Yannick Kälber

@volatilitysmile @W98AB Digga Berlin… gottlos

@YannickKae @W98AB I’m from Berlin, tell me more about it 😂

1. what

@W98AB @volatilitysmile Germans aren't just known for cars & dark history from a century ago, but also for techno & fetish.

@volatilitysmile @YannickKae bro

@RealEmirHan Bale was the perfect cast for Bruce Wayne. He had the charisma of old-money, while still being masculine enough to embody Batman. Affleck on the other hand lacks this elegance. He feels more rugged.

Zack Snyder thinks only real Batman is Ben Affleck because “Christian Bale is still, like, you know, 5’10”. “I wanted Affleck, because, to me, Batman’s a big dude. Affleck’s 6’4”, you know? He’s a legit big dude. When he comes out in costume he’s scary.”

@trikcode There is a professor, Roland Tichy’s brother at KIT, who is famous for keeping everything in German (e.g. Brandmauer instead of fire wall), which is a bit bizarre. „Keeping it German“ seems to be a Tichy family trait…

@trikcode Most German universities teach at least their master's programs entirely in English. Since the majority of computer science literature & online content is in English anyway, Germans also learn & code in English.

Honest question. People who English is not their first language… how do they code?? Do Germans code in German? Do Arabs code in Arabic??

I kind of feel sorry for Jürgen Schmidhuber and his crusade to point out under every LinkedIn post that his academic work wasn't referenced and that he already had Yann LeCun's idea back in 2014.

@systvest @W98AB Love it. Er auch

@YannickKae @W98AB Oh ja, I work wery hard. Ze Germanz are definitely hardwörking

Ironically Germans probably work the least, but they do work efficiently. Greece rating themselves the most hardworking and most corrupt is probably accurate tbh

@W98AB recently saw this music video with EsDeeKid and Timothee Chalamet and I have to admit, this accent goes hard on him.

@W98AB Yeah, shoar lid

@J_lacasse89 @__paleologo 👍🏽, it seems to be common among investment consultants.

@__paleologo surprisingly a number of investment consultants still use it

PSA (obvious, but just in case): no one uses Black-Litterman. Don’t waste time on this. Don’t teach this (I am working on a course for Columbia and will work to remove it). Don’t do complicated things.

@istwine @istwine found another point: The admissible region is in fact the exterior of a ball in Mahalanobis space. And this boils down to a simple lower bound on the regression's R^2.

Extended it to multiple signals: functor.network/user/3370/entr… Thanks again to @istwine, who carefully read through it & correctly pointed out that the admissible region for rho is not a half-space ('cause the norm vector itself would have been function of rho).

@__paleologo Agree, but, as with romantic partners, everyone looks for different things. I have a friend for whom visuals are the most important criterion in movies, or at least one of the most important. For me, that's quite secondary.

Since we are on TV series… “A knight of the seven kingdoms” >> “House of the Dragon”. Proof that story, writing and acting beat everything else.

@joshuarolson Parents who sacrifice their kids' privacy for likes from (anon) strangers, need to touch some gras & rethink their priorities.

"Hey, Josh, why don't you ever post stuff about your kid?" TRULY great PSA. Irish, of course.

@W98AB Good question. Never noticed specifically weird formulations or answers when prompting in German. I'll have a look.

@YannickKae how does it handle german and niche german topics? I tried to use AI for some german bond market stuff last year, found it was a bit shit

Is AI good in languages that arent english? Serious Q for the people using AI who consume serious amounts of content in other languages

Every 2nd post on LinkedIn uses the formulation: "Not y, not x, but z." Not me, not a human, but an AI wrote this slop.

@istwine @macrocephalopod Thanks for reading! True, I'll change it to an inequality. I'm rewriting the article to generalize it to multiple signals anyway.

@YannickKae @macrocephalopod If you set k=0 (trade all positive signals), the formula checks x=0 and calculates 0 return (fail). But the actual strategy (buying all x>0) captures the entire positive tail of the distribution, which is highly profitable.

I’ve written a short note providing a more general & detailed derivation of @macrocephalopod’s heuristic from this nice thread. If you’ve ever wondered about why "3 * beta", for example, here you go. functor.network/user/3370/entr…

New Contributor: A Linear Regression's Predictions are a Relevance-Wtd Avg of Past Outcomes [Yannick Kalber] dlvr.it/TQzZfV