The Idea

ℹ️

This project starts from a simple but somewhat unfashionable belief: most alpha-seeking funds either fail outright or fail to justify cost and complexity over time. A well-constructed beta and factor portfolio already does a lot of the heavy lifting if it is built and implemented with discipline.

We are setting out to make the Shuhari.Capital engine a multi-strategy portfolio system. Instead of picking one approach to investing for all market conditions, it combines different "Books" (more on that in "Our Portfolio Books"), each designed to target different KPIs.

The conceptual work I wrote for the refactoring of our portfolio ("The Stability Core + Momentum Chaser" portfolios) had an ambitious scope:

a Barra-style fundamental equity factor model with full covariance matrix,
a BIS-style arbitrage-free term-structure model for the macro sleeve,
and an alpha lab covering five distinct alpha sources ranging from microstructure to behavioral to structural to domain-specific.

I scoped most of it down. This post is about what I cut and why, because scope discipline tends to be where solo projects in this domain go to die.

We will be building a "Beta / Factor Exposure" book, aiming to optimise the way we are exposed to market beta and factors, predominantly acting as the more stable baseline, as well an "Alpha Book", compiled of strategies that aim to capture residual alpha after stripping all factors.

We then assign a sub-portfolio weight to each strategy and combine them into a single "fund" portfolio.

Personal Context

Personally, I started learning more about "tactical asset allocation"(TAA) in contrast to stock-picking while experimenting with Quantopian (later QuantConnect) about 10 years ago. It's a quantitative research and trading engine that can be programmed in Python or C# and got me hooked on the idea of systematic investing.

The idea of making deliberate, temporary shifts away from a long term "core" strategy to reap returns or sidestep some risk along the way really resonated with me. Mebane T. Faber's research paper "A Quantitative Approach to Tactical Asset Allocation" (2006, The Journal of Wealth Management, Spring 2007) set me off to spend the coming years venturing into this field as a weekend hobby.

When then developing the Shuhari.Capital engine, one of my main drivers was to cap the downside through TAA-typical momentum mechanics, which wave historically shown around 50% reduction in drawdowns. Putting real funds on the line is stressful even in good times. Limiting the exposure to buy-and-hold drawdowns I realised will strongly reduce the likelihood of me making mistakes, even if it comes at the cost of higher returns.

The Philosophy

A few principles that have emerged to guide how I build this:

Rules over gut feel. Every decision the system makes is based on a defined rule with a defined threshold. I do not override the decisions, except from "pausing" the system when we're entering regimes I know I have not accounted for in the model. This removes the emotional aspect and makes our results measurable.

Diversify across strategies, not just assets. Owning 500 stocks doesn't help if they all go down together. This portfolio diversifies across approaches using macro allocation, trend following, sector rotation, and stock selection. Since different strategies work in different market regimes, true diversification also means ensuring that each strategy contributes equally to portfolio risk, not just to capital. (See "ERC" above.)

Respect the research. The strategies aren't invented from scratch or based on any sort of "magic formula" that I claim I have found. They're grounded in academic and practitioner research like tactical allocation, (dual-/multi-) momentum or the CANSLIM methodology. I've adapted them for our system where needed, but the foundations are documented, not invented.

Transparency over performance marketing. I will publish backtests and results, the monthly portfolio and the economic data used. (Code TBD). If the model has a bad month, you'll see it.

Portfolio Design

So we treat the portfolio as two separate problems:

(1) Building a "Factor Book" that harvests market, style, sector, and macro premia using established, academically researched approaches.

(2) Testing whether there is any residual alpha left beyond those exposures and building an "Alpha Book" to try and capture some of it. (Yes, we couldn't resist...)

That separation is not just conceptual hygiene. It avoids calling factor exposure "skill" and reflects the basic logic of alpha-beta separation. They are different engines with different constraints, capacities, and expectations.

I'm not under any illusion about who does this best. There are firms like Citadel, Millennium and Point72 that operate at a level that is simply not accessible without their funding, institutional data, and large teams of quantitative analysts, traders and risk managers.

This project is built with a different constraints. No privileged access, bespoke datasets, or a floor full of PMs. What we have is public data, research and somewhat modern tooling. (Which I might be naïve about.)

The goal is not to replicate institutional multi-strategy funds. It is to build a system that works under my individual capital constraints and is built transparently, rules-based, and robust enough for me to trust it to manage my own money.

Let's take what is documented, strip away what we can not replicate, and implement what remains.

What's Next

I'll be writing dedicated chapters on each strategy, including the specific signals, thresholds, and trade-offs. These will appear as blog posts and be linked from this page as they're published.

If you want to follow along: subscribe to the blog or check the Economic Pulse series for the macro data that feeds our engine.