Monitoring a live EA — the 7 metrics that reveal problems before they become losses

Why monitoring matters — the difference between operation and ownership

A commercial airliner is one of the most thoroughly tested machines ever built. Its systems have been validated, its flight envelope is well understood, and the same aircraft model has accumulated millions of hours in service. Yet no airline would consider flying it without a pilot watching the instruments throughout the journey. Not because something is likely to go wrong — but because when something does go wrong, the few seconds between recognising the problem and acting on it determine whether the outcome is a smooth recovery or a disaster.

An EA running on a live account is in exactly this position. The strategy may be well-backtested, the risk rules may be configured correctly, and the execution environment may be stable. None of that means monitoring is optional. Markets change regime. Brokers change execution policies. Slippage increases during volatile periods. News events create conditions the EA was never designed to handle. Each of these events produces a signal — in the metrics — before it produces a visible loss on the equity curve.

The difference between EA operation and EA ownership is precisely this: an operator sets the EA running and checks back occasionally. An owner has a systematic review process and responds to what the metrics say. The pilot analogy applies directly: a pilot monitoring instruments is not adding noise to a smooth flight. They are performing a continuous check that allows immediate response when conditions change.

The 7 metrics in this lesson are the instrument panel for a live EA. Each metric measures a different dimension of EA health. Taken together, they give you a complete picture of whether the EA is operating within its designed parameters or drifting outside them. Review them consistently — not just when you suspect something is wrong.

The 7 metrics every EA operator tracks

Metric 1: Win rate trend (rolling 20 trades). The win rate over your most recent 20 completed trades, updated after every trade. What it measures: whether the EA's signal quality is holding relative to its historical baseline. Healthy range: within 5 percentage points of the backtest win rate. Warning: 6–10 points below baseline. Critical: more than 10 points below baseline, or a clear downward trend across three consecutive 20-trade windows. Important nuance: a single 20-trade window with low win rate is noise. A trend — three windows each lower than the last — is signal. Look for direction, not just the current reading.

Metric 2: Rolling profit factor (last 20 trades). The ratio of gross profits to gross losses across the most recent 20 closed trades. If those trades produced $1,200 gross profit and $800 gross loss, the rolling profit factor is 1.5. What it measures: whether the EA's edge — its ability to make winning trades larger than losing trades — is intact. Healthy range: above 1.3. Warning: 1.1 to 1.3. Critical: below 1.1, or below 1.0 (meaning losses exceed profits over the recent window). Why this matters more than win rate: profit factor captures both frequency and magnitude. An EA can maintain a decent win rate while producing increasingly small wins and increasingly large losses — which win rate won't reveal but profit factor will.

Metric 3: Drawdown from peak. The current account equity expressed as a percentage decline from the highest balance the account has ever reached. If peak equity was $12,400 and current equity is $11,100, drawdown from peak is 10.5%. What it measures: how deep the EA is currently in a losing streak relative to its best performance. Healthy range: 0–5% (normal trading variance). Warning: 5–10% (concerning, requires daily review). Critical: above 10%, or above whatever threshold triggers your risk rules. Why peak matters more than starting balance: your drawdown cutoff risk rule (configured in Lesson 9) should be measured from peak, not starting balance — so your monitoring metric should match that definition.

Metric 4: Weekly PnL volatility. The standard deviation of the EA's week-by-week profit and loss results, tracked over the previous 8 weeks. What it measures: how consistent the EA's results are — and specifically whether volatility is increasing, which is one of the clearest early indicators that market regime has changed. Healthy range: weekly PnL volatility stable and similar to the backtest's weekly variance. Warning: weekly volatility more than 50% higher than the backtest's typical range. Critical: weekly volatility more than 2× backtest typical range, or extreme single-week outliers. The practical check: if your EA normally produces weeks in the range of +$300 to -$200, and last week it produced -$700, that's a 3× outlier. You don't need to calculate the exact standard deviation — a quick eyeball of the last 8 weeks tells you whether the rhythm has changed.

Metric 5: Average slippage (pips). The average difference between the price at which the EA requested an order fill and the price at which it was actually filled, measured over the last 50 filled orders. What it measures: execution quality at your broker, on your VPS, through your current network path. Healthy range: 0–1 pip average slippage for non-scalping strategies; 0–0.3 pips for scalping strategies. Warning: 1–2 pips (non-scalping), 0.3–0.8 pips (scalping). Critical: above 2 pips (any strategy), as this begins to materially erode the strategy's expected return. Where to find it: MT5's History tab shows entry and exit prices. The Trade Report (Account History → right-click → Save as Detailed Report) shows average slippage per symbol if your EA has sufficient trade count. Third-party tools like MyFXBook Autotrade provide automated slippage tracking.

Metric 6: Execution fill rate. The percentage of trade requests that were filled within acceptable parameters — versus those that were requoted, rejected entirely, or partially filled at a different size than requested. What it measures: broker execution reliability, particularly during high-volatility periods. Healthy range: above 95% of orders filled normally. Warning: 90–95% normal fills (some requotes, investigate timing). Critical: below 90% normal fills, or a sudden drop in fill rate that coincides with a particular session or news event. Where to find it: MT5's Expert Journal logs every order event. A requote appears as a journal entry like 'Request for EURUSD Buy 0.10 at 1.08450 was rejected — requote to 1.08462.' Count how many of these appear per week and divide by total order count.

Metric 7: Expert Journal error count. The number of distinct error or warning messages logged in MT5's Expert Journal (Tools → Terminal → Experts tab) since the last review. What it measures: whether the EA's code is encountering runtime errors — failed function calls, data feed interruptions, calculation failures, connection timeouts. Healthy range: 0 new errors per week. Warning: 1–3 new error types (investigate each one). Critical: any single error that repeats more than 10 times per week, or an error that appears shortly before an abnormal trade. Why errors matter beyond the obvious: many MQL5 errors are silent from a trading perspective — the EA continues running but in a degraded state, making calculations based on stale or incorrect data. An order-send failure followed by a retry, for example, might result in a double position. Read every error in the Journal; do not ignore them because trades seem to be executing normally.

Building the review habit — weekly vs daily cadence

The 7 metrics don't all need to be checked at the same frequency. Some are slow-moving and weekly is sufficient; others can change rapidly and warrant a quick daily scan. Understanding the cadence lets you build a review routine that takes 10 minutes on most days and 30 minutes once a week — without becoming a full-time job.

Daily (2–3 minutes): Check drawdown from peak and Expert Journal error count. These are the two metrics that can change fastest and indicate the most immediate problems. A drawdown that accelerates between Monday and Tuesday is actionable today. An Expert Journal error that starts appearing overnight means something changed on the infrastructure level — VPS, broker connection, data feed — and needs to be diagnosed before it compounds.

Weekly (10–15 minutes on Monday morning, after the Asian session opens): Review all 7 metrics together. Update your win rate trend and rolling profit factor from the previous week's closed trades. Check the week-over-week change in average slippage. Review the fill rate for the previous week. Confirm your drawdown from peak number. Assess whether weekly PnL volatility is within normal range for the strategy. Check for any new Expert Journal error types you haven't seen before.

What a 10-minute Monday morning review looks like in practice: Open MT5. Go to the Account History tab. Filter for the previous week. Count closed trades, note total P&L, calculate win rate and average trade size. Note the largest winning and losing trade — are they proportionate to expectation? Check the Expert Journal — scroll through since last Friday and count distinct error types. Note any new requote patterns. Open your tracking spreadsheet and log the week's metrics. Flag any metric in warning range for closer attention this week. That's the complete review for a normal week. If any metric is in warning range, it gets a 10-minute investigation before you let the EA run through Monday.

When to pause the EA vs when to stop it permanently. Pause (suspend for 1–5 days) when: rolling profit factor is in warning range but above 1.0, or slippage has increased by 50% from baseline but no other metrics are degraded, or you're entering a week with two or more Tier-1 news events. Stop permanently (require manual review and decision) when: drawdown from peak hits your configured cutoff, or rolling profit factor drops below 1.0 for two consecutive 20-trade windows, or Expert Journal shows a repeating error that you cannot diagnose and the EA is still trading. The distinction matters: a pause preserves the option to resume; a permanent stop is a decision that requires you to revisit the EA's entire setup before it trades again.