This is deep work, bringing AI to bear in next-level investing.
The only thing I'd be interested in knowing is getting other information sources outside of SEC filings. The most telling conditions that impact long-term performance often come through other channels.
Absolutely, SEC filings is a starting point and tends to be a good source of truth because of regulatory requirements. Trying a company’s products and services (direct evaluation) and relying on crowd-sourced knowledge on Reddit, X, and the macro environment are great layers in a research process.
I was a floor trader in the Major Market Index (a futures market attempting to clone the DJIA) in October of 1987 when the market broke what in today's terms would be almost 12,000 points (no, not 1,200 but 12,000). I don't know what Henry Shatkin (president of Shatkin Trading) had done to prepare for it, but he wandered into the pit mid-morning, kept quiet with his hands down most of the day and then bought a (forgive me) shitload of contacts right near the bottom. I'm not sure how that applies to Jing Xie's article today, but I promise you, I'll still remember that day when they close the lid on the box and drop me in the hole.
Great story! Thanks for sharing. One easy resource everyone now has is the fear greed index. When it hits low single digits and on days where the entire market is swinging 5-10% in a single day it’s time to buckle up and buy some good names to own for a while. https://www.cnn.com/markets/fear-and-greed
The "kill conditions" checklist is the part I'm stealing. Most people treat AI monitoring as infallible until it flags something wrong, but the IBM buyback example shows the real value is catching what a ratio-only screen misses. To answer the question directly, my line has mostly been drawn by accident, I let AI draft anything low-stakes and still double-check every number myself, but I've never actually written that down as a rule.
The standing limit order story is the part I'm stealing, deciding the price when you're calm and letting the order do the deciding when you're panicking removes the exact moment most people blow the trade. I've definitely treated an AI screen as a verdict before, so the bit about five false "urgent sell" flags coming from empty data fields is a good reminder to keep checking the checker. How often do the kill conditions actually trigger for you in practice versus just sitting there as a sanity check you rarely need?
Hope this is helpful to everyone. Please send me any questions and would love to read comments from subscribers.
Thank you. This is extremely helpful. I’m going to try this. At 69 it’s late for me but I want to pass this along to my children and grandchildren
Thanks! 🙏
So glad to hear, JB!
This is deep work, bringing AI to bear in next-level investing.
The only thing I'd be interested in knowing is getting other information sources outside of SEC filings. The most telling conditions that impact long-term performance often come through other channels.
Absolutely, SEC filings is a starting point and tends to be a good source of truth because of regulatory requirements. Trying a company’s products and services (direct evaluation) and relying on crowd-sourced knowledge on Reddit, X, and the macro environment are great layers in a research process.
100%!! Hrs of research that normally are skipped
I was a floor trader in the Major Market Index (a futures market attempting to clone the DJIA) in October of 1987 when the market broke what in today's terms would be almost 12,000 points (no, not 1,200 but 12,000). I don't know what Henry Shatkin (president of Shatkin Trading) had done to prepare for it, but he wandered into the pit mid-morning, kept quiet with his hands down most of the day and then bought a (forgive me) shitload of contacts right near the bottom. I'm not sure how that applies to Jing Xie's article today, but I promise you, I'll still remember that day when they close the lid on the box and drop me in the hole.
Great story! Thanks for sharing. One easy resource everyone now has is the fear greed index. When it hits low single digits and on days where the entire market is swinging 5-10% in a single day it’s time to buckle up and buy some good names to own for a while. https://www.cnn.com/markets/fear-and-greed
The "kill conditions" checklist is the part I'm stealing. Most people treat AI monitoring as infallible until it flags something wrong, but the IBM buyback example shows the real value is catching what a ratio-only screen misses. To answer the question directly, my line has mostly been drawn by accident, I let AI draft anything low-stakes and still double-check every number myself, but I've never actually written that down as a rule.
But that’s already an excellent guardrail, starting with low stakes dramatically decreases risk :) excellent job, Elaine
Writing the price down before the panic is the whole trick. Everything else is noise.
The standing limit order story is the part I'm stealing, deciding the price when you're calm and letting the order do the deciding when you're panicking removes the exact moment most people blow the trade. I've definitely treated an AI screen as a verdict before, so the bit about five false "urgent sell" flags coming from empty data fields is a good reminder to keep checking the checker. How often do the kill conditions actually trigger for you in practice versus just sitting there as a sanity check you rarely need?