Wahrscheinlichkeits-Oszillator

What the Indicator Measures (Short Version)
The indicator measures, over several different time windows (eight different historyLength values), the probability that the current indicator value (here, a 14-period SMA of the closing price) is higher than past values in that window.
These probabilities (named prob1 … prob8) are expressed as percentages (0–100). The arithmetic mean of these eight percentages is avgLine. Additionally, there are smoothings (SMMA) and a baseline (SMA of avgLine), similar to Bollinger Bands.

Step-by-Step: How the Values Are Calculated
Source:
sma_val = ta.sma(close, 14) → This is the 14-period simple moving average of the closing price. This smoothed price is used as the "current comparison value" (instead of raw close) to reduce noise.

Historical Array & Counting (Function calculateProbability)
For each probX, the function maintains an array of the most recently stored current values (up to historicalLength entries).

For the current sma_val, it counts how many entries in the historical array are smaller than current.

Then this number is divided by the total number of historical entries → result is a decimal between 0 and 1.

Multiplying by 100 gives probX in percent.

Mathematical (Pseudo):

ini
Kopieren
Bearbeiten
prob = (1 / total) * sum_{i=0}^{total-1} [ current > historical ] * 100
→ This is equivalent to the empirical percentile/rank position of the current value within the history.

Eight Windows / Ensemble:
prob1 … prob8 are calculated with different historyLength values (400, 350, 175, 130, 83, 42, 21, 15).

Longer windows measure “long-term” trend strength; shorter windows measure short-term relative strength/momentum.

avgLine:

ini
Kopieren
Bearbeiten
avgLine = (prob1 + ... + prob8) / 8
→ Ensemble average of all eight percentiles. Useful for smoothing extreme values from individual windows.

Smoothing (SMMA):
SMMA on prob1 and SMMA1 on avgLine reduce short-term fluctuations and make signals more stable.

Baseline & “Bands”:
Finally, basis = ta.sma(avgLine, length) and dev = mult * ta.stdev(avgLine, length) are calculated — this is exactly the baseline + band logic of a Bollinger-style representation, applied to avgLine.

Why It’s Meaningful
Percentiles/ranks are robust to scale changes. Instead of absolute price differences, the indicator answers: “Is the current (smoothed) price higher than usual over the last N periods?”

The ensemble of multiple window lengths captures different market regimes: short windows react quickly to momentum, long windows provide context and reduce false signals.

Smoothing (SMA/SMMA) reduces noise, making signals less sensitive to intraday jitter.

Interpretation: When Is the Market “Overheated” / “Not Overheated”?
High values (e.g., avgLine ≈ 80–100 or individual probX > 90):
The current SMA is higher than almost all previous values in the considered window → strong bullish dominance. This can indicate a strong rally (momentum), but also potential overbought conditions, especially if:

Volume growth is slowing, or

avgLine has remained very high for several periods (overextension).

Low values (e.g., avgLine ≈ 0–20):
The current SMA is below most of the historical values → market is under pressure or potentially oversold. Short-term reversal/recovery opportunities are more likely, especially if multiple windows are simultaneously low.

Values around ~50: Neutral — the current value is typical, in the middle of its historical distribution.

Concrete Rule (Your Specification): Buy / Sell
Buy Signal: When all eight lines (prob1 … prob8) are below lowerLine (e.g., lowerLine = 20).
→ Meaning: In all short- to long-term windows, the current SMA is below most historical values → strong, broadly confirmed undervaluation signal (potential rebound or end of correction).
Recommendation: Strong convergent long signal, especially if accompanied by volume increase or support confirmation.

Sell Signal: When all eight lines are above upperLine (e.g., upperLine = 90).
→ Meaning: In all windows, the current SMA is higher than almost all historical values → broadly confirmed overbought / overheating.
Recommendation: Strong convergent short/take-profit signal, especially if coinciding with divergences, weakening volume, or resistance areas.

Important Limitations & Risks (Pay Attention!)

Trend vs. Mean-Reversion: In a strong trend, all windows can stay high for a long time (trend continuation risk). An “all below lower → buy” signal can continue to fall in a strong downtrend → use stop-loss and trend filter (higher TF).

Historical Length & Sampling: Chosen historyLength values determine sensitivity. Very long windows make the indicator slower; very short windows increase noise.

Statistical Stationarity: Percentile signals assume the distribution remains comparable — in crashes/news events, distributions can break.

Smoothing / Lag: SMMA reduces false breakouts but adds delay — trade entries may occur later.

Practical Examples (Concrete)

Example Buy: prob1..prob8 = [12, 15, 8, 10, 14, 11, 9, 13], lowerLine = 20 → all below 20 → strong long signal.

Example Sell: prob1..prob8 = [92, 95, 90, 94, 91, 96, 93, 97], upperLine = 90 → all above 90 → take-profit / short signal.

Conclusion (Short)
Your indicator is a percentile/ranking oscillator over multiple windows — a robust ensemble measuring the relative position of the (smoothed) price to its own history.

Overheated = high, broadly confirmed prob values (e.g., all > upperLine).

Oversold = low, broadly confirmed prob values (e.g., all < lowerLine) → your rule: all below lowerLine = buy, all above upperLine = sell.

Never trade blindly — always use risk management and confirmation (volume, higher timeframes, structure).