iiinova — where your digital self keeps evolving

Paul Johnson @paulrjohnson.net · 01/04/2026

There is no better feeling than new sports gear. It is much better than actually doing the sport tbh.

Paul Johnson @paulrjohnson.net · 19/03/2026

Made a pilgrimage to see the Battle of Cable Street mural round Whitechapel way yesterday. Lived in London for about a decade and only just got round to it... Also grabbed a pint by the Thames in a pub my dad used to drink in in the 60s (but I've done this one many times).

A mural painted on the side of a building, depicting the Battle of Cable Street.

The facade of an old riverside pub called The Prospect of Whitby.

050

Paul Johnson @paulrjohnson.net · 21/02/2026

Me every time I have something to say

Newspaper plastered on a wall with the headline "Best take in town"

040

Paul Johnson @paulrjohnson.net · 25/11/2025

I have to savour these moments.

A screenshot of the full time result of the Borussia Dortmund and Southampton games. Borussia Dortmund beat Villarreal 4-0 (Villarreal had one red card), and Southampton beat Leicester City 3-0 (Leicester also had one red card).

030

Paul Johnson @paulrjohnson.net · 17/11/2025

Because I know no one really cares about the Bundesliga (and because I wanted to test how well the model generalises), I have tested it out on the Premier League too! The results look pretty solid to me, though I have a little less faith in how the model does with the Premier League than the BuLi.

A table showing the Premier League 2025/26 season projections generated from 10,000 Monte Carlo simulations. The model projects Arsenal are the most likely to win the Premier League, with a 63.5% probability, with Manchester City, Liverpool, and Chelsea rounding out the rest of the top four. At the other end of the table, Wolves (72.6%), Burnley (52.3%), and Leeds (35.8%) are the most likely to be relegated.

A table showing the predictions for the next round of Premier League fixtures, showing the predicted expected goals and the match outcome probabilities for each game. Arsenal have a 70.6% probability of beating Tottenham, while Liverpool are given a 68.2% probability of beating Nottingham Forest.

073

Paul Johnson @paulrjohnson.net · 16/11/2025

The model uses Monte Carlo simulation to project outcomes in the Bundesliga for the 2025/26 season, but it also models team ratings across the league and predicts upcoming fixtures.

A screenshot of the Predicting Ball app showing the Team Strengths tab, displaying Borussia Dortmund's Team Metrics. The tab shows cards that display the overall rating (0.4), attack rating (0.39), defense rating (0.41), and project points (64), and below this are two plots showing how BVB compare to the rest of the league (overall they are the third highest rated team behind Bayern Munich and Bayer Leverkusen).

A screenshot of the Predicting Ball app showing the Fixture Predictions tab, displaying the outcome probabilities for Bayern Munich's upcoming game against Freiburg (Bayern have a 86.2% probability of winning) and Borussia Dortmund's game against Stuttgart (BVB have a 54.7% probability of winning).

163

Paul Johnson @paulrjohnson.net · 30/03/2025

I may be a sourpuss but I still found this pretty endearing.

A plate with a pork banh mi and a smiley face drawn in sriracha.

1151

Paul Johnson @paulrjohnson.net · 28/03/2025

A little boy enjoying his silly beers

000

Paul Johnson @paulrjohnson.net · 22/03/2025

I made my own radler earlier today, with a can of Lightbulb and some homemade raspberry lemonade. Because I'm a fancy boy that deserves fancy little drinks.

A teku glass stands on a wooden chopping board, filled with a homemade radler made from a mix of Verdant's Lightbulb XPA and homemade raspberry lemonade. The XPA can and a jug filled with raspberry lemonade sit on either side of the glass.

060

Paul Johnson @paulrjohnson.net · 20/03/2025

Only the finest for my refined palette. You probably wouldn't appreciate it. You don't know the finer things like I do.

A tin of spam held to camera by the person (me) taking the picture. The person's hand and tattooed arm is in shot, with a bin, fridge, and the kitchen floor in the background.

6130

Paul Johnson @paulrjohnson.net · 19/03/2025

Fresh box of treats from Brewser. This month from Pomona Island. Decent mix of beers.

Eight cans of Pomona Island beer, stacked in two rows of four.

The top row of beers are JACK BE NIMBLE, JACK BE QUICK (West Coast Pale), SQUIRREL-SQUASHING, DEER-SMACKING (West Coast IPA), ISÉ-MAIRI (Orange Small Pale), and CHECK OUT THE PARTY (Blackcurrant, Blackberry, Apple and Cinnamon Sour).

The bottom row of beers are I WISH I COULD SHIMMY LIKE MY SISTER KATE (IPA), O'HANRAHA-HANRAHAN (Blackcurrant Gose), MY NAN'S GONNA LOVE HER CIRCULAR SAW (Salted Chocolate & Rosemary Stout), and ВІВBIТY-ВОВВІТY HAT (Pale).

030

Paul Johnson @paulrjohnson.net · 17/03/2025

Getting nice little treats delivered to start the week right. I deserve my treats.

The neon yellow 2024/25 special edition BVB kit, inspired by the 1994/95 kit.

030

Paul Johnson @paulrjohnson.net · 08/03/2025

My pals and I have been coding your vibe from across the bar, and we were wondering if you'd like a sniff of my lizard.

100

Paul Johnson @paulrjohnson.net · 08/03/2025

Can I offer you a picture of a dog in this trying time?

A small dog (breed: ratcha) nestled into blanket looking very cozy, staring at camera as she falls asleep.

030

Paul Johnson @paulrjohnson.net · 03/03/2025

Brewser sent me some delicious treats. Good way to start a week off.

Eight cans of Verdant beer, stacked in two rows of four. The top row of cans are De-Bleaker (IPA), Sundialer (Pale Ale), Light Bulb (XPA), PuTTTy (TIPA). The bottom row of cans are Sniffing the Wrong People (IPA), There is Only Now (Pale Ale), People Money Space Time (Session IPA), Remarks (IPA).

010

Paul Johnson @paulrjohnson.net · 27/02/2025

To celebrate the release of Flashes, please enjoy a splendid dog.

A small dog laying on a sofa submerged by a fluffy blanket, with just her eyes, nose, and feet poking out.

0340

Paul Johnson @paulrjohnson.net · 23/11/2024

Okay this haul feels a little excessive.

14 cans of beer stacked on top of each other, in two rows of seven, on a desk. Eight of the beers are Vault City sours (and one stout), and the others are the six beers produced by various brewery collaborations for Brewser's ProjectX 2024.

090

Paul Johnson @paulrjohnson.net · 11/11/2024

Unfortunately, Southampton are just bad. I didn’t include the figure below in the blog post because I have been forced to accept that no one else cares about Saints, but I’m posting it here because I need the world to know how silly they are.

A time-series plot visualising Southampton's total and predicted points each season from 2012/13 to 2022/23, including a fitted line that visualises the trends in predicted and total points. Southampton overperformed expectations, on average, in the first half of the seasons being studied but started to decline after the 2016/17 season, both in terms of their predicted and total points, before a considerable dip in their total points (which is also a massive underperformance of their predicted points) which resulted in their relegation.

100

Paul Johnson @paulrjohnson.net · 11/11/2024

The model performs pretty well but struggles a bit with the Premier League. It overestimates the worst teams, and the best consistently overperform because the rest of the leagues are dragging the predictions down a bit. Man United did break the model, though (discussed a little in the blog post).

A time-series plot visualising Arsenal, Chelsea, Liverpool, Man City, Man United, and Spurs's total and predicted points each season from 2012/13 to 2023/24, including a fitted line that visualises the trends in predicted and total points. Arsenal significantly underperformed expectations midway through the period but in recent seasons have experienced similarly significant overperformances, while Liverpool and Arsenal have consistently overperformed expectations recently. Manchester United and Tottenham have performed above expectations in earlier seasons but have otherwise met expectations. Finally, Chelsea have the largest swings, with massive overperformances in the early seasons and similarly significant underperformances in other seasons.

100

Paul Johnson @paulrjohnson.net · 11/11/2024

The findings highlight that the Premier League is an outlier. The rest of the Big Five are relatively similar (when you account for the BuLi & Ligue 1 diverging due to fewer games), but the Premier League is just too rich to be normal.

A regression plot visualising the relationship between squad market values and the model's predicted points in the Big Five leagues from 2012/13 - 2023/24. The plot includes fitted regression lines for each league. The five regression lines are similar in shape. However, the Premier League slope is more gradual with a smaller intercept, suggesting that Premier League teams need more expensive squads to compete and increases in squad value are worth fewer points. The Premier League line also extends further on the x-axis because the most expensive squads in the Premier League are worth significantly more than the rest of the leagues. The intercepts for the other four leagues are very similar, but the Bundesliga and Ligue 1 tail off as squad value increases due to playing fewer games in a season and, therefore, having fewer available points that the increases in squad value could gain. Finally, La Liga and Serie A have similar slopes. The endpoint of the La Liga line is further along the x-axis due to the significantly larger Real Madrid and Barcelona squad values.

100

Paul Johnson @paulrjohnson.net · 11/11/2024

I cannot move in case I disturb this perfect angel. My day is over. Potentially my whole week.

A beautiful black and white dog (a ratcha) asleep in the arms of their owner, under a fluffy blanket, with their tongue poking out.

041

Paul Johnson @paulrjohnson.net · 21/08/2024

For reference, these three are on the bookshelf, and I've got The Effect and Causal Inference in R open in tabs in my browser right now too.

A closeup of multiple causal inference books on a shelf, among others about related topics. The three books are Cunningham's Causal Inference: The Mixtape, Pearl & Mackenzie's The Book of Why, and Rosenbaum's Causal Inference.

100

Paul Johnson @paulrjohnson.net · 21/11/2023

Follow me, I look like this

A photoshopped image of Cilla Black with an angry/disgusted face (that is not her own face).

010

Paul Johnson @paulrjohnson.net · 08/07/2023

This little shit woke me up at 6am because he was hungry and impatient, and he decided to bother both me and the other dog until he got either food or attention (eventually both). Me and him gonna fight today.

A brown and white chihuahua puppy sat on the end of the sofa looking cute.

010

Paul Johnson @paulrjohnson.net · 01/07/2023

Since I'm seeing some introductions: I'm a data scientist in the NHS. My background is in social science, and I'm interested in using causal inference and machine learning to model human behaviour. I use both R and Python, so I never remember how to do anything in either. Here's a dog.

A rat terrier/chihuahua mix curled up and sleeping on a pink blanket.

110

Paul Johnson @paulrjohnson.net · 29/06/2023

Another social media account for me to neglect? Oh go on then.

260