538 now has MLS SPI projections available

Only Chicago and MNU have lower defensive ratings? Not sure how much faith you can put into that. We will not give up as many goals as last season for many reasons.
-SJ is better than Saunders all around.
-Our full backline are together at the start of the season instead of midseason moves. Most were with the club last season as well. This should result in a more cohesive unit with fewer breakdowns.
-We should not have any 7 goal losses this year.
 
They have so little data on those teams they probably overweighted home field advantage.

i know its strictly numbers so i guess...but still i felt after week one it should of been closer than that.
 
I'm glad they are doing this. I like Silver and 538. I think they will have trouble with early season predictions because of major turnover in teams like us and Chicago and many others. By June or July I will give their end of season projections considerably more weight.

Even with that, it seems odd that NYC is ranked 3rd in the East yet 9th overall. I don't see how there are 6 teams in the West with a better chance of winning MLS cup than the 3rd team in the East, regardless of who that is.

Also odd that NYC's expected Goal Differential is +4 when the current GD is +3. plenty of other weird things to tease out but, really, it almost seems unfair to pick them apart so early in the season.
 
https://fivethirtyeight.com/features/how-our-club-soccer-projections-work/

I initially assumed the model was running off just the first two games of this season but it seems that they are using some historical data in the calculations as well. Until we get a few more games in this is basically a projection of how last years team would perform. It also means the Atlanta and MNU projections are operating with a much smaller sample size than the rest of the teams and will likely be meaningless for an even longer period of time. I read it quickly but I didn't see any explanation of how they are applying weighting factors to make more recent performances account more towards the model outcomes. It will be interesting to see how much this weeks outcomes start moving the numbers.
 
I'm glad they are doing this. I like Silver and 538. I think they will have trouble with early season predictions because of major turnover in teams like us and Chicago and many others. By June or July I will give their end of season projections considerably more weight.

Even with that, it seems odd that NYC is ranked 3rd in the East yet 9th overall. I don't see how there are 6 teams in the West with a better chance of winning MLS cup than the 3rd team in the East, regardless of who that is.

Also odd that NYC's expected Goal Differential is +4 when the current GD is +3. plenty of other weird things to tease out but, really, it almost seems unfair to pick them apart so early in the season.
I like 538 too, but maybe they should have waited to debut their models until more matches are played, because right now I can't take them seriously if they are projecting Minnesota with a 25% chance to make the playoffs. Unless there is a total roster overhaul/addition of DPs, they aren't even close to a 5% chance.
 
I saw this a few days ago and concluded it's largely garbage numbers, at least for now. The model actually adjusts teams during the offseason based on their net transfer fees. That might work to some extent for a league like the EPL, but for MLS, no way. It would likely be more accurate to assume no offseason changes.
 
https://fivethirtyeight.com/features/how-our-club-soccer-projections-work/

I initially assumed the model was running off just the first two games of this season but it seems that they are using some historical data in the calculations as well. Until we get a few more games in this is basically a projection of how last years team would perform. It also means the Atlanta and MNU projections are operating with a much smaller sample size than the rest of the teams and will likely be meaningless for an even longer period of time. I read it quickly but I didn't see any explanation of how they are applying weighting factors to make more recent performances account more towards the model outcomes. It will be interesting to see how much this weeks outcomes start moving the numbers.

Regarding off-season changes, it mentions this...

As with our Elo-based rating systems, each team’s ratings change in the offseason. Rather than reverting each team toward the same mean, we revert it toward a time-weighted average of its final rating over the past five seasons. In addition, we adjust each team’s preseason rating based on players it acquires or sells in the offseason.

Specifically, these adjustments are based on subtracting transfer fees a team got in the offseason from how much it spent on acquiring players, relative to league average. For every standard deviation of net spend above league average, a team’s rating is boosted by about 0.09 points, split evenly between the team’s offensive and defensive ratings.
 
Regarding off-season changes, it mentions this...

As with our Elo-based rating systems, each team’s ratings change in the offseason. Rather than reverting each team toward the same mean, we revert it toward a time-weighted average of its final rating over the past five seasons. In addition, we adjust each team’s preseason rating based on players it acquires or sells in the offseason.

Specifically, these adjustments are based on subtracting transfer fees a team got in the offseason from how much it spent on acquiring players, relative to league average. For every standard deviation of net spend above league average, a team’s rating is boosted by about 0.09 points, split evenly between the team’s offensive and defensive ratings.
So does it factor in at all when guys are brought in on a free?

For example, if I am understanding this correctly, and please just let this hypothetical go, say if we were not an expansion team a couple of years ago, and we brought in Villa on a free such as we did, would his addition, in essence, not "help us" within this model?
 
So does it factor in at all when guys are brought in on a free?

For example, if I am understanding this correctly, and please just let this hypothetical go, say if we were not an expansion team a couple of years ago, and we brought in Villa on a free such as we did, would his addition, in essence, not "help us" within this model?

Not to mention the fact that MLS doesn't always report fees. Maxi Moralez probably counted for zero.
 
That was super interesting. And that closing paragraph probably belongs in the "sell TMac" thread.
As someone who didn't rely on stats to evaluate him, if the numbers said ol' Jebaroni sucked, they were pretty damn accurate.

But yeah...
 
As someone who didn't rely on stats to evaluate him, if the numbers said ol' Jebaroni sucked, they were pretty damn accurate.

But yeah...
I seem to recall that at some point late in the 2015 season there were "stats" that supported an outrageous argument that Brovsky was the best LB or RB (don't recall which and frankly don't care) in the US player pool.
 
Proper, tangible measurements of a player’s ‘soft’ skills are currently nonexistent. Passion, resiliency, leadership, communication and awareness are all distinct traits of some of the best players in the world and not merely cliches used to sell sports drinks. Looking at a spreadsheet or heat map on Twitter can’t adequately represent the center midfielder who influenced the game by shutting down the opposition’s best player while simultaneously inspiring his team to victory. The numbers also won’t show what goes on inside the locker room, what type of character the individual player has, how the player responds to a mistake or poor performance the previous week, and also the leadership he brings to the table. These are concepts that must be observed using the sensitivity of the human eye, not a camera lens. Soccer is a particular experience that can’t be fully appreciated through computation alone.

Traditionalists: "You can't measure soft skills."

Statisticians: "Prove it matters."

I don't know a ton about soccer metrics, but in baseball, the "soft skills" argument was similar to the "god of the gaps." Soft skills were attributed to every contribution that wasn't a hit, run, or RBI. Until those other contributions started getting measured.

Ask any athlete if being "in the zone" is a real thing, and they'll nod emphatically. But then we measured it. Not really a thing.
 
Bringing this thread back to life --

I'm always amazed at some of the weird results that come up in the 538 forecasts.

For example, right now NYCFC is given a greater chance than NYRB to make the playoffs and a greater chance to have a first-round bye than NYRB, but somehow a lower percentage chance of winning the Supporter's Shield.

So somehow, they think we have a better chance of being 1 or 2 in the East, but somehow NYRB has a better chance of being first overall?
 
  • Like
Reactions: SoupInNYC