Direct answer
In MT5 troubleshooting, “divergence” usually means there is a mismatch: the outcome you expected from your rules, calculations, or indicator logic does not line up with what you see on the chart or in the strategy tester. This mismatch can be due to a true configuration or calculation issue, but it can also come from differences in inputs (price feed, timeframe, session handling), settings (indicator parameters, applied to a different symbol or chart type), or how you interpret results. Divergence is therefore best treated as a prompt to verify assumptions, not as a standalone conclusion.
Mechanism and definition
Start by separating two ideas:
- Expected relationship (your assumption): You believe that given inputs and settings, a process should behave in a certain way (for example, that a computed series will match the displayed indicator line, or that tester results should resemble chart behavior).
- Observed relationship (your measurement): You see a different pattern, different values, or different timing.
Divergence is the difference between those. In troubleshooting language, divergence often shows up as one of these:
- The chart view and tester view show different behavior for the same concept.
- Two series that you assumed should track similarly move in different directions at the same time.
- The same indicator seems to compute differently depending on where and how it is applied.
When you look for divergence, clarify what “same time” means. Platform calculations depend on bar closes versus intrabar movement, the chosen timeframe, and whether values are updated continuously or finalized only at bar close.
Evidence, example, and why it can be misleading
Assume you expect an indicator line to match between two places: (a) the live chart and (b) the strategy tester output. A divergence occurs if the lines do not coincide.
Material reasons this can happen (without any “wrongness” in the indicator itself) include:
- Different history windows or data quality: the tester might use a dataset that does not exactly match the chart’s loaded history.
- Different inputs: even if the symbol looks the same, the trading conditions can differ (tick generation, broker feed specifics), and calculations can depend on the available price points.
- Different settings: indicator parameters, applied price type (for example, using open vs close), or the timeframe can change results.
- Chart-to-tester interpretation: the chart may show values based on how the platform updates during the current bar, while the tester’s treatment of bar timing can differ.
Even after you find a mismatch, be careful with confirmation limits and hindsight bias:
- Confirmation limits: with limited time and a small number of observed cases, you may only notice divergences that support your belief that something is “broken.” The absence of divergence in other cases does not guarantee correctness.
- Hindsight bias: once you know a divergence exists, it can become easier to “see” a reason that fits the outcome, even if alternative explanations (like settings differences) were always present.
Limitations, risks, and at least one failure mode
A key limitation is that divergence is ambiguous evidence. It indicates that expectations and observations differ, but it does not uniquely identify the cause.
At least one common failure mode in troubleshooting is comparing non-identical setups. For example, comparing results on:
- a chart timeframe that differs from the tester timeframe,
- a symbol whose chart history is not fully aligned with the tester’s dataset,
- or indicator settings that were edited after the chart was created.
Because outcomes vary with market conditions, execution details, and historical data availability, historical mismatches do not prove future behavior. Treat divergence as a diagnostic signal to verify inputs and assumptions systematically.
Verification and next question
To independently verify what divergence is telling you, reduce the problem to a controlled comparison:
- Confirm symbol and timeframe match exactly between the chart view and any testing view.
- Confirm indicator parameters and applied price inputs match.
- Ensure the same definition of “time” (bar-close vs intrabar behavior) is being compared.
- Log what changed when divergence first appeared: settings, chart history loading, or the context in which the indicator was applied.
A useful next question is: **“Which specific assumption about inputs or timing is most likely to differ between the two views I am comparing?