SystemTrader's Orion is a residual momentum strategy. It ranks stocks on the part of their momentum that is genuinely their own — the return left over after removing whatever their sector was already doing.
This page explains how Orion separates a stock's own momentum from its sector's, why that distinction changes which stocks get selected, how each stock is matched to a sector from its own price behaviour, and how the portfolio is built and executed.
Ordinary momentum has a hidden problem: when you buy the strongest stocks, you are often just buying whichever sector happens to be running. That works until the sector turns, and then every position turns together. Orion measures each stock against its own sector instead of against the market, and ranks on what is left over.
A stock that rose 30% while its sector rose 30% has shown no individual strength at all. A stock that rose 12% while its sector fell 5% has. Orion prefers the second, and the difference is the whole idea.
Over a long trailing window, Orion measures how strongly each stock has moved with its sector. Some stocks amplify their sector; others barely follow it. That sensitivity is what gets removed in the next step, so it has to be measured per stock rather than assumed.
Each day's return has the sector's contribution subtracted, leaving the portion specific to that company. Summed across the momentum window, this is the stock's own trend — not its sector's trend wearing the stock's name.
The leftover return is divided by how variable it has been. Without this step the ranking collapses onto the most volatile names on the board, whose residuals are large simply because everything about them is large. Scaling by that variability is what makes a quiet, persistent trend rank above a violent, erratic one.
The final days before ranking are deliberately excluded. Very recent moves tend to reverse, so counting them would systematically buy stocks about to give the move back. Orion measures the established trend and ignores the last leg of it.
Two rules deliberately separate the ranking from the portfolio. First, a stock already held stays held while its rank remains respectable, rather than being sold the moment something scores fractionally higher. Ranks move around constantly, and trading on every small reshuffle costs far more than it earns.
Second, no single sector may fill more than a set number of slots. Without that cap, a strategy built to find stock-specific strength can quietly end up concentrated in one corner of the market anyway — which is the exact outcome the scoring was designed to avoid.
Orion trades the most heavily traded names, chosen by rank rather than by any fixed dollar threshold. This matters more than it sounds. A fixed cut-off means something different every year: a threshold that admits a few hundred stocks in one period can admit twice as many in another, quietly changing the strategy without anyone deciding to change it.
Ranking sidesteps that entirely. The universe is the same size in every market environment, so the strategy's historical record reflects its rules rather than the drift of a number chosen years ago.
Each stock is matched to the sector it actually trades like, measured from its own price history, rather than to whatever sector a data provider has filed it under. Classifications are often stale or debatable — a company can be labelled one thing and trade like another for years.
Since the whole strategy depends on removing the right sector from each stock, that match has to be right. Stocks whose behaviour matches no sector clearly enough are grouped together and capped as one, rather than being forced into a category they do not belong in.
Scores are computed after the close using that day's completed data, and any resulting trades are executed at the next morning's open. Signals are published in the evening, giving you the full night to place orders before the market opens.
Nothing is ever decided using a price that had not yet printed when the decision was made. Every input to today's ranking was available to you at the same moment it was available to the strategy.
Positions are held in equal weight and allowed to drift; they are only resized once they have moved meaningfully away from target, so normal day-to-day movement does not generate trades.
Overall exposure is scaled down when short-run market volatility rises above its own longer-run level, and restored as conditions settle. The overlay applies to the entire book rather than only to new purchases — a distinction that matters, because scaling only new entries would leave existing positions at full size through exactly the periods the overlay exists to survive.
All performance figures are based on historical backtesting and are hypothetical. Past performance does not guarantee future results. Removing sector exposure reduces the risk of every position turning at once, but it does not remove market risk — Orion can and does lose money, and its worst historical stretch was a drawdown of roughly a fifth of the portfolio. The backtest assumes execution at open prices with no commission, spread, or slippage; real-world trading will differ, and it will differ more for a strategy that trades often. This is educational content, not investment advice.