Hello friends, Vik here! I'm reviving this account for Confusion Matrix, my sports analytics blog.
I'll be sharing medium-form articles on soccer, football, and whatever else I can turn into a statistical problem.
Check out my recent appearance on the Corridor of Uncertainty podcast! Huge thanks to Dr. Jess Hargreaves and Dr. Rich Bingham for having me on!
Was Tuchel Right to Park the Bus vs Argentina? ft. Vik Shirvaikar https://t.co/FuWvzggKvv via @YouTube
It turns out the coaches might be right: an ugly win is still a win. Early in the season, the most important thing is just to survive and avoid cheap losses, no matter how bad you look doing it. There’s plenty of time later to worry about the eye test.
Every year, we scrutinize the college football teams that get off to a hot start, to separate the future playoff contenders from the frauds.
But are "bad wins" actually red flags? Or are style points in September completely worthless by December?
https://t.co/mbCLNQ2cYc
I looked at every FBS team that finished September undefeated in the last five years.
When I split them into eventual over- and under-performers, their early-season results were basically indistinguishable, by margin of victory, offense, or defense.
That elevates DLS from a rain rule to a strategic tool.
If your advantage is in wickets, you have room to take risks. If it’s in overs, it's okay to be patient.
We already have the tools to explain why a chasing team is ahead or behind - we just only use them when it rains.
Everyone loves to hate the Duckworth-Lewis-Stern method, cricket’s mysterious and chaotic rain calculator.
But DLS can be so much more: learn to read it properly, and it provides deep strategic insight into any run chase, even when it isn't raining.
https://t.co/B2eCC8iagV
Instead of only using DLS to find a par score, hold the runs fixed and read the map along different dimensions to see if the chasing team is ahead or behind of the expected wickets and overs.
Same match, same chart, three different interpretations, regardless of the weather.
I also found three teams — Iowa, TCU, and Boise State — that have been perpetual glitches in the matrix, defying the polls year after year regardless of how much the sport changes.
At some point, it stops being noise and starts being a personality trait.
The AP preseason Top 25 is out, which means college football is close enough for everyone to start hoping that this might finally be the year.
But with the portal constantly scrambling rosters, how much can we actually trust preseason polls anymore?
https://t.co/UqVkSKyaT7
I looked at the last 12 years of preseason and final rankings to see whether the polls really went haywire in the NIL era.
The answer: yes, for a couple years, but then voters caught up — except at the top end of last season’s poll, which was impressively, historically wrong.
Against Mexico, England conceded more shots but kept the chances cheap. Against Argentina, Tuchel made the same bet.
If it had worked, it would have been called genius game management. Because it didn’t, it became cowardice. As usual, the line is thinner than we want to admit.
Everyone knows you shouldn’t park the bus.
Or at least, that’s quickly becoming the new statistically-enlightened conventional wisdom.
But my latest piece uses the England-Argentina match to ask: when is sitting deep to defend a lead actually rational?
https://t.co/VIRtnmkrbo
Attacking may improve your xG margin, but when you’re already up, that isn’t the only thing that matters.
Push forward and you create more opportunities for the match to break open. Sit deeper and you give up that threat — but (hopefully) also make your opponent’s chances worse.
in the Oxford registration system "English" and "American English" are separate options for language proficiency so I'm proud to report I'm now officially bilingual
WE DID IT: with a landslide margin of 1785–912, we are officially the MITGSU-UE!!!
What started 4 years ago with a dozen students in an MIT classroom discussing the needs of graduate workers has culminated in this historic victory for student-workers at MIT. Now, we celebrate!
@sir_miyares This is an obscene question but I found a couple that I think *you* would like better...
- Inception/Incredibles/Indiana Jones/Inside Out
- Caddyshack/Casablanca/Casino Royale/Catch Me If You Can
For the second one you can even add as many "Cars" movies as you want!