Like a zombie at Starbucks right when it opens at 07:00 on the weekend, I ordered an Coffee.
With the help of AI, I started testing the Alpha Logic of my Brawler System.
I am currently using this logic for the US stock market and want to see
if it works in the Korean market as well.
I already subscribed to the Google AI Pro 3-month discount plan for this.
KOSPI: A Familiar Face
The data for KOSPI stocks is well-organized, much like the US market data I usually analyze.
So, the testing went smoothly.
I used to wonder, "Are the US and Korean markets similar?" As expected, KOSPI stocks showed data and results that were quite similar to the US market. I can say,
"Okay, I understand how you move."
KOSDAQ: A Brutal Fight
The real challenge was KOSDAQ.
There wasn't as much data, and the results showed the system getting "beaten up" by the market quite often.
The numbers clearly said, "This market is not easy."
Just by looking at the backtesting results, I could feel the high difficulty of this market.
If something looked even slightly wrong, the Brawler Alpha Strategy suggested it was better to just stay out.
Having a stable performance is more important than having the highest returns.
Same Alpha Logic, Different Survival Ways
For KOSPI, I apply the Brawler Alpha Strategy almost as it is. It’s a "straight-on" style—taking some hits from the market but holding on to eventually take the profit.
KOSDAQ is different. It avoids trades when things look risky (about 1/3 more than KOSPI). It only takes hits and makes profits in the remaining 50% of the sections.
I’m not sure when I will go live with this yet, but I plan to continue backtesting through January.
From February, I will finalize the logic and start a pilot run. I’m thinking about the actual application sometime next year.
Instead of rushing, I’ll take my time every weekend morning at Starbucks with a cup of coffee to make the system stronger.
I want a "smarter" Brawler—one that knows when to take a hit and when to dodge, rather than just getting beaten up by the market.
Comments
Post a Comment