Cival Pattern Lab — Platform Compatibility

The ATAS plugin is the reference implementation. Each port computes what its platform's data allows and labels anything approximate or unavailable inside the indicator itself. The sections below are the READMEs that ship next to each port's source file (ports/<platform>/README.md).

TradingView (Pine Script v6)

Cival Pattern Lab for TradingView

Port of the ATAS original Cival Pattern Lab v1.0.0.0 to Pine Script v6 (CivalPatternLab.pine).

Pattern Lab is a research tool, not a signal service. It evaluates one of six transparent, rule-based candle/order-flow patterns on closed bars, then reports a deliberately conservative simulation of those signals (next-bar-open entry, stop and target in ticks, maximum holding time, round-trip cost, stop-first on ambiguous bars) split into train / validation / untouched-test partitions by date. Results are observed history under the stated rules, not a calibrated probability, and no setting here has been shown to have an edge. The script is an indicator(), not a strategy(): it places no orders, makes no network calls and sends nothing off the chart.

Requirements

Install

1. Open the Pine Editor at the bottom of a TradingView chart.
2. Paste the full contents of CivalPatternLab.pine (or Open > New blank indicator and replace the text).
3. Click Save, then Add to chart.
4. Set the pattern and simulation inputs, then set the Holdout dates (Validation begins / Untouched test begins). Leave a date at 1970-01-01 for (not set).
5. Optional alerts: Create alert > Condition: Cival Pattern Lab > Any alert() function call. Alerts are off by default (input Alerts on closed signal / level touch), as in the original; they fire only when a bar closes.

The status line (second row of the panel) always says which data mode is active: FOOTPRINT (real buy/sell rows), ESTIMATED ... APPROXIMATE or MIXED (both modes occurred in the loaded history).

Verification note: the script was compiled on TradingView's servers (0 errors, 0 warnings). It has not been observed running on a live chart or against a Premium footprint feed in the build environment. Inspect results on your own data before relying on them.

Feature matrix

Values: Full / Approximate / Not available / Not ported.

Settings mapping

Names and groups mirror the ATAS display names. Defaults come from Cival.PatternLab.settings.json.

Port-only inputs (group Port (TradingView only)): Footprint ticks per row (POC resolution, default 1), Estimation timeframe (blank = automatic), Footprint completeness tolerance % (default 1), Panel position, Show results table, Results table position.

Not ported (no equivalent or no effect in the original):

Known differences from ATAS

1. Data source. ATAS reads bid/ask per price from your data provider. TradingView footprint rows come from TradingView's own intrabar data and buy/sell classification, so volumes, delta and POC can differ from ATAS on the same instrument. Expect different signals bar by bar; the rules themselves are identical.
2. Estimated delta (no footprint). Each lower-timeframe bar inside the chart bar is classed as up or down by close vs open (ties keep the previous direction) and its volume is added to buy or sell. This is a proxy for aggression, not measured aggression. It is flagged APPROXIMATE everywhere it is used; bars whose classified volume does not match candle volume within the tolerance are treated as incomplete and produce no signal. Default lower timeframe by chart timeframe: 1D or higher uses 15m, 4h to under 1D uses 5m, 5m to under 4h uses 1m. Charts under 5 minutes fall back to chart-bar direction (no intrabar detail) unless you enter a lower timeframe such as 1S, which needs a plan with seconds data.
3. POC. Taken as the midpoint of TradingView's POC row (rows are ticks per row wide, default 1). ATAS takes the lowest price among equal-volume rows; TradingView's tie rule is not documented. Only the direction of POC change is used.
4. Estimated-bar POC. Estimated bars have no POC, so Require directional POC migration blocks signals that depend on them.
5. Exit time. Set to the open time of the next bar (ATAS: next candle time). For the newest trade it is provisional (bar close) until the next bar exists.
6. Holdout dates. Compared as UNIX timestamps with bar open/exit times. How the date picker maps a calendar time to UNIX time follows TradingView's input widget, not an ATAS "chart time" setting.
7. Arithmetic. ATAS uses decimal; Pine uses floating point. Equality at exact tick boundaries (touching a stop or target exactly) can differ by rounding.
8. Markers. TradingView keeps at most 500 labels; on very long histories only the most recent markers remain visible (statistics still use every processed signal).
9. History depth. TradingView limits how many bars and how much footprint/intrabar history your plan loads. Statistics cover only the loaded, processed bars; the trade list is capped at 10,000 like the original.
10. Historical vs realtime estimates. Lower-timeframe data can be revised between historical load and realtime; estimated bars may therefore differ slightly after a reload (TradingView documents this repainting risk for request.security_lower_tf).
11. Live bar. The original's live-bar handling does not apply: nothing is evaluated until the bar closes.

NinjaTrader 8 (NinjaScript)

Cival Pattern Lab — NinjaTrader 8

Compile on import: NinjaScript Editor → F5. This file has not been compiled against NinjaTrader 8 (NT8 was not available where the port was written). It was type-checked at C# 7.3 against hand-written stand-ins for the NinjaScript / SharpDX types, and every NT8 API name was checked against the NinjaTrader 8 help guide, but only your F5 compile proves it against the real platform.

Summary and requirements

Cival Pattern Lab is a research indicator. It applies one of six transparent rules (Engulfing, DeltaReversal, Absorption, BreakoutRetest, ReversalFiveBar, CandleSequence) to closed bars, gated on volume versus the prior mean and absolute delta %, with an optional directional POC-migration filter. Every match is then run through a conservative simulation: entry at the next bar's open, a stop in ticks, a target equal to the stop times a multiple, a maximum holding time, a round-trip cost, stop-first on ambiguous bars. Results are split into train / validation / untouched test partitions by date and summarised in a panel (n, win %, expectancy, profit factor, drawdown). Trades and bar features can be exported to CSV. It is an Indicator (not a Strategy), places no orders and makes no network calls. It reports observed simulation results on your own data; it makes no profit, win-rate or accuracy claim, and changing parameters is another research trial, not independent evidence.

Requirements:

• NinjaTrader 8 with an Order Flow+ entitlement, and a chart using Order Flow Volumetric Bars (Ticks per level 1 recommended; delta type BidAsk recommended). Bar delta, classified volume and POC come from the Volumetric series.
• On a chart that is not Volumetric the patterns are disabled and the panel shows "Chart is not Volumetric (Order Flow+): delta/POC patterns disabled". Nothing is estimated or invented.
• Calculate = OnBarClose; every signal forms on a confirmed bar. Nothing repaints.

Install

1. Copy CivalPatternLab.cs to Documents\NinjaTrader 8\bin\Custom\Indicators\.
2. In NinjaTrader open New → NinjaScript Editor and press F5 (Compile). The editor adds its own generated-code section at the bottom of the file on a successful compile.
3. Open a Volumetric chart, right-click → Indicators → Cival Pattern Lab.
4. The panel starts in Compact mode (title + two lines, as in the ATAS original). Untick "Compact panel" to see the train / validation / test statistics.
5. Set Validation begins and Untouched test begins (chart time, in order) before judging results. "(not set)" is DateTime.MinValue and shows as 1/1/0001 in the grid.

Feature matrix

Settings mapping

Not ported (and why)

Known differences from the ATAS original

1. Order-flow source. ATAS builds delta and POC from bid/ask footprint rows; this port reads the Volumetric series (BarDelta, TotalBuyingVolume, TotalSellingVolume, GetMaximumVolume). With the default BidAsk delta type the numbers represent the same quantity; with UpDownTick they do not.
2. POC tie-break. ATAS picks the lowest price among equal-volume rows. NinjaTrader returns the highest-volume level's price and, when Ticks per level is above 1, the lowest price of that level; ties are not specified. This only matters when Require directional POC migration is on.
3. Time basis. NinjaTrader stamps time-based bars with their close time. Bar open time (used for the entry time and the Validation / Test comparisons) is derived exactly for Minute and Second bars (including the base period of Volumetric bars) and otherwise taken as the previous bar's stamp. The exit time is the exit bar's close stamp, the same "outcome known at the candle boundary" meaning as the original. Dates are compared in the chart's time (NinjaTrader's own clock); no time-zone conversion is applied.
4. Numeric type. Prices are doubles with a 1e-6 tick tolerance on equality-sensitive comparisons; ATAS uses decimals. A signal sitting exactly on a threshold can differ in rare cases.
5. Compact panel default. As in the original, the default panel shows only three lines. Untick Compact panel for statistics.
6. Panels. Panels are drawn with SharpDX at the Panel top offset; there is no cross-indicator auto-arrangement.
7. Export cadence. Files are rewritten on the last historical bars and then on each real-time bar close. Archive them before loading another chart; use a separate folder per symbol/timeframe.

No profitability or vendor-output parity is claimed. Synthetic and historical simulations validate software behaviour, not market edge.