I get emails every week from traders who’ve spent months perfecting a single strategy. Great backtest. Smooth equity curve. Impressive Sharpe ratio. They go live — and within weeks, the strategy stalls. Within months, it’s in a drawdown deeper than anything the backtest showed.
The strategy wasn’t broken. The market regime changed. And that’s the point.
Every strategy captures a specific market effect. Mean-reversion works in choppy, range-bound conditions. Momentum works in trending conditions. Volatility selling works in calm conditions. None of these effects persist all the time — which means every single-strategy trader is guaranteed to face extended periods where their edge disappears.
This is why professional traders — the hedge funds, CTAs, and serious independent traders I’ve worked with over the past decade — don’t trade one strategy. They trade portfolios of strategies.
The goal isn’t one perfect system. It’s a team of robust systems that take turns carrying the load.
The Power of Taking Turns
Think of it like an engine. Pistons fire at different times — some going up while others go down — but the car moves forward smoothly. That’s what a strategy portfolio does.
A mean-reversion strategy might lose during a strong trending month. But a momentum strategy in the same portfolio is having its best month of the year. A carry strategy is earning its steady yield. A seasonal strategy fires on an end-of-month flow pattern. The portfolio as a whole stays positive.
The math behind this is more powerful than most traders expect. I’ve demonstrated examples on the Build Alpha portfolio trading page where three strategies — each with 40%+ maximum drawdowns individually — combine into a portfolio that never exceeds a 5% drawdown. Not because the strategies are individually great, but because their drawdowns don’t overlap.
This is what low correlation buys you: massive drawdown compression.
How Sharpe Scales (This Changed My Thinking)
Here’s a number that reframed my entire research process. For a portfolio of n uncorrelated strategies, each with Sharpe ratio S, the portfolio Sharpe scales with the square root of n:
Four strategies with individual Sharpes of 0.50 — mediocre by any standard — combine into a portfolio Sharpe of 1.0. That’s institutional quality, from strategies you’d barely trade alone. Nine uncorrelated strategies at 0.50 each? Portfolio Sharpe of 1.5. Sixteen? Sharpe of 2.0.
The catch is that strategies are never perfectly uncorrelated. With moderate correlation (ρ = 0.2), the scaling still works but flattens — you might double your Sharpe with 16 strategies instead of quadrupling it. At high correlation (ρ = 0.5), adding beyond 4–5 strategies barely helps. You’re just repeating the same bet.
The practical takeaway: one hour finding a new uncorrelated strategy is worth more than ten hours improving an existing one. The portfolio Sharpe improvement is structural. The single-strategy improvement from parameter tweaking is fragile and often overfit.
The Strategy Styles That Create Real Diversification
When I say “uncorrelated strategies,” I mean strategies that capture fundamentally different market effects. Here are the major ones, and why each matters in a portfolio context:
Trend following profits in directional markets. Low win rate, big winners. Near-opposite of mean reversion, which profits when prices overshoot and snap back — high win rate, small winners, occasional large loss. Carry earns yield differentials (interest rates, roll yield) and provides steady returns in calm environments but reverses sharply during risk-off events. Volatility strategies trade vol levels or structure — short-vol earns premium in calm markets while long-vol hedges crises. Seasonal strategies exploit recurring calendar patterns (end-of-month flows, holiday effects, FOMC cycles) driven by institutional behavior rather than price patterns. Event-driven strategies trade around catalysts like earnings or economic data releases.
Each style tends to make money when the others don’t. That’s not a coincidence — it’s the structural reason portfolios work. A portfolio spanning 3–4 of these styles across multiple asset classes achieves the low correlation that makes the Sharpe scaling math work.
Parrondo’s Paradox: When Losers Become Winners
Here’s where it gets really interesting. Under the right conditions, you can combine two losing strategies into a winning portfolio.
This isn’t theoretical. It’s a well-documented phenomenon called Parrondo’s Paradox — the idea from game theory that two losing games can combine into a winning game. In trading, the mechanism is rebalancing.
When you periodically reset your capital allocation between strategies, you mechanically sell what went up and buy what went down. If the strategies are uncorrelated and volatile enough, this creates a rebalancing premium that can flip negative returns positive.
The catch: it only works with sufficient volatility, low correlation, and manageable transaction costs. It’s not magic — it’s math. And it requires proper portfolio simulation with shared capital to model correctly.
The Mistake Most Traders Make With Portfolios
Most “portfolio backtesting” simply merges trade lists from separately backtested strategies. Strategy A had these trades. Strategy B had those trades. Add them up.
This doesn’t work because it ignores that strategies share capital. In a real account, profits from Strategy A free up capital for Strategy B — potentially creating trades that never appeared in the standalone backtest. And drawdowns in Strategy A reduce the capital available for Strategy B, changing position sizes.
True portfolio simulation requires re-running all strategies simultaneously with shared capital. This is why I built Build Alpha’s portfolio mode to re-simulate rather than merge — the results can differ significantly.
Getting Started
If you’re trading one strategy today, you don’t need to overhaul everything. Start by identifying which style your current strategy falls into (trend, mean reversion, carry, volatility, seasonal, event-driven). Then look for a strategy from a different style that thrives in the conditions where yours struggles. Measure the correlation between them. If it’s low, you’ve found your first portfolio diversifier.
Then test the portfolio as a whole — with Monte Carlo analysis, noise testing, and proper robustness checks applied at the portfolio level, not just the individual strategy level.
For a complete walkthrough of portfolio trading concepts — including Shannon’s Demon, Sharpe scaling tables, and every portfolio feature in Build Alpha — check out the full Portfolio Backtesting & Trading Guide.
Twitter: @DBurgh @buildalpha
No position in any of the mentioned securities at the time of publication. Any opinions expressed herein are solely those of the author, and do not in any way represent the views or opinions of any other person or entity.
