Rfactory Labs: The Factory, from the top

Rfactory Labs · Orientation

One portfolio. Many systems.
One decision-maker.

Rfactory Labs: a front-row seat on the build process and the research journey behind it.

Rfactory is building a fully automated portfolio of trading systems, run the way a small fund runs money, but in-house. Not one strategy you hope keeps working: a bench of specialist systems, each with a narrow job, coordinated by a single controller that decides who trades and with how much. The portfolio itself runs our own capital and stays in-house. It isn't a product. What we're opening up is Rfactory Labs: a front-row seat on the build process and the research journey behind it.

The big picture

The whole factory on one screen

Three layers. Systems do the trading. The Allocator decides who gets capital. The Market Compass tells the Allocator what kind of market we're in.

MARKET COMPASS reads market conditions: trending · choppy · calm · crisis market weather THE ALLOCATOR Always on. Always watching. Decides which systems get capital, how much, and when. watches risk, correlation and behaviour across the whole bench live results & risk report back capital MONTHLY ROTATION stocks & ETFs rotates monthly TREND FOLLOWING futures holds weeks to months MEAN REVERSION stocks holds days INTRADAY BREAKOUT index futures minutes to hours CRYPTO TREND crypto daily orders BROKERAGE ACCOUNTS: DIFFERENT TYPES, DIFFERENT RULES Retirement account no leverage · long only Margin account stocks · some leverage Futures account leverage · strict limits several systems can share one account as separate "sleeves", and one system can run on more than one account
The three layers of the factory. Capital flows down from the Allocator; live results and risk flow back up. The five systems shown are examples of the kinds of specialists on the bench. The real bench grows over time.

The systems

Each system is one specific strategy: one edge, one set of markets, one timeframe, from monthly rotation down to intraday. No system is a hero. Each contributes a small, measured edge, and because they behave differently in different conditions, the portfolio is steadier than any one of them alone.

The Allocator

Think of it as the CIO of an in-house fund. It never places a trade itself. It hires, sizes and benches systems: more capital to what's earning it, less to what isn't, and an eye on the whole book so two systems don't quietly make the same bet twice.

The Market Compass

A weather station for markets. It classifies the current regime (trending, choppy, calm, crisis) so the Allocator can lean toward systems built for this weather and away from those built for different weather, before the P&L forces the issue.

One level down

Each system is small on purpose

A trading system in this factory is just the strategy rules: roughly a page of logic. Everything hard, dangerous or repetitive lives in a shared foundation that every system inherits, including risk limits, honest cost assumptions, safety checks and record-keeping.

System A just the strategy rules System B just the strategy rules System C just the strategy rules every system inherits all of this, automatically SHARED FOUNDATION: BUILT ONCE, USED BY EVERY SYSTEM risk limits realistic fees & slippage safety checks & tests every result recorded
Thin systems on a fat foundation. The strategy is the only thing that changes from system to system; the machinery underneath is identical for all of them.

This is the engineering discipline that makes a portfolio of systems possible for a small operation. Adding system #10 costs a fraction of what system #1 cost, because the hard parts already exist and are already tested. Just as important: no system can quietly cheat. It can't skip the risk limits, trade without cost assumptions, or forget to record a losing result. The foundation doesn't give it the option.

The discipline

How a system earns real money

Anyone can show you a beautiful backtest. The value is in the gauntlet a strategy has to survive before it touches a live account, and most don't survive it.

Idea hypothesis, not hope Build on the foundation Backtest realistic costs, always Out-of-sample data it has never seen Walk-forward rolling re-test, like live Incubation small real money Live earns its allocation most ideas are rejected here. That is the system working, not failing
The gauntlet. A portion of history is locked away before testing begins, so every system must eventually prove itself on data it was never tuned on. Even after going live, a system keeps re-earning its capital: the Allocator benches what stops working.

The Lab

What membership actually is

Rfactory Labs is a seat inside this build: watching it happen in near real time, with the reasoning attached. It is immersion and collaboration, not a course and not signals.

What's included: watch, learn, ask

  • Weekly lab notes: what was tested, what worked, what failed, and why (including the ugly results)
  • The architecture and the thinking, deep enough to build your own version of the factory
  • Public-domain systems worked end-to-end as real examples: dual momentum, opening-range breakout and others
  • Tearsheets, post-mortems, and honest reviews when things break
  • Live sessions and a small community of serious builders

What's not included

  • Trade signals: no entries, exits, stops or targets, ever
  • Personalised investment advice
  • The production codebase, the proprietary systems, or their live parameters
  • Guarantees: markets don't offer them, and neither do we

The factory itself, meaning the codebase and the production systems, isn't part of the membership. The learnings from watching it get built are: the architecture, the technology, the thinking, so you can implement your own version for your own situation. And if you'd like hands-on help building out your own factory, we're happy to discuss that as a separate engagement.