Rfactory Labs: What is a system?

Members ยท 01

What is a system?

The orientation page showed a bench of specialist systems run by one Allocator. This piece zooms in on a single box on that bench: what a trading system actually is, the five decisions every one of them must answer in writing, and three worked shapes that show how far the same skeleton stretches.

The definition

Five decisions, written down

Every system, from a monthly rotator to an intraday scalper, is a complete written answer to five questions. If any answer is missing or vague, it is not a system yet. It is an opinion.

WHERE the universe which markets it may trade WHEN the schedule when it wakes up and looks WHAT the signal what triggers action HOW MUCH the sizing what it risks per position WHEN OUT the exit what ends the trade all five, written precisely enough that a machine can follow them without asking you anything
The five decisions. Writing them down precisely is what turns an idea into a testable system.

The reason for this strictness is the gauntlet from the orientation page. A backtest can only judge rules that are written down. "Buy strong stocks and sell when it feels toppy" cannot be tested. "Hold the three strongest of these ten ETFs, re-ranked on the first trading day of each month, sized equally" can.

The range

The same skeleton, three very different animals

Here are three shapes we will build out fully in later pieces, using well-known public-domain rules. Look at how differently they answer the five questions, and then at what they still have in common.

Monthly rotationDaily swingIntraday breakout
Wherea fixed menu of ~10 large ETFsa shortlist of liquid US stocks, re-screened as it goesone index future
Whenfirst trading day of each monthevery evening after the closeevery morning after the open
Whatrelative strength: hold the leadersa sharp pullback inside an uptrenda break of the opening range
How muchequal weight across holdingssmall fixed risk per positionsmall fixed risk, one trade at a time
When outnext month's re-rankingstrength returns, or a few days passtarget, stop, or the closing bell
Rhythm~12 decisions a yeardozens of trades a yearhundreds of trades a year
Weather it likessteady, persistent trendsupward but choppy marketshigh-energy, fast-moving days
week 1 week 2 week 3 week 4 Monthly rotation re-rank, rebalance, done then holds all month Daily swing each bar is one position, held 2 to 5 days Intraday breakout short bursts inside single days, always flat by the close
One month in the life of each shape. Same machinery underneath, completely different rhythm on top.

None of these is "the best one". That is the wrong question. They make and lose money at different times, in different conditions, and that disagreement is exactly what the bench is for: a portfolio of specialists is steadier than any single specialist, however good.

The point

Why one skeleton can run all three

Underneath, all three are a page of strategy rules sitting on the same shared foundation you saw in the orientation: identical risk limits, identical honest cost assumptions, identical safety checks, and every result recorded in the same database. That has two consequences worth internalising.

First, adding a new shape is cheap. The intraday system did not need its own risk engine or record-keeping; it needed its five answers written down. Second, the Allocator can compare them fairly. Because a monthly rotator and an intraday breakout report through the same machinery, in the same units, "who is earning their capital?" is one question with one honest answer, not three incompatible spreadsheets.

Next in the series: the tech stack this all runs on. After that, each of the three shapes above gets built in the open as a full worked example, public-domain rules and all.