Troubleshooting

Troubleshooting

This page is organized by symptom — what you see on the chart — followed by the most likely cause and the fix. If something looks wrong, start here.

Runtime errors

"The indicator threw an error and nothing is drawing"

The indicator enforces three timeframe constraints that produce a runtime error and stop the script if violated. Check these in order:

Slot timeframe lower than chart timeframe. Every slot's TimeFrame must be at or above the chart's own timeframe. If you are on a 15-minute chart and Slot 01 is set to 5m, the script will error. Fix: either lower your chart timeframe or raise the slot's timeframe in settings.

Lower TF Precision not below the slot timeframe. Each slot's Lower TF Precision must be strictly lower than the slot's TimeFrame. If both are set to 5m (or the lower TF is set higher than the slot), the script will error. Fix: set Lower TF Precision to a timeframe below the slot.

Window shorter than the slot timeframe. The Window setting (Session or Rolling) must be at least as long as the slot's TimeFrame. A 60-minute slot with a 15-minute window is invalid. Fix: raise the Window to at least the slot's timeframe.

These are the three most common errors on first load and after changing chart timeframe. If you switch from a 5m chart to a 30m chart, any slot configured below 30m will trigger the first error immediately.

Unexpected oscillator behavior

"The oscillator shows extreme values right after the session opens"

Cause: Session cold-start. In Session mode, the accumulation window resets at each session boundary. The first few bars have very little data and a narrow normalization range. Modest directional volume can push the reading to +80 or +90 simply because the range is thin.

Fix: This is expected behavior, not a bug. Wait several bars for the normalization range to build out before treating the readings as contextually meaningful. The session reset markers (vertical dashed lines) show you exactly when the reset happened.

"The oscillator values changed on past bars"

Cause: On Bar Close is off. When disabled, historical bars show the final CVD value of each completed HTF bar — the outcome, not what was visible during the bar's formation. If you noticed a value change by scrolling through history after the bar closed, this is the expected repainting behavior of the "off" mode.

Fix: If you need stable history, turn On Bar Close on. See MTF and Repainting for the full explanation of both modes.

"The oscillator is flat and not updating"

Cause: Likely On Bar Close on, with a slot configured at a much higher timeframe than the chart. The slot only updates when the HTF bar closes. On a 1-minute chart with a 60-minute slot, the slot will hold the same value for sixty consecutive bars.

Fix: This is expected behavior. The flat stretch is the indicator telling you it has nothing new to confirm — the last completed HTF bar is still the most recent data point. If you want live updates, turn On Bar Close off (with awareness of the tradeoff). If the slot is flat even after the HTF bar should have closed, check whether the market is open and whether TradingView is receiving data.

"The oscillator changed direction during a quiet period (Rolling mode)"

Cause: Window aging. In Rolling mode, old bars continuously age out of the sliding window. When strongly directional bars leave the window, the CVD sum changes even if recent bars are neutral. The oscillator can decline without new selling pressure because old buying pressure left the window.

Fix: This is expected behavior for Rolling mode. Check whether the move corresponds to new directional bars entering the window or old directional bars exiting. If price action is genuinely quiet and the oscillator is still moving, window aging is the likely explanation.

"The oscillator looks different on the same symbol on a different exchange"

Cause: Volume data differs by exchange. The oscillator's input is OHLCV data from whatever exchange and symbol your chart is connected to. Different exchanges report different volume for the same instrument, which produces different delta estimates.

Fix: This is inherent to any volume-based indicator, not specific to this tool. If consistency matters, use the same exchange/symbol across all your analysis.

Blend and slot behavior

"The blended line does not match the visible slots"

Cause: A hidden slot is still contributing to the blend. The Hide Plot toggle removes a slot's line from the chart but does not remove it from the blend calculation. If you see two slot lines and the blend sits at a position that does not make sense relative to those two, a third hidden slot is pulling it.

Fix: Check whether any slots have Hide Plot on while still enabled. If you want the blend to reflect only the visible slots, disable the hidden slots entirely.

"I set a slot's weight to zero but it is still drawing and firing alerts"

Cause: Zero weight removes a slot from the blend calculation only. It does not disable the slot's plot or its alerts. The slot still computes CVD, draws its line on the chart, and fires bullish/bearish regime alerts.

Fix: If you want the slot to have no influence and no visibility, disable it entirely (uncheck Enable). A zero-weight slot is for situations where you want the slot's line visible as a reference but excluded from the blend.

"The alignment alert fired but I only have one slot enabled"

Cause: The All CVD Slots Bullish/Bearish alerts require every enabled slot to agree. With only one slot enabled, that single slot's regime determines the alignment. The alert becomes equivalent to that slot's individual regime alert, which may not be what you intended.

Fix: Alignment alerts are designed for multi-slot configurations. If you only have one slot enabled, use that slot's individual regime alert instead.

Alert issues

"I am getting an alert from a slot I cannot see"

Cause: The slot is hidden (Hide Plot on) but still enabled. Alerts are gated by the Enable toggle, not by Hide Plot. A hidden-but-enabled slot fires its alerts normally.

Fix: If you do not want alerts from a hidden slot, disable it entirely.

"The state alert fires every single bar"

Cause: Per-slot regime alerts and blended state alerts are state alerts — they fire on every confirmed bar where the condition is true, not just when the state changes. If the regime is bullish for twenty bars, the alert fires twenty times.

Fix: If you only want notification of regime changes, use the Blended CVD Regime Flip alert (transition alert, fires once on crossover). There is no per-slot transition alert — only the blended transition is available.

"The OB/OS alert thresholds seem wrong"

Cause: The Blended CVD Overbought and Oversold alerts use the Overbought Level and Oversold Level settings in the indicator. If you changed those levels from the defaults (+70 / -70), the alert thresholds changed with them.

Fix: Check the Overbought Level and Oversold Level settings. The alert thresholds always match these values.

ALMA not applying

"I selected ALMA but it does not look different from SMA"

Cause: ALMA with the default parameters (Offset 0.85, Sigma 6.0) produces output very similar to other MA types at short lengths. At a CVD Length of 3, most MA types converge to nearly the same output because there are not enough bars for the weighting differences to matter. The ALMA-specific behavior — its Gaussian-weighted windowing — becomes visible at longer MA lengths or with different Offset/Sigma values.

Fix: ALMA parameters are global — check the ALMA Offset, ALMA Sigma, and ALMA Floor Offset settings under the Global PU section. If those values have not been adjusted from defaults and your MA length is short (5 or less), ALMA will behave nearly identically to SMA or EMA. Either increase the MA length to 10+ where the weighting differences become visible, or adjust Offset and Sigma to produce more distinct ALMA behavior at short lengths.

Configuration issues

"All three slot lines look identical"

Cause: All three slots are configured to the same timeframe (or very close timeframes) with the same tuning parameters. The slots are computing CVD from the same or nearly identical input data, so they produce overlapping lines.

Fix: This is not a bug, but it defeats the purpose of multi-timeframe stacking. The value of three slots comes from running them at timeframes far enough apart that they capture structurally different market dynamics. Set the slots to progressively higher timeframes — something like 5m / 15m / 60m or 1H / 4H / D. If you only need one timeframe, disable the other two slots to reduce processing and visual clutter.