Warmup & Alignment
Two operational caveats affect how closely IPDA signals match a TradingView chart of the same instrument.
Warmup
The Wilder-smoothed ATR needs history to converge to TradingView values. Request a generous
DATA_LOOKBACK:
roughly 150 × TARGET_TF_MINUTES (as 1M candles)With the default TARGET_TF_MINUTES=5, that is about 750 1-minute candles. The forex
example ships DATA_LOOKBACK=600 — raise it if early signals look thin vs the chart.
Signals are suppressed until the series is warmed.
Bucket alignment
Buckets align to UTC midnight so 5-minute bars fall on :00 / :05 / :10 like a chart.
If your feed’s session boundary differs, set BUCKET_OFFSET_MINUTES and verify against a known
TradingView chart.
UTC midnight
│
├─ :00 bucket 0
├─ :05 bucket 1
├─ :10 bucket 2
└─ … (TARGET_TF_MINUTES = 5)Confirm the timeframe took effect by checking that a signal_fired record’s bucket_start
lands on a :00 / :05 / :10 boundary (see Operator runbook).
Chart vs MT5 divergence
Signals are computed from mt5-trader’s candles, so they will not line up exactly with a TradingView chart of the same instrument — different feed, different OHLC, different indicator values. Expect divergence when eyeballing the two side by side.
PIP_SIZE calibration on TradingView is still valid (unit conversion), but price levels differ
between feeds — another reason hard targets travel as distances. See
Hard Targets.
Mid-candle fire-once-lock adds another source of divergence vs closed-bar chart markers. See Signal Logic.