Weekly EA Monitoring Calendar
Why Gold Trading Bots Fail —
And the Weekly Monitoring System That Prevents It
Published 26 June 2026 · 13 min read
Gold trading bots fail because the traders running them wait until the damage is done before investigating. A weekly monitoring routine — checking 7 leading performance indicators every week — catches degradation signals 3–6 weeks before they become catastrophic. The calendar above shows exactly what to check each day. The health widget below gives you a weekly bot health score and a clear verdict: continue, monitor closely, or investigate now.
Why Most Bot Failures Are Predictable
Bot failure rarely arrives without warning. In the weeks before a gold EA produces catastrophic losses, the performance data tells a story — if someone is looking at it. Win rate starts to slip. The profit factor drifts toward 1.0. Drawdown in individual weeks begins to exceed the historical norm. The EA continues running. The trader continues ignoring. The data accumulates. The catastrophe happens.
The monitoring system described here works because it converts vague concern into specific, observable metrics. Rather than asking "is my EA doing okay?" — a question that is genuinely unanswerable from memory — it asks seven specific questions with specific threshold answers. You either cross a threshold or you do not. The ambiguity that causes traders to do nothing ("maybe it will recover next week") is replaced by a concrete traffic-light system.
The most important element is consistency. A weekly check done every week for six months produces 24 data points. A weekly check done "most weeks when I remember" produces noise. The routine matters as much as the metrics.
The 7 Leading Indicators of EA Degradation
These seven metrics are "leading indicators" because they signal declining health before the EA starts producing major losses. Lagging indicators — like total monthly P&L — tell you that the problem already happened. Leading indicators give you time to act.
1. Win rate vs. 30-day average
OK
Within 10% of average
Warning
Down 11–20%
Critical
Down >20%
2. Profit factor this week
OK
Above 1.2
Warning
0.8 – 1.2
Critical
Below 0.8
3. Average spread (London open)
OK
Under 12 pips
Warning
12–18 pips
Critical
Above 18 pips
4. Max drawdown this week
OK
Under 5%
Warning
5–10%
Critical
Above 10%
5. Trade errors in Journal
OK
None
Warning
1–2 minor
Critical
Multiple or serious
6. VPS uptime
OK
100%
Warning
99%+ with brief drop
Critical
Any significant downtime
7. EA producing trades in sessions
OK
Normal volume
Warning
Reduced — below 60% of average
Critical
Zero trades for 2+ active sessions
Spread Monitoring and Goldie Razor
The weekly spread check is worth emphasising because it is easy to overlook and directly impacts whether certain EAs trade at all. If your broker's average XAUUSD spread during London open has been creeping upward over the past month, the EA may be silently reducing its trade count as more entries are blocked by the spread filter — long before performance metrics show an obvious problem.
Goldie Razor V2.8.4's SpreadFilter parameter (default: 18 pips) means this Wednesday spread check is directly actionable: if your recorded weekly spread average is consistently approaching 15–18 pips, you are operating near the EA's filtering threshold. A broker that widens spreads by just a few pips from one month to the next can push you from "trading normally" to "EA mostly blocked" without any visible errors or alarms.
Your Weekly Bot Health Check
Rate each metric based on this week's observed performance. Get a health verdict and recommended action.
Win rate vs. 30-day average
Profit factor this week
Average spread (London open)
Max drawdown this week
Trade errors in Journal
VPS uptime
EA producing trades in sessions
Building the Habit: Making Monitoring Stick
The most common reason traders do not maintain a monitoring routine is that it feels unnecessary when the EA is performing well. "Everything looks fine — why bother checking?" This is the wrong framing. The check is most valuable precisely when everything appears fine, because that is when you establish your baseline data. Without a healthy-period baseline, you cannot recognise when metrics start to deviate.
Make the Saturday review a fixed commitment — 15 minutes, same time each week, with a simple 7-column spreadsheet. Week number, date, and the seven metrics. No analysis needed in the moment — just recording. After 6–8 weeks, you will have enough data to see your EA's normal band. After 12 weeks, deviations from that band become clearly visible in the trend. At that point, the monitoring system is actually working — because you can see what normal looks like and notice when something changes.
The other reason to maintain the routine even in good periods: it keeps you connected to the EA's actual performance rather than relying on memory. Memory consistently overweights recent experience. Traders who do not track data systematically tend to overestimate good periods and underreact to gradual deterioration. The spreadsheet counteracts both biases.
Related Reading
Why do most gold trading EAs fail?
What failure looks like at the population level — the categories these monitoring checks address.
Why does my EA work for 3 months then fail?
What gradual degradation over months looks like and why the regime cycle causes it.
What to do after my EA hits drawdown
The decision framework for when the monitoring system triggers a Critical status.
How to know if your EA is actually profitable
The metrics that confirm profitable operation vs. breakeven vs. genuine loss.
XAUUSD risk management essentials
The broader weekly risk routine that complements this monitoring system.
Frequently Asked Questions
One bad week is almost never a genuine warning sign — it is statistical noise in any strategy with a sub-100% win rate. Two consecutive bad weeks in the same metric are a reason to watch more carefully. Three consecutive weeks showing the same deterioration in the same metric (e.g., win rate declining, not just varying) constitutes a trend and warrants action. The key word is "trend" — a single outlier is noise; a consistent directional movement across multiple weeks is a signal. This is why recording weekly metrics in a spreadsheet matters: you cannot spot a trend from memory, only from data.
A Warning status in one metric means: increase observation frequency but do not change anything yet. Check that metric daily for the next week rather than just weekly. A Warning in two or more metrics simultaneously means: reduce lot size to 50–75% of normal while you investigate. If the Warning across multiple metrics persists for two more weeks without resolving, escalate to "Investigate now" action — which means contacting the EA developer, checking broker conditions, and potentially pausing until you understand the cause. Do not wait for Critical status to act; by then, real damage may already have occurred.
Weekly monitoring is sufficient for an EA that is performing within expected parameters. Daily monitoring is appropriate when: the EA is new and you are establishing your baseline, you have just changed a parameter or broker, any metric is showing a Warning status, or you are in a period of unusual market volatility. Daily monitoring does not mean intervention — it means awareness. You are checking whether the week is on track, not second-guessing every trade. The weekly review is the formal assessment; daily checks are the early warning system.
Seven are most useful: (1) Week's total trades, (2) win rate for the week, (3) profit factor for the week, (4) peak drawdown during the week, (5) average spread observed during London open, (6) number of Journal errors, (7) any trades skipped due to spread filter. Over time, a healthy EA shows these metrics varying within a recognisable band. When they begin trending consistently outside that band — win rate declining week over week, profit factor declining, spread increasing — you have an early warning before performance becomes catastrophically bad. The spreadsheet does not need to be sophisticated. Seven columns, one row per week, calculated over time.
Yes, weekly profit factor variation is entirely normal and expected. A strategy with a 1.5 long-term profit factor can easily show weeks with PF of 0.6 and weeks with PF of 3.2 — this is the normal statistical distribution of wins and losses. The number to watch is not a single week's profit factor but the rolling 4-week average and the 12-week trend. If the 4-week average profit factor is declining consistently — from 1.4 to 1.2 to 1.0 to 0.8 over successive months — that is a genuine deterioration signal. If it is oscillating around a stable mean, the variation is noise.
Goldie Razor V2.8.4
M15 breakout + H4 EMA filter — built for XAUUSD on MT5