If you expected to learn useful finance at a university, you already failed as soon as you applied. If anyone there knew what they were doing financially, they wouldnt be there, and especially not teaching Finance 102. In case you were not aware, nearly all professors hate teaching lower level classes, so they naturally get forced on the worst performing professors (the ones that are closest to getting fired).
Concluding that the entirety of TA can not be useful (the Null hypothesis is true for all input signals) just from seeing that a few strawman TA in finance 102 not work is beyond absurd.
If some version of TA worked, then as soon as enough traders adopt that version of TA, it would stop working.
That is not to say that no versions of TA will work. After all Long Term Capital made a fortune on reversion to the mean...right until they went bankrupt. (And then their portfolio went on to make a fortune again. Too bad they were insolvent.) But it will only continue to work if the trading strategy is not widely known, or comes with trading risks.
The fact that you think a finance degree is centered around trading stocks tells me all I need to know about you and your beliefs. And even on the topic of stocks/derivatives, are they teaching the wrong dividend discount model? Or the wrong black-scholes model? Did they get the equation wrong or something? Or are you just mad because they aren’t teaching TA?
And just wondering, using TA, how would you calculate how much to change an M&A offer in order to compensate for including a poison pill?
Absolutely but there are cases where doing that is not practical. For instance machine learning training tasks that can take a long time to converge (months even). There are good arguments that long running tasks should not be in a notebook but they often find themselves in notebooks regardless especially in the ML domain.
That really sounds like a mismatch: Jupyter is, at the end of the day, a REPL; it seems like the exact opposite of the appropriate tool for anything that takes a long time to run.
Any product/service directly and publicly marketed to criminals is a honeypot (or will soon be). There is no reason for a real business (legal or illegal) to prefer criminal customers in good faith as all they bring is problems.
There are many agencies around the globe that set these up and continue to operate after siezing them to catch the low hanging fruit criminals. Its a lot easier/cheaper to lure criminals expose themselves directly than it is to actually investigate/track/hunt them down. Its pretty much the quantity over quality approach.
Its potentially an even worse problem than overfitting because of the error accumulation and undermining of the fitness functions. Its similar to asking students to create test questions to evaluate themself. As the proportion of self output questions grows, the context eventually drifts to be meaningless. In other words it leads to test "questions" like "The answer is A".
One measurement of this would be to watch the growth of probability of new models to say similar things to older model filters like "As a large language model ..."
Have sold to a lot of governments and large contractors. I wouldn't say its a death sentence, just different business model than most startups. Government and large orgs that send out RFPs are essentially asking for a product/service to be made and maintained specifically for them. Unless your essentially in the consulting business, its rarely worthwhile for a normal startup as it eats A LOT of time for a single customer. Government can be very lucrative if you can find the right niche but its always exceedingly painful and one of the governments favorite pastimes is completely 180ing requirements right after you complete them.
Problem here to not underestimate is that if you are not experienced in RFPs you will spend 10x to 20x the time than someone that has experience int these processes, and then there is a real risk that you optimistically misinterpreted some qualification criteria that disqualifies you in the process. So if you have done this before, do it. If have to do it for the first time, you want to consult someone who has walked the path.
The Federal Reserve: Don't worry my friend, its all just transitory due to <insert latest event>. *Slaps side of money printer* These babies can print us out of any hole!
Nice try FBI. In all seriousness though, has it actually gotten so bad that yandex of all search engines is less censored? Or is it just less censored when it comes to topics controversial to the US (and not russia)? The fact that so much censoring is going on that google has a "hold on while we censor this" page is insane.
It may also be part of their anticompetitive war on other browsers. I get captchas constantly in a new default Firefox profile, but not in a new default chrome profile. Spoofing user agent to recent chrome agent in Firefox makes the captchas happen far less often for me.
I look forward to the day the manager can have a daily standup/progress report meeting oblivious to the fact that (s)he is the only one actually present... One step closer!
Concluding that the entirety of TA can not be useful (the Null hypothesis is true for all input signals) just from seeing that a few strawman TA in finance 102 not work is beyond absurd.