About this site
Shuhari.Capital is an independent publication launched in October 2025. If you subscribe for free today, you'll get full access to the website as well as email newsletters about new content when it's available. Thank you!

Shuhari (Japanese: 守破離) is a Japanese concept that describes the stages of learning to mastery.
Disclaimer First
Everything on this site documents my personal portfolio and research journey. None of it is financial advice. I am not a licensed financial advisor, portfolio manager, or investment professional.
The portfolio I run is my own and I publish it transparently because I believe in open research and the benefits of getting to know other people on a similar journey, not because I'm recommending anyone replicate it.
Past performance, including the backtest results I use for my research that are shown on this site are not indicative of future results.
Do your own research. Make your own decisions.
So what is Shuhari.Capital?
I work in fintech (Open Banking) and have been hobby-programming in Python for years, building apps for personal use and/or early MVPs for startup ideas. I've always been fascinated by the intersection of code and markets and started tinkering with a data collection and stock screening dashboard as a side project. Nothing fancy. Just filtering stocks by a handful of financial metrics to see what floated to the top.
The first versions were rough around the edges, but they got me hooked. I started reading research papers on momentum, tactical asset allocation, and sector rotation. (O'Neil, Meb Faber, Gary Antonacci for example.)
Over time, the screener turned into a portfolio system, with more and more macro signals and layers to diversify. (Since a few versions before the inception of this blog, I signed up for an "Ultimate" tier of "Financial Modelling Prep" API (not affiliated!), which I found very useful so far to fetch the economic and fundamental company data needed for my decision making.
Last year, I (reluctantly) started using Claude Code on weekends to accelerate the development, and that changed the pace entirely. I'm not a professional developer, so the project grew to a scale that became harder for me to handle as a hobby and I have had the chance to learn about the ins and outs of Claude through this application.
The core idea has stayed the same through the iterations: I want to invest with a defined, automated set of rules to remove the emotional inertia that has killed my returns in the past. (Which I have then often tried to make up with erratic action, causing even more harm.) I wanted something between "buy a world equity ETF and forget about it" and "actively trade based on discretionary factors". Something that respects the academic evidence on momentum, diversification, and risk management, but is practical enough to actually run as a non-professional investor.
Where are we today?
I ran an early version of the portfolio engine for several months last year and in Q1 2026 to test some of my hypotheses, and it confirmed the direction I intended to take.
In June 2026 I rebuilt it and let it trade every morning on a paper account at Interactive Brokers: real orders against real prices, with no money behind them. Over the summer our daughter was born and the project rested. In November, before funding it, I laid all its parts side by side for the first time. Only two of its six fixed sleeves had made money against the S&P 500, and the mix as a whole had never been tested. So I started over (see Starting over).
The new engine has one job: no rule gets money until it has passed a test it could have failed (see The design). What passed is now running, in one account at Scalable Capital:
- Half the money in a passive pot: Scalable's five published portfolios, held as their own funds and topped up every month by savings plans.
- 35% in a trend rule: at each month-end it ranks seven markets (US shares, other developed markets, emerging markets, US property, commodities, gold and bitcoin) by their recent trend and holds the best three. A warning signal from US inflation-linked bonds can move the share money into safe bonds or cash for a month.
- 15% in a managed-futures fund, run by its manager and bought every month like the passive funds.
A bond ladder at another broker is tracked as income, outside both pots.
The first half of the starting money went in on 16 November 2026; the rest follows in December. From November 2026, each Economic Pulse issue reads the month's economy through the numbers these rules act on, and ends with what the rules decided and how each pot did against its yardstick.
What this blog can become
I started the blog with the Economic Pulse series to get used to interpreting economic data and putting it into words. Disclaimer: I do use Claude to work on my drafts here, but am explicitly not blindly releasing AI work as it would be against the goal of this exercise. With a few of these released, I will start to add the following in 2026:
- The investment concepts: explaining the financial theory behind the system in plain language and referring to the literature I have taken inspiration from. Not academic, but enough to understand why the model does what it does. Start with the Concepts page.
- The development journey: as I am not a professional programmer, this project is as much a learning experiment in programming and application design as it is for financial investing. The technical decisions, the architecture, the mistakes, I will try and put a lot of it in writing here as well. Maybe it can be helpful to someone on a similar path.
- The monthly rebalancing and results: I will publish the monthly portfolio compilation and the results over time, as well as my interpretations of what went well and what did not.
The goal is simple: get more people interested in thinking about investing beyond "buy world equities and forget about it." I think there is an interesting middle ground between passive and active investing that's worth exploring, and the tools to do it yourself have become very accessible these days.
Find Me Elsewhere
GitHub: shu-ha-ri (most projects are private, I might open-source more over time)
Medium: @shuhari (more personal writing coming soon)