ShikumiBot.xyz

@Jason i’m in

We started an AI founder twitter group... reply with "I'm in" if you're a founder and want to be added

Shikumi's daily session analysis is now publicly available. Risk posture, market structure, and execution guidance updated every morning before the New York session. Check it out 👉 shikumibot.xyz/session

4.20.26 memos are now available: > Bitcoin and Infrastructure Lead Crypto Cycle > AI's Value Migration > Stagflation Uncertainty > Systemic Risks Test Institutional Crypto Conviction > Quantitative Alpha Requires Regime Awareness shikumibot.xyz/memos

@pedma7 thanks, that makes a lot of sense. as you built up the infra for your system, were there any key 3rd party components you were comfortable incorporating off the shelf versus the ones you prioritized building yourself?

i think that infra is really important for specific types of trading. for the bulk of what i do now, it certainly is, and ive put literally thousands of hours behind it by now. especially to iterate through ideas fast right. if you spend a week testing every random idea, you wont get much done. i've been realizing over the past 2 years that this is a game of speed if you dont want to be stuck doing the same thing every year. which is fine, but as a retail independent trader i think flexibility is your advantage and you should use it. so you need to be iterating through ideas all the time. not necessarily from published research, as when those are released the edge is significantly decreased if existent at all. most of what i test these days is observations of my own trading other systems, also what i find people talking about here on twitter or when some friend sends me something to look at. but yeh speed is important, see it like a factory and your job is to produce viable trade ideas each week. that way you probably wont go out of business if you know what youre doing.

@pedma7 how often do you find ideas for your systems in published research? seems the infrastructure to test and validate is as much if not more important than the ideas being tested?

good stuff. always important to have good quality data, especially when more granularity is required. but the ultimate test for this is just deploying the system into live market and benchmarking against your model's output going forward. you'll find the delta at you're running at quite fast and then you can use that to model more appropriately. doesn't help must to nerd out on the best quality data and spend months on it when you can get the answers directly from the market imo. still, good to know, bad data will just lead to poor misjudgments on risk and other stuff, so better to take care of that right off the bat. my point is, test in prod.

@baynkr how much of their strategy do you think can be traced directly back to one of these papers?

People want to believe in the agent harness fantasy. Research agent hands off to a trading agent hands off to a review agent, the whole thing runs itself and of naturally posts a 2.5+ Sharpe. Rest assured this is always curve fit noise. Honestly though, what AI is actually good at for traders is the boring tedious shit. It’s a research partner and a reviewer that doesn’t get tired of you asking the same question six different ways. And it’s genuinely changed the velocity of how much you can test and review and actually interrogate. Stress testing a strategy, for example you can quickly test: • how often are my directional calls right • is the timing right, separately from the direction • do i understand magnitude or am i just getting the sign • how much am i giving back in execution • can i even deploy this at size You can run that across a year of trades in an hour now or pull three testable ideas out of a research paper instead of reading it once and forgetting about it. Or find the actual pattern in your PnL instead of the story you told yourself. Boring work. Also the only part that compounds.

The Shikumi trading system research library is now available: ShikumiBot.xyz/library

@MichaelRoyzen what’s the end game? are you aiming to raise a fund or are you pitching hedge funds to adopt this?

But here's the thing most people miss. The hedge fund is a proof of technology. The model that predicts the price of NVDA five minutes from now and the model that predicts a supply chain disruption six months from now are the same model. Different timescales. Same fundamental capability: reading the world and reasoning about what happens next. Energy. Defense. Commodities. Currencies. And every trade we execute generates proprietary data about how our orders actually move the market. That data feeds back into the model. The model gets better. Our predictions get more precise. This flywheel compounds — and the first mover in a strategy has an advantage in becoming the most informed participant and the hardest to displace. This is why being first matters. This is why we're moving fast.

We built a hedge fund where every single trade is made by AI. No human portfolio managers. No manual research. No one writing trading code. Just a reasoning model that reads the world, thinks for itself, and trades before humans can. Early results in testing: 2.55 Sharpe ratio. 2.79 Sortino ratio. Market neutral. $50M+ capacity. Here's how we got here 🧵

Shikumi memos are now available as soon as they are written. Weekly research from curated sources. Powered by ShikumiBot. shikumibot.xyz/memos

BTC just crossed the yearly VWAP. Rejected from last reclaim attempt in March. Will it hold this time?

@realBigBrainAI 🎯

Peter Steinberger, creator of OpenClaw, on why AI agents still produce "slop" without human taste in the loop: "You can create code and run all night and then you have like the ultimate slop because what those agents don't really do yet is have taste." Peter is direct: raw capability without direction still produces mediocre output. "They are spiky smart and they're really good at things, but if you don't navigate them well, if you don't have a vision of what you're going to build, it's still going to be slop. If you don't ask the right questions, it's still going to be slop." Great AI-assisted work is defined by the human guiding it. @steipete describes his own creative process when starting a new project: "When I start a project, I have like this very rough idea what it could be. And as I play with it and feel it, my vision gets more clear. I try out things, some things don't work, and I evolve my idea into what it will become." Most people skip this part entirely, front-loading everything into a single prompt and wondering why the result feels hollow. "My next prompt depends on what I see and feel and think about the current state of the project." Each step informs the next. The work itself is the feedback loop. "But if you try to put everything into a spec up front, you miss this kind of human-machine loop. And then I don't know how something good can come out without having feelings in the loop — almost like taste." The agentic trap is what happens when you remove yourself from the process too early.

Follow @ShikumiDaily for a preview of our morning market briefings. The end-to-end agentic investing system powering it is now available to select development partners. Request access at ShikumiBot.xyz

The AI industry is undergoing parallel structural transitions that reinforce a single investment thesis: the model is no longer the moat. Read More 👇 shikumimemos.substack.com/p/orchestratio…

Shikumi Memos are live on Substack. shikumimemos.substack.com New content every week crafted by the same agentic pipeline that runs our full cycle trading system. fundamentals 🤝 technicals No fluff. No promotion. Analysis by agents, refined by humans.

BlackRock, DTCC, and major exchanges are operationalizing tokenized securities while on-chain metrics and derivatives positioning signal sustained downside pressure through the current cycle. Read More 👇 shikumimemos.substack.com/p/infrastructu…