2019 By The Numbers: There's A City In My Mind, Come Along

[mgarbowski]

I skipped the Week 15 update, even though NYCFC had a game. Apart from that it was a week of few games, followed by a week off for the entire league, then followed again by a limited schedule week in which the only active Eastern Conference teams are behind NYCFC on the table. Squeezed in there were 2 US Open Cup Rounds, representing NYCFC’s 1st and 2nd USOC wins ever, but those don’t really play into the weekly updates, so there still is not much to say. So even now I’m going to forego the commentary and just put up the charts and graphs.

 

[adam]

I skipped the Week 15 update, even though NYCFC had a game. Apart from that it was a week of few games, followed by a week off for the entire league, then followed again by a limited schedule week in which the only active Eastern Conference teams are behind NYCFC on the table. Squeezed in there were 2 US Open Cup Rounds, representing NYCFC’s 1st and 2nd USOC wins ever, but those don’t really play into the weekly updates, so there still is not much to say. So even now I’m going to forego the commentary and just put up the charts and graphs.

View attachment 9918 View attachment 9919 View attachment 9920 View attachment 9921

For a team that only has 1 loss all season, why is the minimum prediction going forward 5 wins?

Where do we have to be in the standings before we start to compare to 17 & 18?

 

[mgarbowski]

For a team that only has 1 loss all season, why is the minimum prediction going forward 5 wins?

Where do we have to be in the standings before we start to compare to 17 & 18?

What do we have to do to get through a summer without you jinxing us by asking me to revise my graphs to be more optimistic?

 

[413Blue]

For a team that only has 1 loss all season, why is the minimum prediction going forward 5 wins?

Where do we have to be in the standings before we start to compare to 17 & 18?

Well, the last two seasons have been very rough down the stretch for us.

 

[LionNYC]

For a team that only has 1 loss all season, why is the minimum prediction going forward 5 wins?

Where do we have to be in the standings before we start to compare to 17 & 18?

We have to up the ties on those prediction of total points.

 

[mgarbowski]

For a team that only has 1 loss all season, why is the minimum prediction going forward 5 wins?

Where do we have to be in the standings before we start to compare to 17 & 18?

We have to up the ties on those prediction of total points.

I was planning to reassess stuff after this next set of 3 games. That's halfway through the season and the H/A split will be as even as possible given an odd number of games. I have no benchmarks in advance; I'll just do what seems to make sense at the time.

 

[mgarbowski]

Sift through my soul to see what's lost and found
With this week's game against Philadelphia, NYCFC has now played every Eastern Conference team that was in the league as of 2015 at least 10 times in MLS regular season play. This shows NYCFC's Points Per Game against each of those teams, excluding playoffs, US Open Cup, or anything else except MLS regular season.

Here are some more factlets with respect to NYCFC against those teams:

NYCFC's Best Home Record against - Philadelphia 2.60 (next CHI 2.33)
Best Away - Montreal 2.60 (TFC 1.40)
Worst Home - New England 1.33 (MTL 1.5)
Worst Away - RB 0.67 (NER 0.80)

Odd that Montreal has the second best record at Yankee Stadium while having the worst record against NYCFC overall. The gap between NYCFC's best Away record and second best is massive. I knew NYCFC had trouble against New England but wow. A few more items of note with respect to this set of teams and MLS season play:

NYCFC has never lost at home against CHI, DC, PHI, TFC
NYCFC has at least one Away win against all of these teams (expanding the field NYC has yet to win at COL, FCD, LAFC, MNU, RSL & has yet to play at CIN)
NYCFC has never lost Away in Montreal (also COL, LAFC, SJ & again CIN)

Empty my pockets that were weighing me down
Sort the MLS tables by different variables and NYCFC is all over the place, but mostly somewhere good. The biggest factors are the games played and losses. NYCFC is:

Last in Games Played
Tenth in Total Points
Second in Points Per Game
Fourteenth(!) in Total Wins
First for Fewest Losses
First (Tied w/ Vancouver) for Draws (with DC, SKC and Chicago one back)

Of course Games Played is a factor but NYCFC's total amount of tied games no longer seems so ridiculous. Part of the league is is catching up.

Look at this pretty PPG table:


The Union's traveling supporters were singing "We're top of the table" Saturday night, which they are entitled to do. But part of that is a factor of games played. This chart helps show where teams are at similar schedule states:


To help further, I made a mini-table showing each team's points as of Game 15 where NYCFC sits now:


Ties are sorted alphabetically but Atlanta had more wins at 15 than NYCFC so this is fair anyway.

I can hear a howling wind
That sweeps away the pain that's been
2016 and 2015 are -- tentatively -- no longer looking like appropriate comps for 2019. I hesitate on this because this team has had poor second halves two years running.


Gonna walk away from trouble with my head held high
The top half of the potential outcomes chart is looking more attainable:

Tears don't care who cries them
I posted about this at the beginning of June after 33 games because I lacked the patience to wait 3-4 weeks for Torrent to coach his 34th game. But now we have a full season under Domé and his record is 14-8-12 for 54 points and 1.59 PPG. Patrick Vieira's first full season record, which corresponded exactly with the 2016 season, was 15-10-9 for 54 points and 1.59 PPG. I don't draw any huge conclusions from this except to note that it seemed awfully unlikely at the end of 2018 and even several games into this one. Torrent is helped considerably by his 5-1-0 start before he started tinkering much. Vieira took over a bad team and made it good. Torrent took over a strong team and after his excellent start earned just 21 points in 19 games before turning it around. But Vieira also took over between seasons while Domé had to adjust on the fly. Who knows, maybe his rough stretch will be like Tiger Woods early in his career. He had won his first major when he decided to hire a new swing coach and completely revamp his swing. He won nothing for a couple of years (roughly) but then the changes took hold and he won 7 Majors from 1999 to 2002. I'm not predicting that much hardware for NYCFC but some success seems possible, at least.

 

[Gotham Gator]

Great stuff, as usual.

One thing I've noticed over the years - and previously noted - is that our points-per-game typically bounces around for a while and then settles in at about the halfway point of the season and doesn't move much after that. In 2015 and 2016, our points-per-game in Week 18 roughly matched where we'd finish, and didn't vary much. Same is true in 2017 from about Week 16. The 2018 season stretches that narrative a little - for reasons already well discussed, we fell off over the second half, but we did end up just a little below where we were after Week 17, albeit with a little more variance than usual.

Right now, we are sitting on 15 games, with a points-per-game of 1.73, which is higher than we've ever had at the end of a season. After the next 2-3 games, we should be able to project where we will finish within about 2 points up or down - barring some kind of unusual change in circumstances.

 

[Gotham Gator]

It's also interesting to note just how crowded the top of the East is this season. The gap between first and fifth is only 0.14 points-per-game - which translates out to 5 total points at season's end. Montreal, in 6th, is just a hair behind that, only 0.09 points-per-game behind the 5th place Red Bulls. It's going to be a wild ride.

 

[The Toe]

Mark, I look forward to this post every Monday. Excellent stuff.

I'm starting to wonder how much of Dome's success comes from the plethora of individual talents on this team. We are -- by my metric of watching a ton of MLS -- quite easily the most or second-most talented team in the East. I'm not sure how many other teams could lose so many guys for international duty and continue on a 13-game unbeaten streak which has seen the team get better and better. Hats off to proper player management (so far, though the test comes this week and next) from Dome to this point. Guys are getting chances and making the most of them, and I'd like to think that comes from the gaffer.

 

[mgarbowski]

I meant to mention that NYCFC also has the second biggest W - L differential at +5. LAFC is +9

 

[mgarbowski]

Great stuff, as usual.

One thing I've noticed over the years - and previously noted - is that our points-per-game typically bounces around for a while and then settles in at about the halfway point of the season and doesn't move much after that. In 2015 and 2016, our points-per-game in Week 18 roughly matched where we'd finish, and didn't vary much. Same is true in 2017 from about Week 16. The 2018 season stretches that narrative a little - for reasons already well discussed, we fell off over the second half, but we did end up just a little below where we were after Week 17, albeit with a little more variance than usual.

Right now, we are sitting on 15 games, with a points-per-game of 1.73, which is higher than we've ever had at the end of a season. After the next 2-3 games, we should be able to project where we will finish within about 2 points up or down - barring some kind of unusual change in circumstances.

You're right, and my additional spin is that -- as pretty much anyone can figure out -- there are probably 2 main reasons for this. First, that as you keep adding games to the denominator of PPG, every subsequent individual game has a smaller effect than the one before it. It just gets harder and harder to move the needle. Second, every game played increases the sample size upon which we can predict future results, which increases the accuracy. But the real twist is that is that the second factor has its own built in weakness, for which I need to refer everyone back to the chart I took from dummyrun in the first post of this thread this year.

https://twitter.com/x/status/1115589304724664325

There are several lessons from that graph. One is that PPG is worse than almost every competing metric as a predictor until August, when it becomes the best. Another is that none of them are really very good as predictors. A third, and the one I'm focusing on now, is that that they all get worse starting in May, and take a sharper turn worse again around August. I'm pretty sure this is because small sample sizes work in both directions, and even as the sample size of past data gets bigger, the sample size of future results shrinks, and becomes more susceptible to unexpected intervening changes (like signing new players, injuries or coaching changes) or just random fluctuations.

My point is that the the fact that the PPG is not likely to change much once you get past the halfway point of the season is not the same as saying the PPG from any given point forward is likely to be similar to the PPG to date. That doesn't undermine Gator's point at all, but I hope clarifies exactly what it does and does not tell us.

 

[Ulrich]

I meant to mention that NYCFC also has the second biggest W - L differential at +5. LAFC is +9

How about W - T Differential?

 

[dummyrun]

they all get worse starting in May, and take a sharper turn worse again around August. I'm pretty sure this is because small sample sizes work in both directions, and even as the sample size of past data gets bigger, the sample size of future results shrinks

Yup. PPG is inherently noisy so once you get to June (those labels should say "at the end of") and have less than half the season left it gets harder to predict what's left.

 

[Shwafta]

Yup. PPG is inherently noisy so once you get to June (those labels should say "at the end of") and have less than half the season left it gets harder to predict what's left.

I mean, the best way I can think to solve the PPG noisiness is to just get a pair of noise-cancelling headphones,

 

[413Blue]

I mean, the best way I can think to solve the PPG noisiness is to just get a pair of noise-cancelling headphones,

Or to do something crazy, like not hav9ng a 4GP gap between teams!

 

[mgarbowski]

Only The Lonely Can Play

Where did everybody go?

Portland has played 3 games at Yankee Stadium. Portland has won all 3. They won all 3 games by a score of 1-0. But there is something else those games have in common.

• April 19, 2015. Villa was injured and did not play. Same for Mix.
• September 9, 2017. Villa was injured after Spanish national team duty and did not play.
• July 7, 2019. Six starters: Heber, Mitrita, Tajouri-Shradi, Sands, Callens, and Johnson (plus a few potential subs) were unavailable for either injury or international duty.

It is a wonder that NYCFC controlled play for most of the game and almost got a result. It was a shame to end the unbeaten streak with a loss under such circumstances, with more than half of a starting 11 unavailable, including almost the entire front line, and the remaining players reeling in the middle of a 5-games-in-16-days- stretch, but that is how things go.

It should surprise nobody that the 12 game unbeaten streak was the longest in club history. The previous best straddled the last game of 2017 and the first 7 of 2018 for 8 games unbeaten.The longest single season streak was the same first games of 2018.

Brad Stuver just played 6 games in the USOC and league play, most of them without Callens, Sands and Matarrita in the back on defense, and gave up 6 goals. I might have gladly taken 7 or even 8 in advance.

Losses with half your lineup missing still count, and the home loss to Portland put just a bit of a kink into what had been NYCFC’s steady ascent up the tables the last 2 months. Fortunately, just about everyone at the top of the PPG table in the East also dropped points this week.



Philadelphia has retaken the lead on the make-believe table where we compare every team at the number of games NYCFC has played to date:


The good news is the Union (and others) did not grab all the points beyond Game 17 so NYCFC has room to recover and surpass with its games in hand.

The "how to reach" table still looks pretty good:

NYCFC is halfway through its season and I reassessed the comparables for the line charts, choosing to use all of 2016, 2017 and 2018.



The team is just behind the 2017 and 2018 version at the 17-game mark. If it can avoid the extended second half slump of those version it can certainly be the best NYCFC ever.

The remaining schedule might be an advantage, even with one more road game than home remaining. Here are the Home/Away adjusted opponents PPG for the remaining schedules for the top 7 East Conference teams:

Red Bulls 1.51
Philadelphia 1.47
Atlanta 1.47
DC 1.45
Toronto 1.27
NYCFC 1.27
Montreal 1.19

For more detail I counted how many games each of those teams has against an opponent with a H/A adjusted PPG of 1.80 or higher with the following results: RB 3, PHI 5, ATL 4, DC 5, TFC 2, NYCFC 5, MTL 2. That shows NYC has as many or more somewhat arbitrarily tough games remaining as anyone,* despite an average toughness that looks quite easy. As an example, the Red Bulls remaining opponents PPG is so high largely because they still have to play both LAFC (2.78 PPG at home) and Seattle (2.56 PPG at home). But no matter how hard those games are the Red Bulls can at worst suffer 2 losses in them, and they only have 1 other game against a team at 1.80 or higher (Portland 2.0 Home). NYCFC has 5 such games, which is arguably tougher overall even though none of them should be as tough (RB 1.90, RSL 2.25, ATL 2.10, Dallas 2.0, PHI 1.91).

* I picked 1.80 because it generated what seemed to be the right number of results. 1.90 and 2.0 left everyone with hardly any difficult games and 1.70 or lower too many.

 

[mgarbowski]

Communication Breakdown

Corner-gate, throw-in-gate, call it what you want. It happened and the result is that NYCFC lost 3 games in 8 days (2 in league play). Players were missing, the schedule was tough, there was fatigue. But you take the games as they come. Before NYCFC lost to Portland while missing half the starting lineup, NYCFC beat Seattle in a similar state.

Here is where things stand.




Here is how things stood on that mythical day when every East team played exactly 18 games:



With 2 straight losses, NYC drops into a tie for third, three full points behind Philadelphia. The good news is that looking ahead/behind Philly earned/will get just 4 points in the next 4 games, DC got/will get 2 in the next 3, and Atlanta 1 in the next 2 that happened/will happen. Re Philly, it is worth pointing out that they just finished (in actual real time) a tough 5-game stretch with only 1 home game, earning just 5 points out of 15. This brought their H/A split to even after 22 games. Looking forward their schedule is kind of average, but one should not expect their current slump to continue as bad as it has been.

NYCFC’s future schedule is, on average easier, but splits into 3 phases. First, the Pigeons face 2 Away games against teams earning 2.0 PPG at home or better in the next 5 games for an average opponents H/A PPG of 1.42. Then comes a 7 game stretch of which 5 are at home, with a opposing PPG of 0.98. Then the finish sees 3 of the final 4 Away, with 2 hosts getting 1.9 PPG or better, and an average of 1.55. That middle stretch roughly coincides with NYC’s second half slumps in 2017 and 2018, so they are set up to avoid repeating that in 2019.

More immediately, by PPG, Colorado is one of the team’s 3 easiest remaining Away games.

Additional future info in the "how do we get where" table.


Finally, Zeppelin took the post title but Katy Perry gets the last word on Alan Kelly and Cory Rockwell:

“You change your mind like a girl changes clothes
* * *
You’re hot then you’re cold
You’re yes then you’re no
You’re in then you’re out
You’re up then you’re down
You’re wrong when it’s right”

 

[Ulrich]

Communication Breakdown

Corner-gate, throw-in-gate, call it what you want. It happened and the result is that NYCFC lost 3 games in 8 days (2 in league play). Players were missing, the schedule was tough, there was fatigue. But you take the games as they come. Before NYCFC lost to Portland while missing half the starting lineup, NYCFC beat Seattle in a similar state.

Here is where things stand.
View attachment 9998

View attachment 9999
View attachment 10000
Here is how things stood on that mythical day when every East team played exactly 18 games:


View attachment 10001
With 2 straight losses, NYC drops into a tie for third, three full points behind Philadelphia. The good news is that looking ahead/behind Philly earned/will get just 4 points in the next 4 games, DC got/will get 2 in the next 3, and Atlanta 1 in the next 2 that happened/will happen. Re Philly, it is worth pointing out that they just finished (in actual real time) a tough 5-game stretch with only 1 home game, earning just 5 points out of 15. This brought their H/A split to even after 22 games. Looking forward their schedule is kind of average, but one should not expect their current slump to continue as bad as it has been.

NYCFC’s future schedule is, on average easier, but splits into 3 phases. First, the Pigeons face 2 Away games against teams earning 2.0 PPG at home or better in the next 5 games for an average opponents H/A PPG of 1.42. Then comes a 7 game stretch of which 5 are at home, with a opposing PPG of 0.98. Then the finish sees 3 of the final 4 Away, with 2 hosts getting 1.9 PPG or better, and an average of 1.55. That middle stretch roughly coincides with NYC’s second half slumps in 2017 and 2018, so they are set up to avoid repeating that in 2019.

More immediately, by PPG, Colorado is one of the team’s 3 easiest remaining Away games.

Additional future info in the "how do we get where" table.
View attachment 9998

Finally, Zeppelin took the post title but Katy Perry gets the last word on Alan Kelly and Cory Rockwell:

“You change your mind like a girl changes clothes
* * *
You’re hot then you’re cold
You’re yes then you’re no
You’re in then you’re out
You’re up then you’re down
You’re wrong when it’s right”

Listening to your daughter’s music collection for those lyrics?

 

[SoupInNYC]

Listening to your daughter’s music collection for those lyrics?

I don't know, with references to both Zeppelin and Katy Perry, it's possible he's hacked into my Spotify.