Most systematic traders start the same way: build one strategy, test it, refine it, and eventually trust it enough to run live. That process teaches a huge amount about strategy design — but it also trains a habit of thinking in terms of one strategy at a time. Portfolio trading asks a different question. Instead of “is this strategy good?”, it asks “what does this account look like once several strategies are running together, and combined?”
The core idea: the portfolio becomes the decision unit
In portfolio trading, the strategies themselves are still important, but they stop being the thing you evaluate directly. What matters is the combined outcome — the account’s combined equity curve, combined drawdown, and combined trade activity, all generated by every member strategy trading at once. A strategy that looks excellent on its own can still be a poor addition once you look at what it does to that combined picture, and a strategy that looks merely decent on its own can be a genuinely valuable addition if it behaves differently enough from what’s already there.
This is a real shift in mindset, not just a change of vocabulary. Reviewing one strategy is a question about that strategy’s own logic and history. Reviewing a portfolio is a question about interaction — how multiple, partly independent return streams combine over time.
Multiple, partly independent return streams
The phrase “partly independent” is doing real work here. It would be convenient if a group of profitable strategies were completely independent of one another — winning and losing on entirely unrelated schedules — but that is rarely true in practice. Markets share macro drivers, volatility regimes affect many strategies at once, and strategies built with similar tools or instincts can end up more related than their symbols or timeframes suggest.
Portfolio trading doesn’t assume independence; it tries to measure how much of it genuinely exists, and to build combinations where the strategies are different enough from one another that the portfolio’s combined behaviour is smoother, more resilient, or simply more diversified than any one member strategy on its own. See Correlation in Algorithmic Portfolios for a closer look at how that measurement actually works, and its limits.
Portfolio-level performance is not an average
A common misconception is that a portfolio’s performance is roughly the average of its member strategies’ individual performances. It isn’t. A portfolio’s genuine performance comes from its own combined trade stream — every trade from every member strategy, placed in chronological order, treated as one continuous sequence. Two portfolios built from strategies with identical average individual metrics can produce very different combined results, depending entirely on how those strategies’ trades line up (or don’t) in time.
The same is true of drawdown, which is explored in more detail in Why Portfolio Drawdown Can Matter More Than Individual Strategy Drawdown. Averaging each strategy’s worst historical drawdown tells you almost nothing about what the combined account actually experienced.
Why this changes what “good” means
Once the portfolio is the decision unit, the definition of a “good” strategy addition changes. It is no longer only about the strategy’s own Profit Factor or win rate — it becomes a question of contribution. Does this strategy add something the portfolio doesn’t already have? Does it tend to be active, or profitable, during periods when the rest of the portfolio is quiet or struggling? Or does it mostly duplicate behaviour that’s already well represented?
Why the Best Individual Strategies Don’t Always Build the Best Portfolio looks at this contribution question directly — it is arguably the central idea behind portfolio trading as a discipline.
Where this fits in the SIPS workflow
SIPS is built around this exact reframing. Portfolio Builder searches a qualified strategy library for combinations, and Portfolio Comparison presents the resulting candidates using genuinely combined-portfolio metrics — not averages of the member strategies’ own numbers. That distinction matters enough that it’s worth stating plainly: whenever you see a portfolio-level Net Profit, Return, or Score inside SIPS, it was calculated from that portfolio’s own combined chronological trade history, the same way a live account would actually experience it.
The practical takeaway
Thinking at the portfolio level doesn’t make individual strategy quality irrelevant — a portfolio can only be as good as the strategies inside it. But it adds a second, equally important question on top: not just “is each strategy good?” but “what does the combination actually look like once they’re all running together?” That second question is where portfolio trading begins, and it is the question every later article in this series comes back to.