Draw: 32%

Spread: 0.49 (NE)

Chicago Fire versus Toronto FC

The stats show Toronto to be the superior team, but the game is not in Toronto, and home field advantage is a big deal. The following probabilities can't possibly account for Toronto's recent mess, but that likely has a smaller immediate effect on the field than pundits would have you believe.

CHI: 41%

TOR: 28%

Draw: 31%

Spread: 0.25 (CHI)

FC Dallas versus Vancouver Whitecaps

Earlier in the season I raved about Vancouver's excellent start to the season in the Expected Goals department. I jumped the gun, and that is no longer the case. The stats say Dallas is a legitimate #3/4 seed in the West, posting a top-five xGD when tied at home. Vancouver, on the other hand, has produced the single worst xGD when trailing by a goal on a road--a situation in which away teams are actually playing out.

FCD: 66%

VAN: 18%

Draw: 16%

Spread: 0.99 (FCD)

Houston Dynamo versus Columbus Crew

Houston has played well at home this season during even gamestates, and Columbus gives the model reason to believe it can come back and draw on the road by playing well while facing one-goal deficits. The probabilities below reflect that, with relatively high chances of a Houston win or draw.

HOU: 52%

CLB: 13%

Draw: 35%

Spread: 0.68 (HOU)

Colorado Rapids versus Portland Timbers

The Timbers have not produced the quality or quantity of shots that they did last season, and have allowed opponents too many good looks. Though Colorado is in a free fall, and coming off a record-setting 6-0 loss to LA, the model still favors Colorado for its season-long performance. The Rapids have produced a respectable xGD during even gamestates at home, and about the only thing Portland does well is force draws after giving up early leads. Unsurprisingly, the probabilities below look like those of the Houston game.

COL: 57%

POR: 14%

Draw: 29%

Spread: 0.82 (COL)

San Jose Earthquakes versus Los Angeles Galaxy

Our model drools over the Galaxy, but the probabilities below might seem a little conservative. Past data shows that, even in the most lopsided of matchups, home underdogs find ways. In fact, we don't need to look back too far to find a pertinent example

SJ: 32%

LA: 39%

Draw: 29%

Spread: 0.06 (LA)

Friday's MLS Projections

Our next version of Expected Goals is coming, which will include more specific shot location data, as well as additional information about the key pass for each shot. Though Expected Goals 3.0 is not ready yet, one byproduct of its creation is that I now have shot data from 2012 and 2011, and I can make better prediction models. 

So these projections are still based on xGoals 2.0, but the model has more history to work with when evaluating the importance of various statistics.

Seattle Sounders versus Real Salt Lake

We know how much this version of xGoals thinks of RSL, and it's not much. Seattle is favored to win with 61.5 percent probability, and RSL with just 13.5 percent, leaving 25 percent for the draw. RSL's xGD on the road this season is third worst at -0.60 expected goals per 96 minutes, while Seattle's +0.77 at home is second only to the LA Galaxy.

Chivas USA versus Sporting Kansas City

Oh, there's another game tonight? While home teams typically win nearly 50 percent of the time, this is no ordinary matchup. Chivas has just a 39-percent chance of winning, and SKC matches that same 39 percent, leaving a 22-percent chance for a draw. Chivas' home field advantage may come across as the weakest in the league, but its -0.08 xGD at home essentially equals SKC's -0.05 on the road.

Portland throttled San Jose, settled for draw

I'm not sure if the Timbers really could have scored 10 goals, but they produced one of the more lopsided ties in recent history.

Below is a chart of the largest single-game Expected Goal Differentials (xGD) over the past two seasons. 

Team Date Opponent xGF xGA xGD Evenminutes xGDzero GD
SEA 5/11/2013 SJ 4.01 0.00 4.01 27.60 0.91 4
SEA 7/14/2014 POR 3.29 0.50 2.79 71.50 2.27 2
RSL 3/30/2013 SEA 3.96 1.20 2.76 7.80 1.84 1
POR 6/11/2014 FCD 3.93 1.31 2.62 28.90 -0.49 0
NYRB 3/16/2013 DCU 2.76 0.23 2.53 93.80 2.53 0
HOU 9/21/2013 CHV 2.91 0.43 2.48 42.40 0.93 4
LA 7/20/2013 VAN 2.83 0.36 2.48 48.60 1.90 1
LA 3/23/2013 COL 2.73 0.29 2.43 52.30 1.97 1
SKC 7/13/2013 TOR 2.66 0.36 2.29 21.00 0.17 3
FCD 7/20/2014 NE 2.75 0.50 2.25 28.20 0.87 2
LA 3/3/2013 CHI 2.77 0.55 2.22 37.40 0.14 4
LA 8/17/2013 RSL 3.44 1.24 2.20 62.90 1.67 2
POR 9/7/2014 SJ 3.42 1.25 2.17 39.80 0.34 0

Portland's 2.17 edge in Expected Goals ranks 13th in the past 575 games, and third among games ending in a draw. A positive xGD during the 40 minutes played at an even gamestate also suggests that Portland didn't wait for a deficit to turn on the jets--a good sign for a team that has allowed itself to get blitzed early week in and week out.

The average actual goal differential on this list is almost two full goals per game, so it's not any great stretch to say that Portland will see better results if it can continue to produce better xGD results. However, that's a tall order for a team that is still ranked third-to-last in even gamestate xGD.

#FreeBieler

By Harrison Crow (@Harrison_Crow)

When you look at all designated players across the league, there are few that have seen the type of playing time that has been afforded Claudio Bieler, and I'm not talking about that in a good way.

As of right now there are only seven other DP's out of a league wide 45 that have seen less time with their respective clubs this year. Though he's not last in that department, let me actually break it down a bit further to provide adequate context so you can jump on board this train.

Of those seven that have seen fewer minutes, only two started the season with MLS clubs. A grand total of TWO. That's one, two...two. You can't even squeeze a Hobbit into two movies let alone provide for only two other DP's to accrue less time off the bench.

The two mentioned are Alexander Lopez of the Houston Dynamo and Jerry Bengston of the New England Revolution. To make matters worse Bengtson has been on loan for over a month and could have possibly caught up with Bieler as you can see provided by the image below.

The peculiar issues of Bengston and club-vs-country form have been an issue for some time. It culminated with the Honduras national team star being found in a loan situation to help resolve what has turned into a very sour situation.

The situation down in Houston with Lopez is more unique to itself and a bit complicated due to factors of age and... well, let's skip over this and just stay focused squarely on the Argentinian product, Bieler. 

Bieler---once a striker for Sporting Kansas City and now firmly planted on SKC's left bench---hasn't seen much time for a variety of reasons this season. The most commonly cited concern fitness or injuries. This isn't to say he's not able to physically keep up with 90 minutes of soccer, but rather he can't keep up with the 90 minutes of Kansas City soccer. A brand in and of itself that demands high tempo pressing from the first minute until the last.

Bieler isn't probably the best fit for this type of team-emphasized system, and for a lot of reasons that we just don't need to get into. It turns out, the guy that KC wants and needs in that position is already on the team. Dom Dwyer has developed into the perfect piece for the organization, and with the current depth in the roster the need for Bieler at this point seems redundant, and more than anything it forces KC out of their normal element.

 

Forgetting about why he's been locked or frozen out of the starting line-up by head coach Peter Vermes, the opinion from most pundits is that Bieler is still a good player and maybe even a borderline great player with the right team in place around him. As a point of reference look at his xGoals.

Just last year he produced 9.29 xGoals for the season finishing 11th overall on accumulation. That was a nudge over 20% of all of Kansas City's expected goals accumulated. Over a full season, he would be on pace for the 10+ xgoal range, which would put him somewhere in the assumed Top-15 of total goal scorers in the league. Think about that, and then realize that he's currently one of the cheaper DP striking options in MLS---beaten only by Erick Torres ($152,000)---at a gutbusting $225K annually.

El Taca's Designated Player status is largely wrapped around the fee that was paid to acquire him from LDU Quito, which is rumored to be a little over a million. Should that be correct, and based upon what information is available, that DP tag could fall off this off-season. Making him one of the top targets around MLS, assuming Bieler is willing to stay in the states.

The word at last season's end was pretty much that Bieler was done in KC and that he would be looking for... something else. That didn't happen. So you have to wonder if a season with far less opportunities is going to prompt him to "run for the hills".  That said a legit contract and playing time could always tempt a player of his situation into staying with the league.

He honestly makes perfect sense for far too many clubs that will seek to resolve issues with their attack. A couple quick examples would be: The club formerly known as Chivas, Houston and Vancouver would all three be really interesting cases to look at. That's not even accounting for New York, LA and Seattle all having potential questions marks with their men up top going into next season.

This is all a waiting game but I would be excited to see what Bieler could do to this league if given a chance.

 

#FreeBieler

 

Did Toronto FC make a mistake firing Nelsen?

Yesterday morning/afternoon, or some various time during the day, Toronto FC finally parted ways with Ryan Nelsen, making him the eight head coach to be fired in 8 seasons of MLS. It's more than enough turnover to suggest that the club may have a bigger issue.

It's also comes with speculation that Jermaine Defoe could accept a transfer back to the EPL with a reported "many teams" chasing him. Most notably QPR--with whom TFC has done quite a bit of business through the past year.

Getting back to Nelsen; his termination come on the heels of GM Tim Bezbatchenko coming out and saying it is time "to take it up a notch".  He obviously wasn't happy about the teams "current form". As MLSsoccer.com has pointed out, the club is 2-4-1 in their last 7 matches. 

It's been a dubious spell for sure, with defensive injuries and inconsistency in the normal 18. This left Nelsen juggling things in a manner that was apparently inadequate according to the new leadership at TFC. This brings us to the real issue at hand.

One of the most heavily emphasized issue was Nelsen was making excuses when he should have been owning up to mistakes. Which is... well, whatever. It's fine for all I care. However it does lead one to progress the thought and wonder if it was really his inept ability for the "X's and O'x" in tactical realm of MLS that is what caused the front office to make the move.

I tend to be on the liberal side of what tactical acumen should be expected from a manager in today’s era. I'll certainly admit that there is a need to have a good understanding of the principles of tactics. But I think players' skill and circumstance are often credited to the manager and his tactics rather than just saying, "wow, Team X sure got lucky there," as if that takes something away from the team, staff or their players.

Sometimes coaches do their homework and set things up the best they can and the chances just don't go their way. Sometimes there are players that know how to read the game and understand how to take advantage of it. In short, the media and many followers of MLS (and soccer in general) overemphasize outside factors, and underestimate the impact of luck on results.

At any rate; the real issue at hand is that Bezbatchenko obviously doesn't look at the statistics on our site (the nerve!). We currently have their club at a near 90% probability for making the MLS playoffs. This despite there being five teams (spots 3-7) within 3 points of each other. It's not that we're just crazy. A peek over the fences at Sports Club Stats has the reds at 71% odds to make the playoffs but with the #4 seed.

Bezbatchenko has stated from the start that Nelsen's job was simply to bring the first playoff match to Toronto. If that was the case, it would seem the Nelsen was well on the way to complete the task he was brought to do.

Some say there is an immediate (and short-term) improvement a team's performance after a manager is fired. I'd love to see if someone could prove that. With 10 games left in the season, a slight bump could cement Toronto as a #3 seed. If that doesn't work, it's a huge gamble that could go in the opposite direction, leading Toronto right out of the playoffs picture.

As Danny Page pointed out via twitter "manager firing usually happen near the lowpoint of luck based stats, so generally they regress towards mean". The question though becomes whether it's really a low point.

Over the last 7 matches the Reds are running a PDO (which measures luck) of 911 which would play to Mr. Page's point. A change could mean very little in the sense that the club was bound for positive regression.

In short, the data seem to suggest that Toronto FC made a mistake. Ryan Nelsen may not be a good long term manager but they picked a poor time to fire him. I'm not a TFC fan or engrossed in the Toronto soccer/football culture, but they've been a bad club for a lot of years and an estimated 256% increase to their salary inclines someone to think that missing the playoffs would be devastating.  And desperate people do... well, less than smart things sometimes. Like firing their coach with only 10 games left in the season.

It also doesn't help that historically they don't have a propensity for doing a lot of smart things anyway. In short, this move will define TFC's season. While it will certainly bring change, I'm unconvinced it will bring improvement in style or performance. In a couple months, we'll all know for sure.

The Burgundy Waves Goodbye

By Harrison Crow (@Harrison_Crow)

With the Colorado Rapids surrendering a late lead last night against the LA Galaxy, and contributing to the storybook farewell tour that is the now intrinsically tied to Landon Donovan, Colorado will surely be expected to drop further down the ranks of our playoff prediction calculator. It will also make the five teams ahead of them near stone-cold (or should I say adamantium cold?) locks to make the playoffs.

That's sad to me for a few reasons. First, Marc Burch's hair deserves the playoffs or a medal or something that proclaims it's awesomeness through time. Second, Colorado has been a very strong club this season. If they were in the East they would currently be in 4th place in the conference. It is a bit disappointing they'll likely miss out on a playoff game to show off what they are about.

Last season, and prior to the drama laden departure of Óscar Pareja to Dallas, the focus of the club surrounded it's trademark 4-2-3-1/4-3-3 and transition attacks featuring the stunning pace of Deshorn Brown. This season, with new coach Pablo Mastroeni, a new philosophy has emerged that has shown a more complete and versatile attack than the run-and-gun days of Pareja.

Taking the youth uncovered last year with Dillion Powers, Chris Klute, Shane O'Neil, Dillon Serna and the aforementioned Brown, Mastroeni has built upon it them and in reality taken them further than what their points or table placement may indicated.

While their goals conceded is up (1.56 this year to 1.15 last), their total shots allowed are down and based upon shot analysis it implies that the defense has probably more likely carried the club rather than hurt it. This is even more impressive when you consider the host of injuries that have hurt the team in 2014.

While they have had a marked increase in goals conceded inside the 18 yard box, that's not necessarily indicative of their performance. Our data suggests it may have more to do with luck than their backline talent. According to our xGoals table, they're ranked third in the league in expected goals allowed with 1.10.

Drew Moor, the team captain and leader of the backline, has a sort of Chad Marshall disease associated with him. A consistently good player in his time in MLS split between Dallas and Colorado, he's been ranked average among his peers while making a pretty average salary. Losing him for the season with a torn ACL may have been the last nail in the coffin for the Rapid's playoff chances.

Moor's perception as a good but not great defender is due partially because his performances come with the asterisk that ties his cap percentage hit with his on the field value. This is rather a tough way to evaluate a player, especially on defense. His spectacular play this season has gone relatively unnoticed in lieu of the career resurgence of other aging centerbacks such as Bobby Boswell and Chad Marshall.  While it is hard to fit all the names into a nice little feature that recaps the season, it would be a crime to leave Moor off an "top" defensive lists at seasons end.

Though they have lost Moor for the season and are unlikely to make the playoffs, this season has not been a wash.

The development of both O'Neil and Serna have been huge. Their rash of injuries has shown a surprisingly deep and talented club that has quietly floated among the top tier in expected goal differential for most of the season. It's a shame we probably won't see their exciting play fighting for MLS Cup in November, but it's clear Colorado will be back next year and further challenging RSL, LA and Seattle for a piece of the Western Conference pie.

 

Herculez Gomez: American warrior in Mexico

By Harrison Crow (@Harrison_Crow)

Sortable stats can be found at the end of the article for those seeking a more interactive experience.

We embarked on a project some weeks ago to begin focusing on shot data for past and current US Men's National Team players. We came to the realization that these data are not congregated in one place for American players, especially those playing internationally. For example, the Wikipedia page for our first player, Herculez Gomez, shows that he scored 24 goals for Santos Laguna. Try as we might, we could only find 22. 

Wikipedia has been known to be wrong--Matty informs me that he was listed as the creator of Pokemon for some time thanks to the work of his college roommate--but finding Liga MX and CONCACAF Champions League data from games four years ago is tough. Really tough. Poor Drew is probably going to need to go see a counselor, based on some of the emails he sent me. So we could have missed something. But with his determination and Matty's all around know how, they joined forces and dug through more than 175 games for detailed shot data, and we were able to derive some pretty cool stuff.

First of all this stuff comes with a couple of asterisks. 

1) Liga MX has weird seasons. Its time frame, and how ESPN decided to provide game information, doesn't always coincide with the actual season. So we put seasons together as best we could. By the way, ESPNFC was a big help.

2) We had a few games where we couldn't find shot locations. As a result we ended up using a baseline expected goal per shot for those few. It shouldn't skew the results much either way, as it represents less than 5 percent of the data, but it's an important annotation.

3) ... actually there was only two things that really came up. It's just weird to stop at two and not count to three.

2010-11

When Herc left the Kansas City (then) Wizards, it wasn't necessarily on the greatest of terms with the club or the league, as we can see from some of the banter on social media. He left MLS and played his way into the Puebla starting line-up, winning the Mexican Primera Division scoring title with 10 goals. He became the first American to win the Primera scoring title, edging out Javier Hernandez in a tie-breaker by virtue of scoring his 10 goals in fewer total minutes than "Chicharito."

Now, there are plenty of articles from which I could cherry pick to show how people felt about Gomez and his inclusion on the US World Cup roster in 2010. As with any player, it's fair to say that he scored some goals that were a little lucky. But our numbers show that there was more than a little luck to his overall performance. He produced 2.75 shots per 90 minutes and scored nearly half of them.

That's about as "clinical" a finisher as you're going to find.

I don't mean to speak for everyone here at ASA, but I think it's come out in the podcasts that we're all a bit skeptical about finishing ability and a player's ability to reproduce past success. One goal per every two shots is stuff that no striker will sustain.... That is, unless he scores a goal and then retires. I'm sure it's happened.

During his 718 minutes (not including stoppage time) with Puebla, we have him down for 5.65 xGoals compared to the 10 he actually put home. This information is not meant to undercut him, but rather to show that he was outperforming the norm. Lots of players do this from time to time--maybe for one season, occasionally for two. Then most regress, and we chalk it up to unsustainable play, perhaps aided by the residual effect of a teammate or a few too many lucky bounces.

2011-12

The thing is, Gomez continued his crazy pace of outperforming his expected goals numbers. He followed up his 2010-11 season by scoring 17 goals in 2,030 minutes in all competitions spread over three different clubs. I think what stood out the most to me was his ability to do everywhere he played. Despite scoring at what seemed like an unsustainable rate by our metrics and playing with new teammates constantly, he was able to finish as well as he had the season prior.

Thinking about what Brian McBride and Clint Dempsey did over in England as strikers is, of course, admirable. Dempsey still ranks in the top 60 in total goals over the last 20 years of the English Premier League. But these two seasons compiling 27 goals, averaging 0.88 goals per 90 minutes, is a considerable accomplishment and something to savor for a few moments. Go ahead, savor it.

Gomez started etching his name in the CONCACAF Champions League that year, too, scoring six of his 17 goals in those games against MLS clubs (three against Toronto FC, 3 against Seattle Sounders FC). All told, it was perhaps an even stronger performance than the season prior. Over the course of the 2011-12 season we had him for a total of 8.54 expected goals from 71 total shots. He obviously shattered that expectation.

2012-13

2012-13 was a coming-back-to-earth season of sorts for Herc. He played nearly the same number of minutes (1,691) in the Liga MX regular season as he did in the two prior seasons combined (1,794). That, in addition to Santos' dependency on him in the CCL, perhaps resulted in him taking fewer shots per 90 minutes (2.4) than in either 2010-11 (2.8) or 2011-12 (3.1). He tallied 13 goals at a conversion rate of "just" 20 percent. Recall that the MLS average, for comparison's sake, is just a shade over 10 percent.

That's about the extent of his "down year." But in truth, accounting for his shot opportunities, it was another plus year for the American. Again, he outperformed expectations, finding 13 goals versus the 8.60 we expected him to score.

2013-14

This time regression hit Gomez across the board. He scored just eight goals across all competitions versus our expectation of 8.88, representing his first truly down year finishing. But one thing to point out was that his shots per 90 were up and his minutes were down. He was more productive creating shots on a per-minute basis, though this may have more to do with his usage. He was often used as a substitute for Santos, and almost exclusively as a substitute for Tijuana. Evidence suggests substitutes cover more ground and increase the chances of their teams scoring goals. It's a not a great leap, then, to conclude that it might also increase the individual's shot totals productivity.

However, the first study linked above showed that a substitute's pass completion percentage remains unchanged, indicating that perhaps the boost is restricted to gross motor skills. Despite being used as a sub often in his career, especially this particular year, Herc's overperformance in the finishing department over the past four years shouldn't be entirely chalked up to the substitution effect. 

Still, he scored just four of those eight goals in regular season play over a whopping 1,691 minutes played. We had him for an expected 6.74 goals over that time. Assessing his play in Liga MX that year helps us to better understand the common narrative that he was struggling at that time.

That said, he produced an expected goal once every three full matches in Liga MX. If we were to compare that to our MLS data, that same shot production would put him in the same company as someone like Mike Magee or, on the high end, Chris Wondolowski. While neither are having seasons that would blow you out of the water, both have been consistent contributors to their squads, and I think Gomez could still be in that range now.

Timeout

At this point, we have Herc down for 48 goals scored. If we were to allow thousands of typical finishers to take the same shots that Herc got, our expected goals suggest that those players would average 34 goals scored with a standard deviation of about 5. That places him 2.8 standard deviations above the mean, and into the 99.7th percentile. While it's important to note that these expectations come from MLS play, we already found that MLS teams converted opportunities similar to World Cup teams. It's probably fair to say that Liga MX is not that different from MLS, and that Herc possesses a finishing skill not common among MLS or Liga MX players.  

2014-15

We only collected the first half of the Liga MX season data for 2014-15, so the last game in our data set is from March 28th. To that point, Gomez had not yet scored for Tigres, and it seemed as though he'd fallen off the map. I wondered how much he could still contribute to an MLS squad if he ever made the move. However, a more in-depth look revealed that through those seven documented matches, in limited minutes, he still managed to compile 18 shots worth nearly 2.2 expected goals.

He's still finding good looks and he's still taking those shots. I have him at just more than 2.1 shots per 90 which, in combination with his 2.3 expected goals, would indicate that he's a still producing enough shots. It's far too early to worry about his finishing rate right now.

Herculean Summation

Herculez Gomez has been one of the most underrated American players to play the game over the last decade. I would actually go so far to as to say maybe the most underrated. Because he plays in the Mexican league rather than in Spain or Portugal or even the Eredivisie (where the goals are plentiful and the defense is bad), it feels like we sometimes compartmentalize what he does since it's on the wrong side of the ocean.

During the last four years in the Primeria, he has compiled 34 total goals (according to our records), and he is unofficially in the top 10 of goals scored during that period, tied with several others and most notably Raul Jimenez of Club America.

If he had accomplished this feat in England, Agentina or Brazil, there would be much more prestige awarded to him for his dues. However, sometimes we just look at Liga MX as only slightly better than MLS. It's fun to think what could have been if a better league had given him a chance.

Feel free to take a look at the data below and think about all that he's done. It's quite significant.

Season Shots OnTarget Goals xGoals G - xG
2010-11 38 21 10 5.65 4.35
2011-12 71 45 17 8.54 8.46
2012-13 64 31 13 8.60 4.40
2013-14 61 30 8 8.88 -0.88
2014-15 17 6 0 2.27 -2.27
Total 251 133 48 33.94 14.06
Season Goals (Prim) Goals (CCL) xGoals (Prim) xGoals (CCL) G/xG (Prim) G/xG (CCL)
2010-11 10 0 4.87 0.00 2.05 0.00
2011-12 11 6 6.14 2.40 1.79 2.50
2012-13 9 3 4.76 2.85 1.89 1.05
2013-14 4 3 6.74 1.07 0.59 2.80
2014-15 0 0 2.27 0.00 0.00 0.00
Ratio of goals to xGoals was used here due to the differences in quantity and quality of his opportunities between the Primeria and CCL.
Location Shots OnTarget Goals xGoals G - xG
Away 107 51 17 14.46 2.54
Home 144 82 31 19.48 11.52
Total 251 133 48 33.94 14.06
Shot Zone Shots OnTarget Goals xGoals G - xG
1 28 17 8 8.41 -0.41
2 113 68 29 20.19 8.81
3 44 23 7 3.05 3.95
4 26 11 2 1.36 0.64
5 39 13 2 0.90 1.10
6 1 1 0 0.04 -0.04
Total 251 133 48 33.94 14.06
Team Shots OnTarget Goals xGoals G - xG
Pachuca 32 19 5 4.19 0.81
Puebla 22 14 9 3.72 5.28
Santos 123 64 22 16.35 5.65
Tijuana 27 11 3 3.78 -0.78
UAG 29 18 7 3.35 3.65
USA 18 7 2 2.56 -0.56
Total 251 133 48 33.94 14.06


Our Playoff Chances Model is underselling FC Dallas

By Matthias Kullowatz (@MattyAnselmo)

Our on-site playoff chances finally gave FC Dallas a better-than-50-percent shot to make the playoffs after its away win over Chivas on Sunday. I say, "finally," because despite being in a playoff position much of the season, that particular playoff model hasn't been too convinced. 

It's important to understand how our playoff model works, and what its weaknesses are. It is based on overall shot ratios and finishing ratios, though at this point in the season it's the shot ratios that dominate the model's predictions. If you take a look at our MLS Tables, you'll see why the model isn't too keen on FC Dallas--its shot ratio of 0.84 shows that it only tallies 84% of the number of shots that it gives up to opponents. So despite being in fourth place by points per game, that model likes Vancouver and Colorado more because of their respective 1.03 and 1.30 shot ratios.

I don't yet have enough granular shot data to form a full playoff projection model using our Expected Goals 2.0, but we can still use that to intuitively tweak our expectations. Here's why FC Dallas is better than its 0.84 shot ratio.

Expected Goal Differentials (xGD) are more predictive of future success than simple shot ratios. At least, the 1.5 seasons of data we have say so. xGD takes into account not just shot quantity but also shot quality, based on the shot's origin and which body part was used. Based on 2013 and 2014 data--when controlled for number of home games--just eight games of xGD information predicts the following eight games of actual goal differential with a linear correlation coefficient of 0.33. Bump that up to 17 games of xGD information, and one could predict the following 17 games of actual goal differential with a linear correlation coefficient of 0.89 (based on 2013 data only). xGD is strong stuff.

FC Dallas' current xGD of +0.05 ranks fourth in the Western Conference, better than its seventh-place ranking in shot ratios. That's good news for Hoops fans, but I can get an even better idea of how they'll play if I break down xGD further into some specific gamestates.

Watching your favorite MLS team on the road while it's tied or ahead is a nerve-wracking experience. Away teams sitting on points often hang back and allow the home team to suffocate them, hoping to bend but not break. These scenarios are not exactly indicative of the away team's ability. But home teams usually play how they want to, regardless of gamestate, and thus all of a team's past Expected Goals data from when it was at home is helpful for projection.

Based on 2013 and 2014 data, the two best Expected Goals statistics to use when projecting game winners 20 weeks into the season are the home team's past home xGD and the away team's past away xGD in -1 gamestates. The data tells me that the best time to really see an away team's ability is when it finds itself behind by a goal. Interestingly, most teams have played fewer than 300 minutes while down a goal on the road--only about three game's worth--and yet that data in combination with home xGD is more predictive than overall xGD. So far, anyway.

As an example of how these two distinct models think of FC Dallas, we need look no further than its next two games---home against Colorado and away against San Jose. The playoff model we use (based only on shot totals, remember) has them with 38% and 21% chances of winning each game, respectively, and an expected point total of 2.3. Compare that to the Expected Goals model utilizing scenario-specific data, which projects them to win with 51% and 58% probabilities, and an expected point total of 3.7.

It turns out, FC Dallas ranks pretty well in both the aforementioned categories, but it should be noted that FC Dallas has played the third-fewest minutes while trailing by one goal on the road. So its variance in that department is greater than that of the typical team. But while the model's estimates as well as the team's outputs are subject to a modest margin of error, there is little doubt these are important gamestates. I leave you with a sortable table for home performance (xGDhome) and away performance when down a goal (xGDaway(-1)).

Team xGDhome Minutes xGDaway(-1) Minutes
SEA 0.93 957 0.17 249
SKC 0.90 1056 1.45 115
LA 0.89 955 0.10 123
COL 0.73 1047 -0.24 195
NE 0.61 963 -0.52 315
TOR 0.61 963 0.92 141
VAN 0.52 966 -1.12 94
FCD 0.46 1067 1.83 103
NYRB 0.43 966 0.00 434
PHI 0.41 863 0.74 326
RSL 0.31 1065 0.02 126
CHI 0.22 1061 -0.92 78
HOU 0.12 964 -0.88 256
SJ 0.10 1150 -0.34 271
CLB 0.05 953 0.89 225
POR 0.04 1074 -0.35 162
DCU 0.02 1153 0.47 124
MTL -0.13 1054 0.02 298
CHV -0.36 868 0.51 162

Expected Goal Differentials are per 96 minutes of play. 

US Autopsy - World Cup 2014

By Matt Hartley (@Libero_Or_Death)

Well the transfer rumors coming off the back of the United States’ World Cup are ending in a depressingly familiar half-exciting, half-exasperating muddle. A steady flow of rumors about foreign suitors for Matt Besler ended with the revelation that he could choose between the damned (Fulham) and barely spared (Sunderland). Little wonder that being a one-club legend in Kansas City was more appealing.

We can still salivate over where DeAndre Yedlin might end up, and while that is a totally valid use of your day, he will be more of a project for clubs like Roma or Lyon than an immediate contender for playing time. Just because the US went further than England doesn’t make Yedlin better than Glen Johnson, does it? Anyways, a few interesting statistical tidbits:

Goalkeeper

Howard - sure, he made a record-setting 16 saves against Belgium, but his best was the recovery to save from Eder after he misjudged Nani’s shot. That kept the score at 1-0, allowing the US to take advantage of their best 90 minutes of soccer and get the result that would see the US out of the group.

The most incongruous stat for Howard was his distribution distance of 30 meters. This was the second shortest among teams that made it out of the group stages, but was that part of the US game plan? While Jozy Altidore’s absence affected the ability to play long, if Klinsmann had instructed his players to build from the back, it didn’t quite come off, as the United States was 8th out of the 16 second round teams in passes completed per game. Things broke down too quickly when the US had the ball, leading to a rather high amount of chances for the opposition.

Center Backs

Thankfully, the US centerbacks were pretty adept at protecting the castle. In examining how the centerbacks did, CBI (Clearances+Blocks+Interceptions) nicely conveys how busy our defenders were, and we’ll look at that stat in its per90 form.

Besler - I wrote a World Cup preview piece for Paste in which I posited that due to having the most secure spot on the backline, Besler would have to be the rock for the US. He finished with a very respectable 13.83 CBIp90, good for fourth in the tournament. In fact, finishing ahead of him was…

Omar Gonzalez, emerging from what seemed to be a long-term demotion to rack up a  15.07 CBIp90 rating, coming from an outstanding 12.14 clearances per 90. The US was certainly relying on Omar to dominate as they conceded the flanks and allowed crosses to rain in.

The third primary center back for the US, Geoff Cameron, was 11th overall for CBIp90 with 12.60. Spending time in midfield as well, Cameron is well on his way to using that versatility to become the American Phil Neville.

Main thing to touch on:

Looking at the top 20 defenders by the CBI metric, there aren’t a lot of big names there. Medel has a good background, Vlaar at Villa, Cameron at Stoke, Nigeria’s Omueruo is on the books at Chelsea, and a couple of guys in Ligue 1. Hell, there are four current MLSers in the top 20 CBIper90 rankings. If the US really wants players to move to “big” clubs, then the national team will need to start producing more performances that aren’t backs-to-the-wall, man-the-pumps nonsense. Matt Besler had a really damn solid World Cup, and his options were the 14-20 slots from the Premier League. It’s certainly a chicken and egg situation, but it makes you hope that Juventus will come in for Erik Palmer Brown so that we can see some US players grow into regular slots at teams that seriously compete for the Champions League.

Fullbacks

This can be the hardest position to judge in the game, I think. You’ve got to be all things to all people at fullback, and that can make the position difficult to analyze. For the US there seemed to be a fairly clear hierarchy going into the tournament:

    1. Fabian Johnson, the best player for all 10 outfield positions

    2. DaMarcus Beasley, well, we like him better than Chandler

    3. Timothy Chandler, the source for a million overstated concerns about   German-Americans’ Americaness

    4. DeAndre Yedlin, there for the experience.

Of course, Beasley played solid two-way ball, Johnson was a useful offensive tool while on the field, and Yedlin became one of the breakout players of the tourney. Since the United States played a very narrow midfield for large swaths of the tournament, offensive contributions from the fullbacks were always going to be vital to our success. Looking at key passes, Fabian Johnson ended up with a .90 KPp90, which was 36th among defenders, placing him behind such noted playmakers at Vincent Kompany, and oh holy crap, DeAndre Yedlin.

That’s right, our little roadrunner, with his limited minutes, contributed a very nice 2.27 KPp90, good for fourth among Squawka’s defenders, and that’s right, one place ahead of Glen Johnson. Sign him up, Brendan!

Midfield

This was the part of the field where the United States’ struggles seemed rather stark. The US ended up with 326.5 completed passes per game, which put them smack in the middle of the 32 team field, and above Brazil, Costa Rica, and Colombia. But looking at things a little more closely, the United States played in its own half 34% of the time, more than any other country, and 22% in the opposition’s half, fourth worst in the entire tournament.

Looking at individuals, Michael Bradley came in for a lot of criticism, but despite playing mostly in a new role further up the field, he managed to complete more passes per 90 (47.77) than Luka Modric (46.00), Sami Khedira (45.36), and Steven Gerrard (44.09). Sadly, this involvement didn’t translate into chance creation, as Bradley finished with 0.67 KPp90, somewhere in the 139-160 range overall. Sure, there’s where Ronaldo finished, but so did Gary Cahill.

Jermaine Jones did everything, winning 65% of his aerial duels, 54% of his take ons, and running a very competitive race for the USMNT’s “Holy crap, I can’t believe that went in” award.  Alejandro Bedoya and Brad Davis weren’t statistically significant, while Kyle Beckerman finished 14th among midfielders in the CBIp90 metric. Graham Zusi provided two assists, but otherwise seemed very forgettable. There just wasn’t a lot to hang our hat on offensively.

In Closing

The United States failed to make the transition to a more progressive style of play this World Cup, but the US did show that they can defend fairly well. Klinsmann’s challenge will be to integrate more players comfortable with keeping and moving the ball through midfield to ally with a decent defense and a serviceable striker corps. There’s a lot of potential in the pool to meld into a strong corps for Russia 2018. I’d expect Fabian Johnson to become a full-time midfielder in the future, and see extended run-outs given to players like Julian Green, Joe Gyau, and Will Trapp. Future columns will look at the players who are making a strong case for the national team, starting with September’s friendly in Prague against the Czech Republic

New England's roller coaster ride

By Matthias Kullowatz (@MattyAnselmo)

This site purports to be one that covers all of American soccer--the United States part of America anyway--but outside of my obsession with Sporting Kansas City, Harrison's love for Federico Higuain, and Drew's commitment to DC United, we don't cover the Eastern Conference as much. So let's talk about New England.

Currently, our playoff projections have the Revs at 60-percent chances of making the MLS playoffs, and our Expected Goals data shows they are just about league average. Considering New England sits in fourth place in the East, and the teams chasing it for playoff spots are the Columbus Crew and the recently humiliated Chicago Fire, 60 percent makes a lot of sense. The New England Revolution is probably a playoff team despite its recent skid.

As fans, it's hard not to get caught up in streaks. After all, streaks affect our team's chances of making the playoffs, and our criticism tends to follow those streaks of the losing variety. Typically, unless a major injury or other personnel change occurs, the team is not getting any better or worse relative to the rest of the league. Shit just happens. New England has lost seven consecutive games. But before that, it completed a seven-game undefeated run, tying just one of those. Which team is New England? The data tells an interesting story.

Period Final 3rd Pass% Final 3rd Ratio GD xGD xGDzero
Hot Streak 0.610 1.14 1.86 0.11 0.26
Cold Streak 0.680 1.56 -2.00 -0.05 -1.53
Season 0.630 1.28 -0.37 -0.02 -0.14

New England has gotten hammered in even gamestates over the past seven games. Even gamestates are perhaps the fastest-stabilizing of the Expected Goals statistics, so though this is a small sample, it still suggests that the Revs are not simply getting unlucky. Luck is likely a factor, but not the only factor. And I think the only reason the overall xGD wasn't so different (0.11 versus -0.05) is that New England played so often from behind that its opponents were willing to give up more shots for longer periods of time. 

While New England actually completed a higher percentage of passes in the final third during its cold streak, and spent more time than its opponents in the attacking third, that could probably also be explained by opposing teams being ahead and willing to allow shots and possession.

New England didn't play nearly as well as it looked during its hot streak, and it probably didn't actually play as bad as it looked during its cold streak. But the recent cold streak still shows scary and somewhat-sustainable signs that the Revs aren't as good as our 60-percent playoff chances say they are.