Loading...
Test six readable footprint rules with next-bar-open simulation, costs and train/validation/test holdouts on your own ATAS chart.
The ATAS plugin is compiled; the ports included for other platforms are source software supplied for the purchaser's own use. Pattern Lab is a research tool: its results use simplified assumptions (entry at the next open, a fixed cost, stop-first inside a bar), describe a past sample and are not a forecast. Its markers, results and alerts are not trading advice, and nothing here is certified for automated execution; the indicator never places orders.
A match is a research observation, not a signal: the demo rule (profit factor 0.61) lost money after costs in all three periods, which is exactly what the Lab exists to show. Every setting you change after seeing a result is another trial, so the test period is only untouched if you leave its date empty until the rule is frozen.
Windows: Get-FileHash .\your-download.zip -Algorithm SHA256. Linux: sha256sum your-download.zip. macOS: shasum -a 256 your-download.zip.
Get-FileHash .\ATAS\Cival.PatternLab.dll -Algorithm SHA256
Select-String 'Cival.PatternLab.dll' .\SHA256SUMS.txt
.\ATAS\Install.ps1
powershell -ExecutionPolicy Bypass -File .\ATAS\Install.ps1On every closed bar it checks one of six readable rules (Engulfing, DeltaReversal, Absorption, BreakoutRetest, ReversalFiveBar or a custom U/D/N candle sequence) against your volume and delta gates. A match is marked and simulated: entry at the next bar's open, a fixed stop and target, a holding limit and your round-trip cost, with stop-first assumed when one bar touches both levels. Results are scored per train, validation and test period, and trades that cross a period boundary are excluded from both sides.
These defaults are a starting point for comparison, not recommended trading parameters or a loss limit. Candle counts depend on the interval. Refer to the shipped source for validation rules before changing them.
| Setting | Default | Meaning |
|---|---|---|
| Pattern | "DeltaReversal" | Which rule is tested: Engulfing, DeltaReversal, Absorption, BreakoutRetest, ReversalFiveBar or CandleSequence. |
| Prior lookback bars | 20 | Bars used for the mean-volume gate and for the BreakoutRetest range; also the warm-up before the first possible signal. |
| Volume / prior mean | 1 | The signal bar's volume must be at least this multiple of the lookback mean volume (0 switches the gate off). |
| Minimum absolute delta % | 20 | The signal bar's absolute delta must be at least this percent of its volume (0 switches the gate off). |
| Stop distance (ticks) | 20 | Simulated stop distance from the entry, in ticks. |
| Target / stop multiple | 2 | Simulated target distance as a multiple of the stop distance. |
| Maximum holding bars | 20 | A simulated trade that has hit neither level exits at the close of this many bars (the entry bar counts as 1). |
| Round-trip costs in price units | 0 | Cost charged per simulated trade in price units, not ticks or currency. The default 0 is not realistic. |
| Validation begins (chart time) | "(not set)" | Start of the validation period; earlier trades are train. Needed before the Train and Validation lines appear. |
| Untouched test begins (chart time) | "(not set)" | Start of the test period; must be later than Validation. Leave it empty until the rule is frozen. |
Work in ATAS replay or on a demo account. The exercise teaches installation and signal interpretation; the indicator never sends an order.
Close ATAS, run .\ATAS\Install.ps1, start ATAS and add Cival Pattern Lab to a chart that has footprint history. When the calculation finishes the panel title reads Cival Pattern Lab — simulation and arrows appear on bars where the rule matched.
Untick Compact panel (title + two lines) in the Information panel group. The panel now shows all nine lines, including All n=… PF=… DD=… and the Ambiguous / censored / pending line.
Enter Round-trip costs in price units, then Stop distance (ticks), Target / stop multiple and Maximum holding bars, and choose Pattern. The All line updates after the recalculation; n changes with the rule, PF and E change with costs.
Enter Validation begins (chart time) for the last third of your history and leave the Test date empty. The Train and Validation lines appear and the Test line reads Test not configured.
Compare the win rate with the break-even rate (S + C) ÷ (S × (1 + R)), where S is the stop, R the target multiple and C the cost, all in price units. A rule needs positive expectancy after costs in validation as well as train. Check Ambiguous, censored and pending.
When no further changes will be made, enter Untouched test begins (chart time) later than the Validation date and read the Test line once. If the Test line is negative or the sample is tiny, reject the rule or gather more data instead of tuning.
Set CSV export folder to a folder used only by this chart. PatternLab-trades.csv and PatternLab-bars.csv appear after the calculation; check the partition, reason and ambiguous columns and the share of the result that comes from the best three trades.
Untick Compact panel (title + two lines) and read the All, Train, Validation and Test lines together with Ambiguous, censored and pending. Then open PatternLab-trades.csv and check that the result does not hinge on a few trades.
Close ATAS fully, confirm Cival.PatternLab.dll is in %APPDATA%\ATAS\Indicators, start ATAS again and search for Cival. If the installer was blocked, run powershell -ExecutionPolicy Bypass -File .\ATAS\Install.ps1.
Compact panel (title + two lines) is on by default. Untick it in the Information panel group.
Check the data before the rule: complete footprint rows for the period, warm-up (the first Prior lookback bars + 1 bars cannot signal), enough history, the right symbol and timeframe, and older saved template settings. A silent panel is not a measured zero. Do not loosen the gates just to force signals; each change is a new trial.
A simulated trade needs later bars to reach its stop, target or time limit, so signals near the end of the history are counted as pending, not as completed trades. Wait for more bars.
The Validation or Test date may be outside the loaded history, or no completed trade fell in that period. ATAS may show an unset date as 01/01/0001 00:00:00, which means not set. Compare the dates with the first and last bar.
The sequence must be 2 to 20 characters made only of U, D and N. Re-enter it, for example DDDUU; no warning is shown for a bad sequence.
Set the Validation date first and a later Test date, or clear the Test date.
ATAS did not report a tick size for the instrument. Check the instrument settings and reload the chart.
ATAS is still running. Close it, end leftover ATAS processes in the task manager and run the installer again.
The whole history is recalculated when a setting changes, when a closed bar is revised or when the plugin version changes. Record the history range and settings with every result you keep and export the CSV files.
Include your order reference, product/version, operating system, ATAS version, failed step and a redacted error. Never send environment files, passwords, private keys or seed phrases.