CryptoPainter
CryptoPainter
An old friend calls me a "painter", technical/data analysis and quantitative trading, providing various tricky angles to see the market, and using time to leverage. The real account is an agent account, a self-evolving strategy system is being tested, please do not copy!
32Following
3Kfollowers
Feed
Feed
I think MicroStrategy's limitation on BTC's long-term upside is similar to the leading figure in the MEME coin market...
When you see a KOL enthusiastically promoting a MEME coin, and then check on-chain to find out they already hold 5% of the supply, would you trust their hype and buy in?
Existing funds in the crypto space might still buy, but the problem is that external market funds might not...
So when I see MicroStrategy selling, I actually think it's a good thing!
The narrative of BTC is deteriorating...
Today I discovered an interesting phenomenon: NVDA and BTC have had a very high correlation in price trends over the past 10 years, but after 2025, this correlation suddenly disappears and turns into a negative correlation...
Could it be that AI is harmful to BTC?
A similar negative correlation actually appeared in 2019 as well...
Added two more trading pairs overnight and successfully tested the single strategy running multi-asset data.
The results indeed improved; the more uncorrelated the volatile assets are, the better the drawdown resistance under the same strategy. The Sharpe ratio rose from 1.7 to 2.0. According to this expectation, the more assets added, the more stable the overall strategy system will be.
Based on the usual logic of halving expectations for live data, this currently counts as just barely building a passable strategy with a Sharpe close to 1. The surprising part is the out-of-sample performance, which actually exceeds the in-sample results. Even Claude praised my strategy approach for its strong "robustness"!
Luckily, this wasn’t said by Gemini, otherwise I wouldn’t believe a single word...
CryptoPainter
The old strategy has recently started to show signs of a revival with the help of AI!
The ASR trend-following strategy, after incorporating state machines and adaptive functions, has gradually become adaptable to the vast majority of trading pairs.
So, these past few days, while the kids were asleep, I gradually built a strategy system using Cursor, specifically running ASR and its variant strategies across various trading pairs. The entire system has been deployed to the cloud, and the backend has a complete set of processes for AI to perform backtesting, optimization, and deployment.
I expect to gradually add more adaptive strategies for additional coins tomorrow!
Honestly, I didn’t expect ASR itself to be adaptable to various trading pairs. I hadn’t put any effort into this aspect for a whole year, which was a complete waste...
So I specifically chose DOGE, which is completely different from BTC, for testing. The whole strategy indeed adapts well—it's really amazing...
By the way, here’s an idea: LLMs are generally mediocre at writing strategies, but if you have them optimize existing mature factors or strategies and require them to follow the framework of a professional quantitative architect, they will provide you with a more stable strategy that yields lower returns, has smaller drawdowns, passes out-of-sample data tests, and is more robust.
This approach is a compromise given the current limitations of LLM intelligence. Compared to the Agent Trading I worked on before, this logic is much more reliable, at least the decision-making layer isn’t a black box.
Finally, all of this is premised on having a good strategy as the foundation...
If you flaunt your uniqueness excessively, showing off unconventional ideas and rebellious behavior, people will think you're just seeking attention and feel that you are looking down on them.
They will try to punish you because you make them feel inferior.
Blending in with the crowd and cultivating affinity is much safer.
Reserve your originality only for tolerant friends and those who will surely appreciate your uniqueness.
——— "The 48 Laws of Power"
The old strategy has recently started to show signs of a revival with the help of AI!
The ASR trend-following strategy, after incorporating state machines and adaptive functions, has gradually become adaptable to the vast majority of trading pairs.
So, these past few days, while the kids were asleep, I gradually built a strategy system using Cursor, specifically running ASR and its variant strategies across various trading pairs. The entire system has been deployed to the cloud, and the backend has a complete set of processes for AI to perform backtesting, optimization, and deployment.
I expect to gradually add more adaptive strategies for additional coins tomorrow!
Honestly, I didn’t expect ASR itself to be adaptable to various trading pairs. I hadn’t put any effort into this aspect for a whole year, which was a complete waste...
So I specifically chose DOGE, which is completely different from BTC, for testing. The whole strategy indeed adapts well—it's really amazing...
By the way, here’s an idea: LLMs are generally mediocre at writing strategies, but if you have them optimize existing mature factors or strategies and require them to follow the framework of a professional quantitative architect, they will provide you with a more stable strategy that yields lower returns, has smaller drawdowns, passes out-of-sample data tests, and is more robust.
This approach is a compromise given the current limitations of LLM intelligence. Compared to the Agent Trading I worked on before, this logic is much more reliable, at least the decision-making layer isn’t a black box.
Finally, all of this is premised on having a good strategy as the foundation...
CryptoPainter
While researching new strategies for the Agent and configuring a pure algorithmic state machine, I suddenly realized this mechanism could also be applied to the previous ASR strategy. So I hurriedly started modifying the code, and then I was moved to tears...
An old strategy that hadn’t been successfully optimized forward for a whole year suddenly came back to life!
As for the specific changes, it’s just using the state machine to record market volatility in real time, then fine-tuning the volatility into the original strategy’s parameters. Altogether, it’s less than 20 lines of code, but it made the original ASR channel differ in many subtle details...
The overall return of the 5-year BTC strategy improved by over 75%, while the maximum drawdown decreased by 14%!!!
Previously, when researching pure algorithmic quantification, I looked down on state machines. Only when it truly produced positive optimization effects did I realize how great it is...
I guess an updated version can be released soon!
It’s really been so tough...
