Here's a question that keeps quant traders up at night: if you've got two solid signals — say, a trend-following indicator and a carry signal — should you blend them equally, or assign weights based on some measure of quality? It sounds straightforward. It absolutely is not. This is one of those decisions that quietly determines whether your portfolio is resilient or fragile.

The direct answer is this: equal-weight (unweighted) aggregation is almost always more robust in live trading, while optimised weighting can look spectacular in backtests and quietly fall apart when markets change regime. Think of it like a recipe — a skilled chef might swear by their precise 60/40 ratio of spices, but a good home cook who just adds roughly equal amounts consistently makes a decent meal every single time.

CONCEPTEqual-weight signal aggregation sacrifices theoretical optimality for real-world robustness — and that trade-off usually pays off.
WARNINGHistorically optimised signal weights are fitted to noise as much as signal — expect them to degrade out-of-sample.
KEY IDEATrend and carry signals have low correlation across regimes, making their combination more powerful than either alone — regardless of weighting method.

The reason this matters so much in multi-asset portfolios is signal correlation. Trend-following signals — which bet on price momentum continuing — and carry signals — which bet on high-yield assets outperforming low-yield ones — tend to perform at different times. When trend struggles (choppy, mean-reverting markets), carry often holds up. That diversification effect is the real prize, and you capture most of it just by combining them at all.

Signal Aggregation: Sharpe Ratio Comparison 1.2 0.9 0.6 0.3 1.20 1.00 In-Sample 0.38 0.79 Out-of-Sample Optimised Weight Equal Weight OOS Degradation

When researchers at AQR studied diversified trend and carry combinations across equities, bonds, commodities, and currencies, the consistent finding was that the diversification benefit dominated the weighting method. Optimised weights extracted maybe 5–10% extra Sharpe in-sample, then underperformed equal-weight out-of-sample. The estimation error in deriving those weights swamped the theoretical gain. It's Stein's paradox dressed up in a trading suit.

Where weighted aggregation does earn its keep is in risk-adjusted signal scaling — not static allocation, but dynamic adjustment based on realised volatility or signal strength. Volatility-targeting each signal before aggregation is a form of implicit weighting that genuinely improves outcomes, because you're correcting for a real, observable characteristic rather than fitting historical correlations. That's a meaningful distinction. For a deeper grounding in how signals combine, the momentum investing framework on Investopedia explains the underlying logic well, while the carry trade entry on Wikipedia covers why carry generates persistent premia. For the statistical argument against over-fitted weights, Investopedia's explainer on overfitting is a sharp refresher.

The practical takeaway: start with equal-weight combination of your trend and carry signals, volatility-scale each one individually, then track whether adding weights actually improves out-of-sample performance over 12–24 months before committing to them.

Complexity is only your friend when it earns its seat at the table — and in signal aggregation, simplicity usually outperforms the impressive spreadsheet.

This content is for educational purposes only and does not constitute financial product advice. Past performance is not indicative of future results. Profit Logic Ltd (ACN 688 669 936) accepts no responsibility for errors or omissions in this content or anywhere on this website. Always seek advice from a licensed financial adviser before making investment decisions.