Very cool. The growth in variance from ~14 games to 82 looks pretty close to the growth expected of a random walk.

Do you have a version of this from last season? I'd be interested to see what the magnitudes look like after a full 82.

You are talking about teams that finished 44 points apart last season.

"Talented software engineers"

Have you ever used any version of the Spotify app?

💀💀💀

Thinking vs Listening

Hype

Sounds like 9 OT periods in 7 games

I thought +/- excluded special teams

I would prefer he beats Gretzky with a 10ft empty netter and then retires

Assuming you still use goalie terms in your xG model, would the xG values be deflated here because Shesterkin is the goalie?

Look, I think ML tech is fascinating too. I build models for fun in my free time. But it's unrealistic to assume that there is no ceiling. As we discover breakthroughs, we will also discover limitations. The tech growth curve is more sigmoid than exponential. Physical limitations are inevitable.

It will slow down eventually. It happens with any widely adopted technology. We'll discover all practical applications. The growth-at-all-cost VC $ will dry up. Resource consumption will make significant advancements impractical. We'll keep refining the tech, but any tech boom like this is ephemeral

He's on pace for 43g 4a

That's because the mooterus is a great logo