Systematic strategy monitoring is the process of comparing a strategy’s current behaviour with the evidence and operating range that justified its deployment, while distinguishing normal variance from potentially meaningful change.
Why monitoring is harder than watching P&L
A strategy can lose money without being broken, and it can make money while becoming less reliable. P&L is an outcome; monitoring should examine the behaviour behind that outcome. The useful question is not simply “is this strategy up or down?” but “is the strategy still behaving in a way that is consistent with the evidence we approved?”
For a systematic team, that can include changes in realised volatility, drawdown depth and duration, hit rate, payoff distribution, turnover, exposure, correlation to other strategies, sensitivity to market conditions and the stability of the signals that drive the model.
Start with an explicit behavioural baseline
Monitoring requires something to compare against. A baseline can be derived from research, walk-forward validation, paper operation and subsequent live history. The exact metrics depend on the strategy, but the principle is consistent: define what “normal” looks like before the system has to decide whether current behaviour is unusual.
Expected range
What variation was visible in testing and validation?
Recent state
How is the strategy behaving now relative to its own history?
Portfolio effect
Does the change matter more because of the rest of the portfolio?
Behaviour drift is a decision input, not an automatic stop signal
“Drift” can mean many different things: a gradual change in return distribution, rising correlation, reduced signal efficacy, execution slippage, increased tail losses, or sensitivity to a market regime the strategy was not designed for. A monitoring system should therefore avoid collapsing every change into a binary good/bad flag.
Prospect’s intended approach is to represent change as evidence that can alter a strategy’s reliability view. For a more technical breakdown of the signals and methods involved, see strategy drift detection. That view can then interact with portfolio-level constraints before any capital response is proposed.
Example: a strategy entering a historically plausible drawdown may require no allocation change. The same drawdown combined with rising correlation, abnormal volatility and a poor regime fit may justify a different review.
What to monitor
From monitoring to capital response
The objective is not to generate more alerts. It is to make the next portfolio decision clearer. A useful monitoring layer should show what changed, why it matters, how confident the system is, which portfolio constraints are relevant and what action — including no action — is justified.
That is why Prospect treats strategy monitoring as the first wedge into a broader decision layer. Once the system can understand how each approved strategy is changing, it can connect that evidence to portfolio risk and explainable capital allocation.