I recently had a conversation with a friend who is a swing trader. I wanted to understand his process and figure if I could use AI to help. He told me that he uses screens to get down to about 200-250 possibly tradeable stocks every day. Then he looks through charts for *all those names* to whittle down the list to 25-30 names. And then, finally, puts on maybe 4-5 trades. I asked him if AI might not be used to read the charts and at least reduce his workload from 250 charts a day to 40 charts a day? His answer told me everything about why he is so good at what he does. He said that he wants to look at those 250 charts. It builds his pattern recognition muscle. It builds up background knowledge on what is going on in the broader market, even if he does not trade on all those names. This conversation may have been about stocks and markets, but the lesson echoes across all the cognitive work we do. When Google started answering questions like "which is the second biggest snake in the world" directly with an answer instead of pointing you to an article about snakes, it made life easier, but it removed the opportunity to build out a much larger knowledge base. Now in some instances, it might actually make sense to just get the answer. The specific domain may not be of primary concern to you. But we run the risk of doing this with our core areas of work as well. Directly getting an answer about a regulation instead of reading the original document - in case you are a compliance officer somewhere. Or getting boilerplate code for your web application, if you are a young developer. Or getting a relevant quote for your essay without struggling through content by some incredible writers. There is value in apparently "wasted" effort. Be careful what you automate away.
There are two perspectives to how one uses AI In this example- Regulations are written as manuals not as learning materials, so one could still use AI to extract info but still use as learning tool and keep asking why on each subtopic one come across Whether you want to get job done, remove dependencies, or build your own LLM Choice is yours One way it makes the job easier, other way give you wealth of knowledge to build up own intelligence
Trust the process with patience - is the summary that GPT gave me for this post. 😅 That was on lighter note, but insightful post, Parijat.
I see ur point. But u r discounting how much more u can learn by just asking chatgpt to explain it to u. I would have never dreamt of things that I am doing today just because I asked chatgpt and it told me a direct answer on how to go about it which gives a very big headstart. Googling wouldn't have cut it.
A very important reminder of balancing between our spotlight mode vs floodlight mode.
https://x.com/itsalexvacca/status/1935343874421178762?s=46&t=HLTN5QRPMDLmOpC7CImKrg There is research on the same
Don’t waste your time using traditional methods, use TeakAI. It has already done most of the techniques, metrics and back testing interfaced through an AI chatbot.
Very true. Its not about the answers - the process to get there matters and can be more meaningful.
Skin is signal.
3moAbsolutely agree Parijat Garg, CFA . I often place handicaps on myself when I can afford to balance the trade off between getting something done right now and learning something new. Not using maps when you're not in a hurry, recall without looking something up. Go on a journey to construct an answer rather than being given the answer. I believe there's some intrinsic motivation required to do this or maybe even obsession. Extremely rewarding to show up in the world this way especially when something important hinges on your bare self, and there are plenty of those in life.