This started simple. On Monday I wrote
I wanted to check if the underlying stats confirmed that. But the project grew. A lot. I give summary conclusions at a couple of points, but what's the fun of just reading that?
So, first I did confirm that NYCFC's xG figures have gotten worse both on offense and defense. The first 15 games run through the July 30 4-1 win over Columbus.
View attachment 11430
Also the expected points per game went down:
First 15 Games 2.08 xPPG
Recent 8 Games 1.60 xPPG
But then I noticed that the first 15 games included 8H-7A, and the recent 8 was 3H-5A. In small samples that's enough to maybe be significant so I dug further.
View attachment 11431
EVen adjusting for the H/S difference, you can see that everything got worse, with the bigger changes coming on defense both Home and Away. Similarly for xPoints:
View attachment 11432
Then I turned to trying to figure out what is driving the poorer play, and after consideration came up with the following theories:
- Random fluctuations in play that will just happen
- The schedule was harder in the Recent 8 Games
- A change in tactics, formation, and/or a learning curve incorporating our new players
- The team is fatigued due to schedule congestion and/or poor substitution and rotation patterns
- I received a new Mandalorian coffee mug for my birthday the last week of July and have been using it instead of my NYCFC mug for daily coffee since that date, and that was a lucky mug.
I also looked at the game rosters for any patterns, or missing players who really stood out. Nothing really seemed to jump out. There were no absences, or weird rotations, that were not roughly similar to things that happened in the First 15 Games. The poorer play coincided with the return of Sands and Johnson after missing several games due to the Gold Cup, but I don't think anyone believes their return could be the cause.
Random Fluctuations
I put this first, even though it is unprovable and unfalsifiable, because it should always be considered as the reason for anything that happens over small sample sizes, and both 15 games and 8 games are
very small sample sizes. Yet it is often the very last thing that fans consider because everyone really wants to point at something or someone to get blame or credit for a change in fortunes and explain things. Yet random chance and fluctuations is more often than not a big part of it, even when a more substantive reason is also in play.
Schedule Difficulty
The numbers are simple. The schedule as measured by H/A Adjusted Opposing PPG was harder recently:
First 15 Games 1.33 xPPG
Recent 8 Games 1.58 xPPG
That's pretty substantial, and it probably is a part of what drove this change. But, in the most recent 8 games, NYCFC also did worse against the poorer teams. For example, NYC played 7 games in the First 15 against team with a PPG of 1.00 or lower and went 5-1-1. NYC played 2 such games in the Recent 8 and drew both. In contrast, NYC won against teams with 1.36 and 1.85 in the more recent 8. In fact the win at Home against the Revs who sport a 1.85 Away PPG is - by that math - NYCFC's most difficult win this season. Yes the Revs were missing some players, but if you start trying to adjust for that, including when NYCFC wins despite missing players, your brain breaks and it really adds more noise than heat to the data overall. I definitely think schedule was a factor, but I don't think it explains everything.
And by the way, here is a list of teams whose
entire schedule to date has been easier than the "easy" part of NYCFC's schedule to date:
New England, SKC, Seattle, Colorado, Orlando, Nashville, DC United, and Miami. That list includes almost every very good team in the league. NYCFC's schedule has been very, very tough.
Changes in Tactics, Etc
I include this, but it's not really my thing. I notice some but not all such changes, and track none of it. There are people on the board who do remember when we switched formations and similar things and feel free to weigh in. I cannot really say if this is a potential reason. I'm
very interested in good theories about why our defense has gotten so much worse. Opponents xGoals is largely independent of goalkeeping, so GK play is unlikely to be a factor.
Fatigue, Including Schedule Congestion, Substitution and Rotation Patterns
Schedule
Superficially the schedule seems like a logical explanation. NYCFC played a normal schedule for 2-3 months, then suddenly had double games almost every week, with occasional random short breaks for international play. But (1) every team in MLS had a similar schedule, and (2) roughly half the teams in the league have played 1 or 2
more games than NYCFC. NYCFC players might be tired, but there is no reason why the schedule particularly should be a cause that affects NYCFC more than anyone else.
Subs and Rotation
A favorite forum topic. I dug deeper.
First, an aside. Taty Castellanos has played more minutes than anyone on the team and his play objectively improved in the Recent 8 Games. Of course, he can be a weird outlier/exception and seems exceptionally fit but it's still worth mentioning.
Second, I checked the Games Played and Minutes Played stats with the most minutes on each team. I looked at both the Top 7 and Top 10 field players on the following teams in addition to NYCFC:New England, Nashville, Orlando, Seattle and SKC.
I figured the top 10 players with the most minutes covers the usual starting 10 while the top 7 covers the really most core players who get the least rotation. Of course, every team deals with absences due to international play, injuries, and discipline, which are not exactly evenly balanced, but I think we have enough teams to even out those differences. I looked at Minutes Played per Team Games, Minutes Played per Player games, and Total games Played per Player. I did not use Total Minutes Played because the difference between playing 22-25 games swamps all other factors unless you look at averages. Also, I used ASA data, which is easier to work with. ASA minutes played are different from the official MLS figures for Minutes Played. Basically, ASA adds stoppage time, so, for example, Revs CB Andrew Farrell has played 99 minutes per game on average (I
think they use 9 minutes per game as a standard covering both halves). It looks weird but it is the same for every team so it evens out.
View attachment 11434
Only Orlando plays its starters for significantly fewer minutes per game than NYCFC does. A few extremely successful teams play them more.
The idea that Ronnie Deila is an outlier on making subs or rotating players compared to other coaches just needs to die. There is no evidence for it. All you have left if you want to keep banging that drum is to claim either that Oscar Pareja knows something that the head coach of every other top MLS team does not, or look up the sub patterns of the mediocre to bad teams and try to build an argument out of that, and then exp hand waving.
So does this completely eliminate fatigue as a cause? No, because NYC's players might just be less fit. But I don't know how to measure that, and there is not even anyone who has seen enough of every team to make reasonable subjective assessments.
Lucky Coffee Mug Theory
I know of no evidence that contradicts it. I stopped using my NYCFC coffee mug at almost exactly the same time the club started playing worse. On the other hand, I was due for a new mug, and I'm not giving it up. I mean, look at it.
View attachment 11436
Conclusion re Causality
Unclear, but probably some combination of random fluctuation, and a tougher schedule. Can't rule out the possibility it is all my fault because I stopped using my lucky NYCFC mug. If anyone has thought on tactics or formation changes around the operative time please chime in.
Performance Below Expectations
One remaining question is whether NYCFC's results in recent games are still below what one would expect from the underlying stats or has that also changed? It depends. Mostly yes, but not really in the way many people think about it:
Looking at Goals minus xG:
First 15 Games: -0.21 per Game
Recent 8 Games: -0.19 per Game
So that improved, a wee bit. But scoring more or less than your xGoals is only part of it, and actually a very small part. Matt Doyle has repeatedly said and written that for NYCFC's results to start matching their stats, Taty Castellanos would have to start finishing to match his xGoal production. Lots of people agreed. He was completely wrong. Taty went on a scoring binge and was MLS player of the Month in August and he has scored 5 goals on 3.83 xGoals and NYCFC did worse.
What really counts, where the giant variances really kick in, is whether your results are in line with your xGoal Differential or your actual Goal Differential.
First, let's look at Expected Points:
View attachment 11441
NYCFC's record on this got worse in recent games, even as its xPts went down. But now comes the real fun, basic stuff. We'll put advanced stats completely to the side and measure real Points compared to real Goal Differential.
View attachment 11438
This has nothing to do with advanced stats. This is pure Goals and Points, and it swamps the effect of NYC's xG underperformance. If you are annoyed that NYCFC is not performing in line with the statistics, your problem is not advanced statistics, it is this. You don't like or trust advanced stats like xG and xPts? Ignore them.
Portland has just 1 less point than NYCFC, despite a negative Goal Differential that is 21 goals worse than NYCFC. The Galaxy has more points than NYCFC with a net GD of zero. Orlando also is off the charts successful compared to actual scoring, with 38 points and a measly GD of +2 . Only Nashville among the Top 10 teams in MLS is doing worse on this metric than NYCFC. This is basically the issue we discussed a few days ago: do your goals scored and goals allowed come at opportune times, or not? Do you waste goals by winning 5-0 or 4-1, and losing 2-1, or do you win 1-0 or 2-1, and lose 5-0? Are you efficient?
Then the debate switches to whether this is due to luck, will, grit, or some knowledge of "how to win" possessed by the coach or players. The sport where this is most studied is baseball (look up Pythagorean formula). The consensus opinion among those who study this is that luck is the main factor. Every year some baseball teams routinely have very "lucky" or "unlucky" results.
Right now the Seattle Mariners have 12 more real wins than expected wins (all based on Runs Scored, Runs Allowed, and the Differential), while most teams are within +/- 3 and almost all within +/- 6. There have been scattered results that are significantly more extreme than Seattle's +12 in the past. Generally teams who do especially well or poor one year, tend to regress towards the mean the next. There are occasional exceptions and this does engender some debate. But mostly, people write thousands of words discussing bullpens, clutch hits, strategy, and veteran experience, to explain why a big differential between wins and expected wins for their favorite team is due to some perceivable and earned factor, but in most cases they are rendered moot the next year when the same team and same manager and largely the same roster ends up with a differential of +/- 3 or lower.
Baseball is a game with a 162 game season and if your intuition is that good or bad luck in soccer cannot possibly last a full season of 34 games or even just 23 games, and therefore your argument based on [____] must be correct, you might want to think on that some more because it surely does cover entire 162 game seasons in baseball. That does not mean whatever you see is not real. But it is extremely likely that you overweigh it because we are wired to generate stories to make sense of what we see, and not to ascribe it to random chance. Also, if focus you too closely on the results obtained by teams like Portland you start to think the entire enterprise is pointless, and that's no fun.
Wrapping Up This Shaggy Dog
NYCFC has played objectively worse the last 8 games than it did in the first 15. There's no hard evidence that fatigue or sub patterns have anything to do with it, but it could be related to the schedule or normal fluctuations, or my coffee cup. NYCFC was very unlucky the first 15 games, and even less lucky recently. That does not mean Ronny has not made poor decisions that cost us games. Maybe he has! But even if he did, NYCFC has still been pretty damn unlucky (and Nashville even more so) #GarySmithOut!