I'm generally more in agreement than disagreement with what you've written. I'm experiencing a bit of a "wow" effect, as this is the first time I've seen something broken down step-by-step with a chat. I've worked with AI before, but I didn’t realize the process had advanced this much. I'm interested in using AI as a tool rather than some "holy grail" answer machine. Right now, I’m losing a lot of time parsing data and finding the right pieces because I constantly have to write small utility scripts with a lifespan of maybe an hour or two. Do you have any of your own experience documented somewhere? I’d genuinely love to read about real, practical experiences in trading rather than the YouTube fluff where AI is more of a novelty than a serious tool.
Fidonet? we might be of a similar age. ChatGPT helps but it is not able to replace anyone or anything, simply because you have to ask the right question. All of those LLMs suffer from a very simple flaw. Garbage in -> Garbage out If you ask the wrong question, the model will reply like it is an expert on that shit. You will be happy with the answer because you don't know if that answer is correct and fits your task. So at the end you need someone that is able to understand the answer to value if what you got from the LLMs is pure garbage or not.
I used to think AI was overhyped but I use it to generate some code, documents, and get explanations of things (easier than looking at *overflow sites). It makes creating language bindings (PyO3 rust to python) and parsers quick; not complete but a start
One of the comments on that video sums everything up. When AI recommends you exactly what you just have bought. Today I bought a simple table for the living room, a TV table. I got it already. Well, amazon keeps sending me tables to buy, and not only that, on Youtube I get the same shit. How many times are you going to buy another table when you already have one? Never. So you can tell that AI is just mixing data together and it is far away from giving you a good advice. Another one that is very simple is "film recommendations". Netflix keeps recommending movies that I hate. That shitty algorithm just gather what people watched, and if I watched one the movies in the list they will send me the whole list at once. "you must love this" . No I don't. They have no clue of what I like
Agreed, these recommendation algos need some serious revisions and I don't think they're easy to make because your points have been argued for at least a decade, and yet. Looking forward to the day when after you purchased your table, Amazon and others will promoted goods to complement that table.
Try the prompt in chatGPT or Claude: "show me an example proof in coq" Now you can write a clickbait blog article or youtube video about how AI will let anyone automatically prove mathematical theorems. See though how the real problem is not the calculation, it is knowing what to calculate? Then even if you know what to calculate, you need to be able to use the answer. Then you also have to answer why are there these different flavors of trading software to start with? Why did one not just solve the problem at hand and put the others out of business already?
No, I don't have any Youtube, I didn't write any book etc. You know, in my experience, people who do good - most not share anything - they usually put big effort, so they expect others to put those hours and effort, or they were in right place in right time and as the edges can still appear again there is no business in sharing those. From the other side, most of the stuff available online is very rubish (including many things praised on this forum). And the problem is not only that those people want to make money selling you something. The biggest issue is you losing time, or dig yourself in some rabbit holes. The time is our most valuable asset. I think there will be special level in hell for those dream-sellers, who know exactly it is not like they say, but still sell those dreams to fill their own pockets. What more I can write in some qucik answer. I think most important is: - The markets evolve, and trying to be more and more effective, so it's not only about chosing the market, but also the approach. Many things taught online - it's not like it never worked, many of those things worked - but it was in different time in different market (and I know, now there can appear some oldschool guys who will say it still work etc. But vey easy thing - if they built their bankroll in the times of less effectivness, they can now be picky, they can seat and wait for the setups that long ago appeared often and now appear less and less often. But I'm focusing on people who start the jurney or look how to "make it") - Many older approaches can be still useful - just in other type of markets (look at crypto markets in last years, many inefficiencies) - Another thing - people don't understand that edge can be not only your research/strategy, technology etc. In example - can be costs structure - The best way to learn the stuff is to get into solid place (company) and learn from the inside. But comment here - yours skills will increase, your knowladge will increase, but many times you will also realase, that you cannot make same things outside those companies (reasons should be easy to understand) - About props (but real props, not challange props, and I'm saying about leftovers of some manual/semi manual props) - First of all - not many left (comparing to long ago), second - it's not that good as people think, more of that - those can be very scammy places (you can be robbed from your strategies, cut off when not needed, overpriced on turnover/ desk fee, fed by dreams as long as office has some money from you etc. etc.). To be honest I wouldn't advise most people to start any trading now (I mean very short term or short term). And for sure not in tradfi. If there are people who because of this or other reason want to make excesive returns (taking additional risk) then crypto is the place (although not that great as it was), betting markets (in example: Polymarket), or if there will appear any NFT exchange in some time - that will also be the place. I go, I normally don't post much, so it's already long post, and second in some time.