thenimblemm

$BTC's key question is simple: Is adoption still growing? So far, yes. Across 5,788 daily observations: Non-zero addresses: t^~3 R² = 0.976 Price vs adoption: N^~2 R² = 0.951 Price vs time: t^~6 R² = 0.962 If the foundation is still working, the strategy is simple: DCA, hold, and ignore short-term noise.

a 50-day moving average on BTC. long when above, flat when below. outperformed both BTC and SPY this is why i always start with simple strategies built on principles i understand. no machine learning. no complex signals. one rule. people hear "moving average" and immediately dismiss it as too simple to work. but the reason it works is because it captures a real behavioral pattern. humans herd. trends persist. one clarification though, i wouldn't call this alpha. it's a risk premium. you're getting paid to ride volatile trends and endure the whipsaws and drawdowns that most people can't stomach. this is why i always prefer strategies where i can clearly explain why i'm making money. if i can't explain it, i can't trust it during a drawdown. and if i can't trust it during a drawdown, i'll abandon it at the worst possible time. there are obviously ways you can improve this, but it's a great simple reminder.

some random ai thoughts: - for code, i went from 80/20 claude/gpt to 80/20 gpt/claude in <3 months. surprised by this tbh, and interested to see where the split is at in another 3mo. - claude still mogs gpt for non-coding agent stuff. codex feels like an engineer (which is great for coding!), whereas claude still feels like a general purpose coworker. gpt still lacks that coworker magic - i’m pretty meh on opus 4.7. my experience hasn’t been *bad*, but it certainly hasn’t been good. sideways if anything. - anthropic has got to figure out the compute thing. you can feel it as a user. vibes are all out of whack bc of it. my opinions above are all likely downstream of this. it’s an issue. - anthropic labs continues to be the goat of ai product. claude design is another hit. it’s fantastic. idk why it’s not talked about more? a+ - updated claude code app is great. i finally switched out of the terminal for it. very well done. - how are people STILL sleeping on the claude agent sdk? i feel like i’m going insane. - gpt 5.5 is incredible. the level to which i trust it for engineering is amazing. if i could only have one model rn, it would be this one just bc of strong need for the coding use case. - codex team is killing it. app has been the gold standard since 5.3 release (buuut i credit conductor team for the ui innovation that everyone is using now). though i could do with a little less passive aggressive shots at ant from the codex team. TARS, dial up class by 30%. it’s a long race guys haha - i uninstalled cursor this month and am now back to vs code for my ide. composer just can’t hang with claude/gpt, and the product feels a bit all over the place. pretty stoked about the xai thing though, because their team is absolutely stacked and i’m excited to see what they might be able to do with that compute. codex and claude code are t1, cursor is t2. i would love if this deal got xai/cursor to t1 for a real trio there. - gemini…? seems like this is 2-3 models now where the model seems like a great release and then nobody ever uses it? i’m bullish google/deepmind but weird it hasn’t translated to product use in any form. kinda disappointed still - no open source models have hit the opus 4.5 level. was hopeful the new deepseek would get there, but nope. good oss agents will have to wait a few more months it would seem…

tokenized equities + VTFs unlock what was impossible before Flowdesk on DimeTV: "the key is getting the smaller investors, those guys who can't access the big private banks, but still want to invest their funds in some sort of managed system" "there's a lot of middle money out there who can't access the big structured products but they want some sort of structured mechanism for investing in equities. this is where (VTFs) become really interesting" VTFs are the next evolution of ETFs watch the full episode below

self-evolving skills in Hermes agent. i found this to be the most powerful feature in Hermes. the agent doesn't just solve problems, it remembers how it solved them. memory handles facts, and skills handle procedures. the difference matters because knowing that a Kubernetes pod crashed is useless without knowing the exact sequence of commands that fixed it. here's how the self-improvement loop works in six steps: 1. the agent encounters a problem 2. works through trial and error: 5+ tool calls, errors, retries 3. finds a working solution 4. calls 𝘀𝗸𝗶𝗹𝗹_𝗺𝗮𝗻𝗮𝗴𝗲(𝗰𝗿𝗲𝗮𝘁𝗲) to save the successful approach as a 𝗦𝗞𝗜𝗟𝗟.𝗺𝗱 file 5. the skill is saved to ~/.hermes/skills/ 6. next session, the agent reads the skill and follows the proven procedure instead of rediscovering from scratch skill creation triggers automatically when the agent completes a complex task, hits errors and recovers, receives a correction from you, or discovers a non-trivial workflow. now here's the important part: without maintenance, agent-created skills can pile up fast. that's why Hermes ships with a Curator that runs in the background, merging overlapping skills, archiving unused ones, and never auto-deleting anything. worst case is archival, and rollback is one command. in this way every hard problem the agent solves once, it solves faster the next time. the agent compounds. i wrote a full deep dive covering hermes agent's self-evolving skills, three-tier memory, GEPA optimization, and setting up multiple specialized agents. the article is quoted below.