My sample of shots for my xG model is at 313,319 (and growing, I'll likely double that soon), so I can dig into some non-xG data...

And I found that the more a team leads by, the more likely they are to score with their shots. It's statistically significant as well The graph shows all non-penalty shots in my sample taken when the team was losing by 4/3/2/1, drawing (gamestate = 0), or winning by 1/2/3/4

Conversion rates when losing are all effectively the same (10.5%)

Shots are statistically more likely to go in when drawing than losing..

why the hell do the USMNT stans care so much about shouting into the void that the US would win the Asian Cup?

Being in the concacaf gold cup final almost by default has made them so entitled Like, yes, we are a very good team.

But so is Japan

And Korea

And Iran. & Iraq. Saudi Arabia, Qatar, Australia... And then teams like Uzbekistan, UAE, Jordan, Oman, they can beat anyone on their day

Thailand, Vietnam, Kyrgyzstan, Tajikistan, Lebanon, Bahrain... not easy!!!

🚨🚨 Python package update

I have a treat for y'all today!

I've added a new function to my python package. It is code to create a post-match dashboard for a given match on Whoscored

Please see this link for more info on the package

👉 github.com/griffisben/gri…


Please note that you will need to uninstall and then re-install the package if you've already installed it before

And IMPORTANT NOTE, you will need ChromeDriver for this to work. I apologize, but I can't be tech support for everyone. Here is that link
chromedriver.chromium.org/downloads

My thoughts on FBRef's standard scouting report metric when using it on a CB

Can really only look at 4-6 metrics, and they're all on-ball, in-possession ones

Not really a reason to look at almost all the attacking metrics, and nearly every defensive metric is bad Non-penalty goals: not really useful. Most CBs only score when up for corners, so dependent on too many factors

npxG: same as npG

Shots: fine to look at but will by and large be shots from corners, so keep that in mind. Still not good to *compare* off just this metric

If Khvicha Kvaratskhelia ends up leaving Napoli, are there any possible replacements similar to Kvara?

Here are the top 20 similar style players, in leagues with relatively similar styles to Serie A, and teams in those leagues with relatively similar styles to Napoli...

🧵 If we look at Kvaradona, he was a young, high-potential prospect that, while incredible, may not have been expected to be THIS good in season 1 in Naples.

Any on this list that meet that? I think so.

I'll look at a few U22 talents with somewhat similar styles to Kvara

For a little expansion, some vizzes of the data

Brighton are a good example because the goal kicks and their long passes right after short goal kicks tend to come from similar areas (central)

A bit higher cmp% on long after short GK than on long GKs

https://twitter.com/BeGriffis/status/1668276112642678786

Fulham another example, but with different areas of the pitch both passing from and being targeted.

If teams want to get the ball out of the def. 3rd with a goal kick...

... should they boot it from the goal kick, or should they take it short first and THEN boot it long?

Data from 193 teams across 12 leagues.

The answer & method in the thread 🧵 To answer this, I calculated the completion % of all "long" goal kicks, which I define as going at least 40% of the pitch (about the base of the center circle). I also calculated the completion % of passes going to this area where the preceding action was a goal kick into the box

Post-Match Data 📊🏴󠁧󠁢󠁥󠁮󠁧󠁿🇮🇹

Manchester City 1-0 Inter

#UCLfinal #MCIINT Manchester City starting XI passing sonars

At least on TV (can't speak for in-person obv), the AFC CL final atmospheres are just so much better

Really something to be said for the 2 home matches vs a corporate-centered neutral venue Al-Hilal atmosphere

https://twitter.com/TheAFCCL/status/1652408271506923524?s=20

The county that Lionel Messi is moving to...

66.5% Spanish speaking

24.9% English

People in Miami should be learning Spanish not English, if they have to learn a language to live there

Oh wait they don't, because by law you can speak any language in this country

https://twitter.com/BeGriffis/status/1667510176893509638

Source: 2021 ACS: data.census.gov/profile?g=050X…

This is an example why data alone can't be the final word on players.

As we all know I'm a data person. But I'd like to think I'm one of the good & realistic ones...

🧵 I make role scores, where several key metrics are all importance-weighted to make a final score

And....... You may notice everyone's favorite GK, David de Gea, above Ederson in a "modern gk" score

The metrics I include with varying weights are: psxG +/- per 90, save %, percent of passes being short, & long pass completion %

psxG +/- is most important, save % least

Probably a good thing we didn’t have social media in the 1800s when British statisticians were pioneering experiments and general statistics in agriculture so we don’t die of malnutrition and food shortages

https://twitter.com/theeuropeanlad/status/1658546242622742563

Josiah Smith, 1855: “The 'data people' need to get out of the farming world as soon as possible. Otherwise I will quit eating in 3 years time. I am being serious."

Benevento Tactical Analysis 🧵🇮🇹

From their 1-2 loss to Venezia last weekend

👇👇 I'm posting this because Benevento just sacked Fabio Cannavaro & hired Roberto Stellone to try to save them from relegation.

I'm planning on analyzing this weekend's match vs Cagliari as well to see what immediate changes Stellone makes in their (likely) final attempt to survive

Which Serie B 🇮🇹 teams may have the most effective press this season?

Bari allow 0.44 passes per possession fewer than expected, based on their pressing intensity. Despite not pressing as intense as the rest of the league, they may be good at breaking up opponent possessions Those two teams overlapping are Reggina & Ternana, they have basically the exact same number for both metrics lol

Ben Davies as a progressive LCB or defensive LWB (basically the same position depending on coach lol)

Cannot stress how good he is at the few things needed for those roles. Reads the game incredibly well, can beat an opposing winger with a cute disguised dribble, has a good pass

https://twitter.com/EiFSoccer/status/1623041195231350790

His issue is that he's not the fastest player on the pitch so the rare times he does make a mistake he can get caught out (it is rare, often he's just been instructed to join the attack and another player loses the ball so he *looks* out of position)

People see that and slag him

A Bundesliga club needs Jairo Concha 🇵🇪

Really strong mover/progressor of the ball from midfield. Is great at driving into acres of space, pushing defensive line back, & spraying a nice pass to facilitate attacks

Can shoot, but main use should be as enabler of all your attacks Radar does show his ability to score (5 non-pens, 1 pen) last season. I wouldn't say that's his main thing though he is pretty decent at a run into the box if it's on, as this goal shows

(had 2 this game actually, the other also a nice run into the box)

Inspired by a post from @MLS_Buzz...

I mapped out 348 US soccer clubs to see what a Brazil-like Regional League system might look like here.

This article has the link to the interactive map & the sample of promotion/relegation leagues for each region
cafetactiques.com/2023/02/07/usa… Here's the interactive map on Tableau
public.tableau.com/app/profile/be…

🚨🚨🚨
Melbourne Victory record a victory for the first time since December 11th
🚨🚨🚨

Post-Match Report 📊🧵(🇦🇺/🇳🇿)

Melbourne *Victory* 3-1 Wellington Phoenix

#MVCvWEL #ALeagueMen Melbourne Victory's starting XI passing sonars

Arthur Theate a really up-and-coming LCB. Already very good, but constantly improving too.

Took a look at some of his action success rate development over his career.

Very nice to see his consistency in defensive duel rate, & progression in passing The fairly wild variance we're seeing now in his defensive duel success rate came right when he moved to Rennes, and looks like he's steadied out since. He didn't have that same effect when moving to Bologna, but his move to Rennes also went from LCB in back 3 to LCB in back 4

Serie B team radars as of this weekend's matches.

I like making these when I'm learning about a leagues' teams, so wanted to share.

Image is huge, so zoom in! Teams sorted by current position. Thread has all teams individually Frosinone

A quick thread showing why even the best xG models can be "broken" by any player, especially in the short term

🚨 Variance! 🚨

So, let's look at some coin clipping 🪙 A fair coin is perfectly 50-50, heads-tails.

But even over 100 flips, you aren't guaranteed 50 heads!

🧵 xG can be thought of as the probability of any given shot going in.

So if a shot has an xG of 0.10, we can think of that as the average player scores that exact shot 10% of the time.

A coin flip would have an "xH" (expected heads) of 0.50: 50% of the time, it's heads