iiinova — where your digital self keeps evolving

Reposted by devinpleuler

Matthew Doyle @mattdoyle.bsky.social · 11/02/2026

Fifteen years ago, I got my dream job writing for MLSsoccer. Last week, that came to an end. I'm off on my own, and it's time to build something new – with a little help from all of you. tacticsfreezone.ghost.io/an-end-to-my...

100835136

devinpleuler @centralwinger.com · 30/01/2026

En route to Houston for @americansoccerinsights.com summit. Really looking forward to catching up with some old faces.

250

Reposted by devinpleuler

cata @catabush.com · 23/01/2026

In the wake of FBref's defacto death, I saw a lot of posts suggesting alternate data sources. Unfortunately, many of them focus exclusively on men's soccer. To measure what options remain for woso, I conducted a survey of the most popular FBref alternatives. What I found is disappointing: 🧵

Bar chart
Title: Best sources for women's soccer data
Subtitle: Women's competitions with xG Data | 8 comps surveyed
Chart shows # of comps on y-axis. Only FBref scored 8/8. Next closest is Fotmob at 4.

410937

devinpleuler @centralwinger.com · 03/11/2025

The best thing about LLMs in data science is that nobody forgets to label their axes anymore.

130

devinpleuler @centralwinger.com · 02/06/2025

When your step count is larger than your sample size. Ahead of the election, I walked 65 km around my electoral district and took a picture of every political lawn sign. www.centralwinger.com/p/getting-st...

080

devinpleuler @centralwinger.com · 15/05/2025

I have the same suspicion, but you're better qualified to answer that question!

100

devinpleuler @centralwinger.com · 15/05/2025

So if Distance is flat, and Sprints are up, that means ... Walking must also be up? There has to be some sort of reciprocal change to the ensure balance. I'm having a very difficult time believing this data, especially after the inflation of stoppage time added minutes.

5191

devinpleuler @centralwinger.com · 02/04/2025

This is great! Very intuitive. I suppose pooling will have this effect with any sort of aggregation along an axis? Pretty useful.

020

devinpleuler @centralwinger.com · 31/03/2025

Not silly! I don’t think there’s anything wrong with eyeballing it (which is what I did here). It’s difficult to “average” a set of points because of the same order-invariance issue. I’ve found success with something called a Wasserstein Barycenter.

000

devinpleuler @centralwinger.com · 31/03/2025

First blog post in a while. Tracking data research probably wasn’t worth it until recently. And some unexpected utility of the Hungarian algorithm. www.centralwinger.com/p/tracking-d...

centralwinger.com

Tracking Data was a Red Herring

It's not anymore.

2367

devinpleuler @centralwinger.com · 24/02/2025

Lucasfilm should have hired @mikeduncan.bsky.social to do a Revolutions-style accompanying podcast for Andor. Such a missed opportunity.

2181

devinpleuler @centralwinger.com · 27/01/2025

Is there a conceptual issue with performing a bayesian update on xG measurements after a shot? The outcome is pretty useful information, and it's a mathematically valid approach. I assume someone has already tried this?

130

devinpleuler @centralwinger.com · 13/01/2025

That's what I've been playing with! But would prefer some of your feature engineering out of the box!

020

devinpleuler @centralwinger.com · 13/01/2025

What I’d really love is if unravelsports could be used to make node-level predictions instead of just frame-level.

200

devinpleuler @centralwinger.com · 13/01/2025

One of the first things I want to try with this is plugging in your GNN model.

120

devinpleuler @centralwinger.com · 13/01/2025

Big thanks to @probberechts.bsky.social – this uses his lovely 𝗱𝟯-𝘀𝗼𝗰𝗰𝗲𝗿 package under the hood.

030

devinpleuler @centralwinger.com · 13/01/2025

And a live demo on Streamlit Community Cloud:

st-soccer.streamlit.app

120

devinpleuler @centralwinger.com · 13/01/2025

I built 𝘀𝘁𝗿𝗲𝗮𝗺𝗹𝗶𝘁-𝘀𝗼𝗰𝗰𝗲𝗿, a custom React component for Streamlit for working with soccer tracking data. And a blog post on 𝗕𝗶𝗴 𝗗𝗮𝘁𝗮, 𝗧𝗶𝗻𝘆 𝗧𝗲𝗮𝗺𝘀

centralwinger.com

Big Data, Tiny Teams

Introducing streamlit-soccer and some thoughts on team tools

36119

devinpleuler @centralwinger.com · 01/01/2025

First search term of the 2025 (and my parental leave) was “baby ate cat food”, so we’re off to a good start.

0162

Reposted by devinpleuler

Tiotal Football @tiotalfootball.bsky.social · 27/12/2024

The theory here is that some kind of basic property of “goal probability” is “option-ness” open.substack.com/pub/absolute...

open.substack.com

Options have Value

Zone-reads, pick 'n rolls, cut-backs

24714

devinpleuler @centralwinger.com · 13/12/2024

Really enjoyed the idea of using the sideline to apply equal-and-opposite pressure relative to the player velocity. Super clever!

140

Reposted by devinpleuler

Joris Bekkers @unravelsports.bsky.social · 12/12/2024

🏃‍♂️⏪ 🏃‍♀️ 𝐀𝐧 𝐈𝐧𝐭𝐮𝐢𝐭𝐢𝐯𝐞 𝐌𝐞𝐚𝐬𝐮𝐫𝐞 𝐟𝐨𝐫 𝐏𝐫𝐞𝐬𝐬𝐢𝐧𝐠 𝐢𝐧 𝐒𝐨𝐜𝐜𝐞𝐫 It's been over a year now since I was invited to present my 𝐏𝐫𝐞𝐬𝐬𝐢𝐧𝐠 𝐈𝐧𝐭𝐞𝐧𝐬𝐢𝐭𝐲 metric to a small group of people at a SkillCorner event in Dallas, TX. I've finally put some effort into writing a blog on it! unravelsports.github.io/2024/12/12/p...

unravelsports.github.io

🔬 An Intuitive Measure for Pressing

For a long time I’ve had the aim to develop an intuitive measure for pressing that can be used by coaches, assistants and (data) analysts to identify and analyze pressing situations, compute advanced ...

43718

devinpleuler @centralwinger.com · 09/12/2024

I think you can get away with relative features like angle and distance to the team centroid combined with a few other heuristics like the number of teammates ahead/behind/left/right. Similar principals.

010

devinpleuler @centralwinger.com · 09/12/2024

I'm proud of that one! Definitely elegant. But I think there are more straight-forward methods for achieving similar results.

100

devinpleuler @centralwinger.com · 09/12/2024

These are some of my favourite "tricks" that I've picked up over the last few years that have been particularly useful in soccer analytics.

centralwinger.com

Béziers, Bivariates, and Beyond

Some Tricks I Wish I Knew Earlier

25615

devinpleuler @centralwinger.com · 04/12/2024

For the most part we do! Some data suppliers have been refreshingly open about their approach.

020

devinpleuler @centralwinger.com · 04/12/2024

I suspect this is negligible, but I’m willing to be proven wrong.

000

devinpleuler @centralwinger.com · 04/12/2024

I don’t think I necessarily agree here. Third party tools are actually really good now and off-the-shelf foundational models (e.g. xG) are production ready. This helps teams go from zero to 75% really fast and help justify continued investment.

110

devinpleuler @centralwinger.com · 04/12/2024

Score effects are proxies for important features that better data can just measure directly. Like pressure. You can actually measure how good your model is by how little score effects matter if included. But you shouldn’t.

120

devinpleuler @centralwinger.com · 02/12/2024

Is this a market for lemons argument!? Analytics twitter is fully back.

140

devinpleuler @centralwinger.com · 02/12/2024

I think this is a pretty good framework to think about this. But Arsenal was a "big club" and led the analytics arms race for a decade.

010

devinpleuler @centralwinger.com · 28/11/2024

Where can I buy these, and have them shipped to Canada?

010

devinpleuler @centralwinger.com · 27/11/2024

Enjoyed speaking with the folks at Analytics FC. I haven't spoken a whole lot online about my new role. It's exciting to be working across multiple sports. Learning a bunch.

1153

devinpleuler @centralwinger.com · 25/11/2024

New post (!) – on xG, rebounds, conditional probability, and the need for more storytelling. There are still a lot of problems with how we use this model.

centralwinger.com

Penalties and Conditional Probability

Expected Goals still need storytelling.

25612

devinpleuler @centralwinger.com · 20/11/2024

I think this is the generalizable learning across analytics in pretty much every sport. Shift dramatically in baseball. Go for it on 4th downs in American football. Take more corner threes in basketball. Pull the goalie earlier in hockey. We’re just unsure what the soccer parables are.

260

devinpleuler @centralwinger.com · 19/11/2024

I think that I mostly agree with this, though I suspect that creative players are likely to generate more value in a given situation. But the direction of the causal arrow is important.

010

devinpleuler @centralwinger.com · 18/11/2024

You probably can't. But I don't think my suggestions here invalidate existing approaches.

010

devinpleuler @centralwinger.com · 18/11/2024

Unfortunately not. They pulled most of their historical session recordings off YouTube this past year for _reasons_.

100

devinpleuler @centralwinger.com · 18/11/2024

For sure. I think you could validate the player-adjusted decision models against a generic baseline. Any sort of stable performance increase would should give you confidence that you're capturing real signal.

110

devinpleuler @centralwinger.com · 18/11/2024

Feels like a very natural direction of research for SkillCorner!

100