Control When Lorentzian Classification Signals Pass
Lorentzian Classification filters add conditions that an entry signal must satisfy. They do not retrain the distance model or establish that the remaining signals will be profitable. The daily kernel filter described here is a Premium control.
Interactive Settings Explorer
Use the information buttons to explore the Premium filters. The controls are a documentation sandbox; they do not run the signal logic, save a preset, or change TradingView. Read the active defaults from your installed version.
Documentation example. Changes stay on this page and do not update TradingView or calculate signals. Check your installed version for current labels, defaults, and available controls.
🧮 FILTERS SETTINGS
Setting Help
Detailed Settings Reference
| Filter | What it tests |
|---|---|
| Volatility | Whether the 1-bar Average True Range (ATR), a measure of price movement, exceeds the 10-bar ATR, meaning the latest bar moved more than recent bars did on average. Passing it does not certify that the market is trending. |
| Regime | Whether the slope of a smoothed price curve, up or down, is steep enough compared with a 200-bar exponential moving average of that slope. A Threshold of 0.1 requires a slope at least 10% steeper than that average. It is an indicator rule, not knowledge of the true market regime. |
| ADX | Whether Average Directional Index clears its trend-strength threshold. ADX does not itself give the trade direction. |
| EMA | Whether price is on the appropriate side of an exponential moving average (EMA) for the signal direction. |
| SMA | Whether price is on the appropriate side of a simple moving average (SMA) for the signal direction. |
| Daily Kernel | On intraday charts, lets buy signals pass while the daily kernel estimate is rising and sell signals pass while it is falling. It tests the slope, not whether price is above or below the kernel. Check how your installed version calculates and confirms the daily value. |
An EMA is not inherently a short-term filter or an SMA inherently a long-term filter: their lengths and the chart timeframe determine that context. Use the installed defaults as the baseline rather than treating the explorer values as a recommended strategy.
A filter that rejects a new signal does not necessarily set the classifier's remembered signal to neutral or emit an exit. Read the distinction between votes, events, and position signals before translating a filter result into a trading rule.
Filter Combinations
Compare the baseline against one additional filter first. Record both the events removed and the events retained. Combining many filters can leave too few trades to make a useful comparison and can exclude favorable events as well as unfavorable ones.
For a threshold experiment, keep the dataset and other settings fixed and compare several nearby values. Investigate whether the pattern persists outside the period used to choose the threshold. Do not assume the highest score or fewest signals identifies the safest setup.
Check Interactions
Check the script version as well as Kernel Settings. Enhanced kernel smoothing (Enhance Kernel Smoothing, or Use Kernel Smoothing in AI Edge Optimizer) disables dynamic exits in Premium v3.1. The original open-source script additionally restricts dynamic exits with its EMA and SMA filters. Use the rule for your installed script before comparing statistics or recreating alerts.
For the daily kernel filter, a lower-timeframe bar can close while the daily bar is still open. A bar-close alert on the lower timeframe does not by itself confirm the daily input. Compare live timing with the installed daily-series logic before treating historical agreement as information that was available then.