Lightnews — Scholar-powered news

Jamon M - MLS and footy analysis @jamonm.bsky.social · 9h

It might take heading towards Morgan Hill, but you should be able to get some at the pumpkin patches.

Reposted by Jamon M - MLS and footy analysis

Tutul Rahman @tutul.bsky.social · 2d

The MLS record for non-penalty goals in a single season is 27 (Vela, 2019).

Messi is at 26, going into Decision Day.

1 1 19

Reposted by Jamon M - MLS and footy analysis

g+ GameFlow @gameflow.bsky.social · 3d

Note: GameFlows are paused until I get back from vacation next weekend. Until then, @mlsstat.bsky.social, @nwslstat.bsky.social, and @uslstat.bsky.social should have you covered.

-Eliot

1 3

Jamon M - MLS and footy analysis @jamonm.bsky.social · 3d

In the end, because Save % shows more stability as a metric, shows career Save % and season-over-season change are valuable tools for evaluations in addition to the PSxG metrics, particularly if G-PSxG is highly volatile. Comparing the YOY Save %'s can tell you a lot.

More to come.

/end

1

Jamon M - MLS and footy analysis @jamonm.bsky.social · 3d

(The R-squareds are a very low 0.04 to 0.06 on any of these, so it just doesn't matter a whole lot.)

Plus, just try bringing up G/PSxG in conversation with even an MLS sicko and explaining why it shows Bond has had a worse 2025 than Pulskamp, even though Pulskamp's G-PSxG is worse by two goals.

2/

1

Jamon M - MLS and footy analysis @jamonm.bsky.social · 3d

FSD follow-up: some questions on Save % came up, so I dug around more.

First, YOY Save % is more stable than even G / PSxG. (see viz)

Second, Save % predicts future Save % better than G-PSxG (SAE) and G/PSxG predict themselves, but not quite better than G-PSxG at predicting itself.

1/

1 1 2

Jamon M - MLS and footy analysis @jamonm.bsky.social · 3d

Was there. Can confirm.

Arielle Dror @arielledror.bsky.social · 3d

We played a whole game of soccer across the street and the Mariners are STILL playing

4

Reposted by Jamon M - MLS and footy analysis

Catalina @catabush.com · 4d

1 25

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

This is a very short thread for a very tricky topic, but if you are a club looking at GKs, you might want to consider YOY save % as much or moreso than just GA-PSxG, because it's generally more stable.

That's all I'm saying (for now).

1 1

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

That seems right to me.

2

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

There's a lot of fog on this iceberg. I'll say this for now: The y-axis (SAE / GA-PSxG SD) ranges from about 1 to 8, while the x-axis (save % SD) only ranges from about 0.02 to 0.08. Even accounting for different scales, SAE shows much larger swings YOY.

bsky.app/profile/jamo...

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

It's obvious from the first two visualizations in 🧵 that goals against has a better correlation to PSxG against than save % in a single season. From that POV, the data analysts look good.

The issue is that PSxG is much less stable than save % for keepers season-over-season as this shows.

3/

1

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

I've got a lot more where this came from. But I'm my own bottleneck to getting it out there.

2 1

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

What this tells us at a simple level is historical Save % for goalkeepers is almost certainly a better indicator of future GA-PSxG performance than past GA-PSxG performance is.

If a keeper is in one tier today, what will his tier be next season? Viz below.

Enjoy the games.

/end

3 8

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

The five tiers are based on these season save %:

* Elite: ≥ 0.78
* Above Average: 0.74-0.78
* Average: 0.70-0.74
* Below Average: 0.65-0.70
* Poor: < 0.65

While I've demonstrated GA-PSxG is not stable YOY, it does move within a range with these tiers.

5/

1 5

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

For this analysis, I am looking at keepers in the Big 5 European Leagues and MLS in select seasons who have played at least 5 seasons and have 50+ saves and 1,200+ minutes per season.

Through trial-and-error, I put each keeper season into 5 tiers: Elite, Above Avg, Avg, Below Avg, and Poor.

4/

1 3

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

It's obvious from the first two visualizations in 🧵 that goals against has a better correlation to PSxG against than save % in a single season. From that POV, the data analysts look good.

The issue is that PSxG is much less stable than save % for keepers season-over-season as this shows.

3/

2 4

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

This really needs something like a Where Saves Come From series to fully unpack. Today we aren't even looking at the tip of the iceberg...we're in a fog looking at just a few feet while standing on it.

My basic hypothesis is that save % has much more utility than we analysts have been saying.

2/

1 8

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

Friday Stat Dump: Saves and Save %

The analytics folk, me included, have for the past decade or so extolled the virtues of evaluating goalkeepers on something like goals against - post-shot xG (G-PSxG).

I don't think that's wrong, per se, but, if that is the only metric, we've got some issues.

1/

3 7 16

Jamon M - MLS and footy analysis @jamonm.bsky.social · 4d

What a great run, Jeff. Looking forward to hearing what's next.

4

Reposted by Jamon M - MLS and footy analysis

Paul Harvey @paulharvey.theoutfield.nyc · 5d

I have been working on a new USA Soccer stat hub for a couple of months now in Tableau, as I grew increasingly dissatisfied with what I put together at the start of 2025. It's *finally* ready for public consumption. The images here don't show the half of it:

public.tableau.com/views/MLSAdv...

3 17 69

Jamon M - MLS and footy analysis @jamonm.bsky.social · 7d

Sure, it's not Stuver that's my concern.

1

Jamon M - MLS and footy analysis @jamonm.bsky.social · 7d

Not gonna blame club PRs for trumpeting when their players do something with an aggregate stat that actually means the team is bad at something, but we've still collectively failed to define what "good" looks like for goalkeepers.

Look for something on Friday Stat Dump this week.

Austin FC @austinfootballclub.bsky.social · 7d

Etching his name in the MLS History Books.

Congrats to @bradstuver.com for breaking the record of most saves made in a 5-year-span!

1 10

Jamon M - MLS and footy analysis @jamonm.bsky.social · 7d

No team better at making slow, lethargic, sideways passes in the final third in MLS 2025.

Paul Harvey @paulharvey.theoutfield.nyc · 7d

Crazy, almost unbelievable stat:

Sporting Kansas City is 2nd in MLS for possessions started in the opponent third. First place is Shield winner Philadelphia Union.

6

Jamon M - MLS and footy analysis @jamonm.bsky.social · 11d

I would be remiss not to point out that the best version of the Points - xPoints method is done by Eliot McKinley at least once or twice every season with his "good, bad, and unlucky" version.

Take a look and see what's changed since 14 games in:

bsky.app/profile/elio...

Jamon M - MLS and footy analysis @jamonm.bsky.social · 11d

The tables can still right themselves with 6 or 9 points at stake left in the season for each team. We won't know for sure until after Decision Day.

We can use Points - xPoints to more easily see who is above or below expected right now and by how much.

4/

1

Jamon M - MLS and footy analysis @jamonm.bsky.social · 11d

If you are wondering why the R-squared values (coefficient of determination) are so low here, I refer you once again to this important work from our analysts at @americansocceranalysis.com:

www.americansocceranalysis.com/home/2022/7/...

1 1