Just stumbled across an insane article: Finding a Trading Idea with ChatGPT and Claude: From Data to Backtesting. A trader shared how he used ChatGPT to brainstorm trading ideas, generate charts, and even run backtests—all without touching any of the usual trading platforms. Why do we even need these trading platforms anymore if AI can spit out ready-to-go code that connects directly to a broker’s SDK or exchange and starts trading? All the backtests are done right there in the chat. It’s wild to think that by 2025, tools like MetaTrader, QuantConnect, StockSharp, and TradingView might just be irrelevant. This is a total game-changer. PS: I'm still in shock. Looks like even quants and algo traders don’t need to know how to program anymore.
Yes, and the best is, all backtesting is useless. So just skip all software AND ChatGPT and you made one step further in progress. You should not backtest because there is no any good consistent profitable trading system or algo which work good in the future. Past data is not indicative, so you need to adapt all the time trading discretionary. That is my experience from coding and testing 10000s of strategies over 10 years being systematic.
Got it, now I understand what you were getting at. It seems like you updated your comment with more details, so now we have something to discuss. First, the article isn’t mine. Second, what you mentioned is pretty obvious. Third, saying “no need to backtest” is a bit extreme. Fourth, the article isn’t about the pros or cons of backtesting as a concept. It’s about how you can instantly get results without spending too much time coding. That last point is crucial because it addresses not “Why” but “How.”
There are few things in your question I would like to address, so step by step: 1. "Why do we even need these trading platforms anymore" - would indicate that we need those now. I'm not really aware of any serious place where ppl use those. So the straight answer is - no we won't need those same as we don't need those now. 2. As you put weight on "backtest" (just my deduction from your message, so I can be wrong on that) - it's really not that relevant, as stated before. Of course some kind of modeling, or super simple tests are needed to distinguish if your idea has any potential or is pure randomness or deviation of short term results in long term randomness. I mean, what you need you can do in super simple scripts already today, and don't need AI and crazy complicated models for that. If the specific market is highly effective - no crazy model will help you anyway, if specific market is not effective - then you don't need anything crazy to spot it, check it and trade it and make money on it. 3. All those things are just the tools, the problem is many people think, that the tool will find the way to make money for them. That is not correct. Same with AI, (for now) it's just a tool. And if you don't have any valid idea or hypothesis, that you can use that tool to check, then it doesn't matter. It's not the tool that make money, it's the tool operator ideas and experience. 4. Again about tools - different people want different tools, and that's why those platofrms probably will stay with us for much longer than you would expect. You think people will make rational decisions, no they won't. Ppl will paint crabs and bats on the chart with connections with rainbow Fibonacci with level 0,5. This is what they want. I contacted Tradingview many times, with solid ideas of improvements, with solutions that I'm aware of from very robust platforms - they were not really interested in implementing, yet they were interested in implementing rainbow Fibo and emojii on charts. So that tell you also what is the client. There is supply there will be demand. For long time. 5. Chat is not that great for now, vey overrated in my opinion. I used it for some trading related topic, some things were ok, but in general Chat made a lot of mistakes, it took long time to finally created correct versions of things. Same with programming - as for now, Chat cannot compete with experienced programmers and devs. So, the main answer to your question is: no. The aditional info - I don't know, maybe you or someone reading it will find it useful anyhow. And at the end - I also wish you a good trading
The point of trading is to take on risk when probabilities say you should - no matter what tool you use to determine that. AI is just the latest greatest, actually it's be around awhile. Something else will follow it. The more things change, the more things stay the same. Probabilities, or edge per se, is all you need and should focus on. Not what you use to get it.
There should be a special place for all of those that said "AI is going to replace..." So they all talk to each other there.