Wick is in development, not yet a registered investment adviser, and currently offers no advisory services. joining the waitlist creates no advisory relationship.
Chapter 7

Managing a live algo strategy

how to read the signal. not the noise.

Running a live algo strategy means making real decisions: what to monitor, how to read a drawdown, and when to act. Done well, it's less about reacting to what happens and more about knowing in advance what to expect — and what to do when you see it.

What to monitor — and what not to

What should you monitor when running a live algo strategy?

The most common mistake when managing a live strategy: checking it too often. For longer-timeframe strategies, daily P&L is noise. A strategy that rebalances monthly will have dozens of days where it looks wrong before it looks right. The right cadence for review depends on the strategy's timeframe — an end-of-day strategy might warrant a weekly check-in; an intraday strategy warrants more frequent attention.

What to actually monitor

Compare live performance against the track record — not just the return, but the drawdown profile and trade frequency. Is the strategy behaving the way the track record said it would? Small differences are normal. Large, persistent gaps warrant investigation.

Watch for changes in market regime. A momentum strategy will underperform in a choppy, mean-reverting market. That's not strategy failure — it's expected behavior. Understanding why a strategy is underperforming matters more than reacting to the fact that it is.

Market regimes are the context in which a strategy operates. A regime is a period of broadly similar market behavior — trending, mean-reverting, high-volatility, low-volatility. Most strategies are designed for specific regime conditions, even if that's not explicit in how they're described. A momentum strategy is implicitly designed for trending regimes. A mean reversion strategy is implicitly designed for range-bound ones.

Regime shifts are normal and expected. What changes when a regime shifts is which strategies are likely to perform and which are likely to struggle — not whether any strategy is fundamentally broken. The key question is: is this strategy underperforming because conditions are temporarily unfavorable, or because its underlying edge has decayed? See types of algo trading strategies for a breakdown of which types perform best in which conditions. Comparing a strategy's behavior against the prevailing market regime — whether the period has been trending, choppy, or high-volatility — can help distinguish regime-driven underperformance from strategy decay.

Normal drawdown vs. strategy decay

How do you tell a normal drawdown from strategy decay?

Most traders react to drawdowns. Knowing what a normal drawdown looks like — and having the framework to read it — is what separates managing a strategy from just watching it.

Every algorithmic trading strategy draws down at some point. The track record tells you the historical max drawdown — but in live markets, that number can be exceeded. The question is whether the drawdown is within the expected character of the strategy, or whether something has fundamentally changed.

normal drawdown vs strategy decay — what to look for
Normal drawdownWithin historical range. Strategy still trading as expected. Market conditions explain the underperformance. Track record suggests it has recovered from similar periods before.
Worth investigatingDrawdown significantly exceeds historical worst. Trade frequency or behavior has changed. Live returns persistently diverging from track record. Market regime has shifted structurally.

Time is the best diagnostic. A strategy drawing down in conditions that have historically caused drawdowns is not necessarily in decay. A strategy that underperforms persistently across varied market conditions — including periods where it historically performed well — warrants a harder look. The relevant timeframe depends on the strategy's own cadence: a two-week drawdown means something very different for a monthly-rebalancing strategy than for an intraday one.

When to stop

When should you stop an algo trading strategy?

There's no formula for knowing when to stop an algorithmic trading strategy — but there are clear signals. Stop when the strategy's live behavior diverges materially from its track record: different trade frequency, drawdowns that exceed the historical profile, or persistent underperformance across conditions where it historically performed well. The principles below give the framework.

Stop if your stop-loss hits
You set a stop-loss threshold before going live for exactly this reason. If it triggers, honor it. The purpose of a stop-loss is to remove the decision from the moment of maximum emotional pressure.
Stop if the logic no longer makes sense
If the market condition the algorithmic trading strategy was designed to exploit has structurally changed — and you can articulate why — that's a rational reason to stop. Not because it's losing, but because the hypothesis no longer applies.
Drawdown discomfort is not the same as strategy failure
Stopping a strategy during a normal drawdown locks in a loss and removes the possibility of participating in the recovery. A drawdown within the historical range of the track record is evidence the strategy is behaving as expected — not evidence that something has gone wrong.
Recent performance and future performance are different things
A strategy that looks best right now has typically been performing well recently. Recent performance is real — but it reflects what conditions have been, not what they will be. Understanding the relationship between a strategy's type and current market conditions is more informative than comparing recent returns in isolation.

Knowing when to stop is the hardest call. Knowing when to hold — because the strategy is behaving exactly as its track record said it would — is the skill that compounds over time.

Adding a second strategy

When and how should you add a second algo strategy?

Running multiple uncorrelated strategies is one of the most effective ways to manage portfolio-level risk. When two strategies are uncorrelated — meaning they don't tend to lose at the same time — a drawdown in one may be offset by stable or positive performance in the other. The result is a smoother overall equity curve without necessarily sacrificing return.

Check correlation before adding anything. Two momentum strategies targeting the same market will tend to lose at the same time — that's concentration, not diversification. Momentum strategies across different asset classes or timeframes may be less correlated. When evaluating a second strategy, apply the same framework you used for your first — see how to evaluate an algo trading strategy for the metrics that matter.

The second strategy is easier than the first. You have a baseline, you understand your own risk tolerance, and you've seen what a real drawdown feels like. That's experience most traders don't have — and it makes every subsequent decision sharper.

Common questions
Managing a live algo strategy — FAQs
When should I stop an algo trading strategy?
Stop if your pre-set stop-loss threshold is hit — you defined that number for exactly this situation. Stop if the market condition the strategy was designed for has structurally changed and you can clearly articulate why. Don't stop because the strategy is uncomfortable during a normal drawdown, and don't stop to switch to a strategy that has been performing better recently. Chasing recent performance is one of the most reliable ways to underperform.
Is it normal for an algo strategy to have losing weeks?
Yes — every algo strategy has losing periods, including the most successful systematic funds. What matters is whether the drawdown is consistent with the strategy's historical behavior. A losing week during market conditions that have historically caused the strategy to draw down is expected. A losing week during conditions where the strategy historically performed well is worth investigating.
What is correlation in a multi-strategy portfolio?
Correlation measures how closely two strategies' returns move together. Strategies with low or negative correlation tend to lose at different times — when one draws down, the other may be flat or positive. Running uncorrelated strategies together reduces portfolio-level volatility without necessarily reducing overall return. Before adding a second strategy, check that it is not highly correlated with the first.
What is strategy decay in algo trading?
Strategy decay is when a strategy's edge gradually diminishes — usually because the market conditions it was designed to exploit have changed, because the underlying relationship it relied on has broken down, or because similar approaches have become more crowded. Signs include live performance persistently below the track record across varied conditions, significant changes in trade frequency, and drawdowns exceeding the historical worst. Underperformance in conditions that have historically been difficult for the strategy is not decay. Persistent underperformance across varied conditions — including periods where the strategy historically did well — warrants closer examination.
How often should I check my algo trading strategy?
It depends on the strategy's trading frequency. A monthly-rebalancing strategy doesn't warrant daily monitoring — less frequent check-ins are appropriate. An intraday strategy warrants more regular attention. The main thing to avoid is checking too often and reacting to normal short-term noise as if it were a signal. Set a review cadence that matches the strategy's timeframe and stick to it.
Can I run multiple algo strategies at the same time?
Yes. Running two uncorrelated strategies — momentum and mean reversion, for example — reduces portfolio drawdown depth because they tend to underperform in different conditions. Before adding a second strategy, check correlation: two momentum strategies targeting the same market will tend to lose at the same time. Check the asset class, strategy type, and trading timeframe of any strategy you're considering adding.
key takeaway
most traders stop a strategy during a normal drawdown. you now have a framework to tell the difference.