# Indicator source and licensing

## RedK MMTStop
Original © RedKTrader, **RedK Multi-Modal TrailingStop v1.0**, short title RedK MMTStop v1.0, internal build v1.01. Official public Pine revision 2.0 (published 2026-09-23).
Source: https://www.tradingview.com/script/ZxCXrnij-RedK-Multi-Modal-TrailingStop-RedK-MMTStop/
Original source provided as reference/RedK-MMTStop-v1.0.pine.
JavaScript port in redk.js, modified 2026-10-24 for BIT SIGNAL. That file remains available under Mozilla Public License 2.0, https://mozilla.org/MPL/2.0/ . Modifications: JavaScript arrays/state, web outputs, validation; higher-timeframe ATR is not offered. Default same-timeframe calculations, five modes, ratchet, trend, boundaries and events are ported. Display colors and layouts differ. Source author notices are retained.

## Breakout Probability (Expo)
Original © Zeiierman, official public Pine revision 2.0 (2023-03-30).
Source: https://www.tradingview.com/script/Qt7fqntR-Breakout-Probability-Expo/
Original source provided as reference/Breakout-Probability-Expo.pine.
Original and adapted expo.js are licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International, https://creativecommons.org/licenses/by-nc-sa/4.0/ . The adapted source is provided alongside the app under the same noncommercial/share-alike terms. Do not commercially reuse this adaptation without the author's separate permission.
Modified 2026-10-24: JavaScript translation, state snapshots, validation, forward projection extension. The original hit-only probability updates, current-close step, prior-bar colour denominators and original after-update backward statistic are preserved. Web UI shows zero-probability levels instead of hiding them. Forward mode deliberately projects from the most recent completed bar and is labelled a project extension. Both probabilities may be hit in the same bar; they are not complementary probabilities. No endorsement by the original author is implied.

## Project confluence model
confluence.js is an additional project model, not a published trading strategy by RedKTrader or Zeiierman. It groups historical next-bar log returns by RedK trend, candle colour and Expo bias. Each origin trains only on outcomes known by that origin; minimum 20 matching samples or abstention. Displays empirical 10th–90th quantiles, not calibrated probability bands. Chronological final-30% predictions report MAPE, constant-price baseline and direction accuracy for available signals only. No fees or execution model. All comparisons depend on loaded history and settings. No guarantee of profitability or exact TradingView parity is made.

Original Pine text is distributed unmodified. Public source retrieval used the public no-authentication Pine facade returned by each official publication. The original Pine scripts are not executed by the TradingView embed; overlays are drawn by the app's local chart renderer. There is no account/API-key trading integration in the website.

## DMI / ADX filter
Independently implemented standard Wilder DMI/ADX formulas in dmi.js. Reference documentation: https://www.tradingview.com/support/solutions/43000502250-directional-movement-dmi/ and https://www.tradingview.com/support/solutions/43000589099-average-directional-index-adx/ . Default 14/14/25 heuristic and output gating are project choices, not author-endorsed optimization. No changes to RedK or Expo source/license terms. The model still trains the original features; DMI filters displayed origins only.

## Similarity research engine and market service
analog.js, analog-worker.js, market-data.js and research-ui.js are project additions. The similarity model is not a strategy published by the RedK or Expo authors and no endorsement is implied. Existing indicator source and license terms remain unchanged. The research engine uses public OHLCV, standard technical features and causal historical neighbors; probability/quantile output is an empirical summary, not a guaranteed or calibrated forecast. See README.md for failed-baseline real-data example and validation scope.

## Direction candidate
direction.js and direction-ui.js are original project additions. The new classifier is an experimental candidate, not RedKTrader's or Zeiierman's strategy and not an improved-accuracy claim. On the documented BTCUSDT1h sample it failed its probability reference and selected no signals. Prospective records are browser-local and not independently audited. See README.md for parameters and limitations.

## Chart engine
TradingView Lightweight Charts5.0.9 ©2025TradingView,Inc. distributed unchanged at vendor/lightweight-charts-5.0.9.js with its license header and vendor/LICENSE-lightweight-charts.txt (Apache2.0). TradingView attribution is retained in the chart and footer. pro-chart.js is a project adapter. This is not TradingView's proprietary Advanced Charts platform.

## Execution study
execution.js and workspace.js are project code. Candidate long-only fills are idealized next-bar-open/close simulations with assumed costs and no actual orders. No improved predictive accuracy or profitability is claimed.

## Five-frame model
multiframe.js, multiframe-worker.js and multiframe-ui.js are project research code. Inputs are1/5/15/30/60minute completed bars, target is5minutes. They reuse publicly attributed indicator ports under their existing licenses. No author endorsement, improved accuracy or profitability is claimed. On the documented real-data sample this candidate underperformed the matched single-frame and frequency references; UI explicitly reports that result.

## Forecast5 and actual-flow history
forecast5.js, forecast5-worker.js, history.js and forecast5-ui.js are project additions. They use public Binance kline quote/taker-buy volumes and standard formulas; no TradingView proprietary source is implied. Existing RedK/Expo license obligations remain unchanged. Model selection is historical, data-dependent and not proof of an optimal predictor. Actual final evaluation underperformed the frequency reference and is reported as such.

## Precision target and nonlinear challenger
boosted.js and precision70.js are original project research additions.70% is a user-specified evaluation target, not demonstrated performance or a guarantee. The nonlinear challenger failed the historical final comparison and is not promoted. The frequency benchmark is preserved unchanged.