Stochastic MAs+ (K Logit Bands)

Below is a ready-to-paste **English TradingView publish description** that is detailed enough to satisfy the “Originality & usefulness” and “Description” house-rule expectations. It explains **what is original**, **why the components are combined**, **how they work together**, and **how to use it**, including practical presets and cautions.

---

## Title

**Stochastic MAs+ (K Logit Bands) — Extreme-Zone Reversion with Adaptive Percentile Bands**

## Overview

This script is a **Stochastic-based extreme-zone tool** designed for traders who want signals that occur **near statistically-defined extremes**, while reducing noise and overtrading.
It combines three ideas into one coherent workflow:

1. **Stochastic %K/%D with selectable smoothing MAs** (EMA/ZEMA/SMA/KAMA)
2. **Adaptive Logit Percentile Bands** computed **on %K** (not price) to define “extreme” zones dynamically
3. A **two-step signal workflow** (Touch → Re-entry → First K/D Cross) with **cooldown + invalidation rules** to suppress repeated signals in choppy markets

This is not a “mashup for convenience.” The logit-percentile bands and the signal state-machine are explicitly built to **solve a common Stochastic problem**: fixed 20/80 levels are often too generic, and raw K/D crosses can fire repeatedly in ranges. The components here work together to make Stochastic extremes more **context-aware** and signals more **selective**.

---

## What makes it original / useful

### 1) Dynamic extremes based on the oscillator’s own distribution

Instead of using fixed 20/80, the script builds **percentile-based bands on transformed %K values**:

* **Logit transform** is used to expand sensitivity near 0 and 100 (where Stochastic tends to compress).
* A rolling buffer stores recent transformed values.
* **Percentiles** (e.g., 15% / 85%) define adaptive low/high bands that respond to changing volatility regimes.

Result: “Extreme” zones are **relative to recent market behavior**, which is often more practical than static thresholds.

### 2) A structured signal process to reduce overtrading

Classic Stochastic crossovers can spam signals. This script uses a **state-based trigger**:

**Long logic**

1. %K drops below the **adaptive low band** (touch/arm)
2. %K re-enters above the low band (re-entry)
3. The first bullish crossover occurs (K crosses above D) while K remains below the mid-band

**Short logic** is symmetrical.

Then it adds:

* **Cooldown**: prevents clustered entries during noisy periods
* **Max wait**: invalidates old setups if confirmation takes too long
* **Mid-band invalidation**: if K moves too far (crosses mid), the setup is considered late and discarded

This turns Stochastic into a **controlled mean-reversion trigger** rather than an always-on crossover machine.

---

## How it works (plain-language)

### A) Stochastic with selectable smoothing (MAK/MAD)

* `%K` is computed from the standard Stochastic formula, then smoothed with your chosen MA.
* `%D` is computed by smoothing `%K` with a chosen MA.

**MA options**

* **EMA**: baseline responsive smoothing
* **ZEMA**: reduced lag (faster reactions)
* **SMA**: heavier smoothing (less noise)
* **KAMA**: adaptive smoothing (reacts faster when price moves, slower in noise)

### B) K-based Logit Percentile Bands

The script builds bands from **%K**, not from price:

* Convert K into logit space → store in rolling buffer
* Compute low/high percentiles in logit space
* Convert back to 0–100 space with logistic function
* Produce: **kLo / kHi / kMid**

This keeps the bands stable and meaningful even when volatility changes.

### C) Signal state-machine

* **Touch**: K enters extreme zone
* **Re-entry**: K exits the extreme zone
* **Trigger**: first K/D cross after re-entry, while still in the “early” half of the band (before mid)

The idea is to catch reversals **early**, but not on the very first noisy bounce.

---

## How to use

### 1) Baseline setup (recommended starting point)

These defaults are already aligned with the script’s intent:

* Stoch: **21 / 3 / 7**
* Bands: **bandLen 200**, **low/high 0.15/0.85**, **logitGain 1.0**
* Signals: **cooldown 8**, **maxWait 24**, **Use D Direction Confirm ON**

This typically produces fewer, more selective signals than traditional 14/3/3 style settings.

### 2) Interpreting the plots

* **%K (purple)** and **%D (yellow)** are the smoothed oscillator lines.
* **kLo / kHi / kMid** are the adaptive bands.
* Labels:

* **“L”** appears near the low band when a long setup completes
* **“S”** appears near the high band when a short setup completes

### 3) Practical trading workflow

* Prefer using signals as **timing cues**, not as a complete strategy by themselves.
* Many traders combine this with:

* a trend filter (e.g., EMA200 direction)
* a volatility filter (avoid low-vol chop)
* or higher timeframe confirmation

The script is designed to give **high-quality entry timing near extremes**, but you still need a trade plan for exits and risk management.

---

## Tuning guide (fast)

### Want signals closer to extremes (more selective)?

* Decrease / increase percentiles:

* lowPct **0.12** and highPct **0.88**
* Increase logitGain slightly:

* logitGain **1.1–1.2**
* Increase cooldown:

* cooldown **10–14**

### Want earlier signals (faster confirmations)?

* Use faster MA for %D (or reduce periodD):

* maD = **ZEMA** (or EMA)
* Reduce cooldown a bit:

* cooldown **5–8**

### Getting too many signals in ranges?

* Increase periodK to reduce chop:

* periodK **34**
* Increase cooldown
* Keep D confirm enabled

---

## Strengths

* **Adaptive extreme zones**: bands adjust to changing regimes (better context than static 20/80)
* **Reduced noise**: the Touch→Re-entry→Cross structure avoids many “random” crosses
* **Configurable smoothing**: lets you tune response vs stability via MA type
* **Risk-friendly by design**: cooldown + invalidation reduce repeated entries during chop

## Limitations

* **Not a full strategy**: no position management, take-profit/stop rules, or trend filter included
* **Mean-reversion bias**: in strong trends, Stochastic can stay overbought/oversold for long periods
* **Band buffer needs history**: percentile bands are more reliable after enough bars have accumulated (bandLen)

---

## Notes on repainting / confirmations

* The percentile band buffer uses **confirmed bars** (optional) to avoid unstable band updates during an incomplete candle.
* Signal labels are plotted when the full signal conditions are met (you can enforce confirmed-bar signals via settings).

---

## Suggested disclaimer (TradingView-friendly)

This indicator is for research and educational purposes and does not constitute financial advice. Always test settings on your market/timeframe and use proper risk management.