Thompson-Enhanced Derivative Pattern Engine

Indicator Disclaimer | Report this indicator

Thompson-Enhanced Derivative Pattern Engine

By: DskyzInvestments | DAFE Trading Systems

Traditional oscillators (RSI, CCI) rely on static thresholds built for a market that doesn't behave statically. An RSI of 70 means something different in an accumulation phase than in a low-volatility drift. TED-PE was built to stop treating price oscillators as lines to compare against fixed levels, and start treating them as the displacement term of a physical system.

The Physics of Momentum

Take the base oscillator and differentiate it four times: Velocity (d1), Acceleration (d2), Jerk (d3), Snap (d4). In a real trend exhaustion, the breakdown doesn't start at the price chart — it starts in the higher-order derivatives. Snap decays first, then Jerk crosses zero, then Acceleration flips, then Velocity stalls, and only then does price itself roll over. By the time RSI crosses overbought, the structural breakdown already happened three derivatives ago.

Differentiating a noisy series four times amplifies high-frequency noise exponentially, so the base indicator first passes through a causal, Gaussian-weighted low-pass filter (a smoother alternative to a boxcar average) before differentiation, and every derivative is rescaled with a rolling Z-score so d1-d4 live on the same comparable footing.

The Pattern Engine

Once normalized, [nd1, nd2, nd3, nd4] defines a point in a 4D kinematic phase space. Instead of hardcoded if/then rules, the engine scans its own recent history for prior states that look like the current one, using a Bayesian-weighted Euclidean distance. Matches are aggregated into a similarity-weighted ensemble — not a plain average, not a raw win rate — and combined into an Expected Value estimate (probability-weighted average win minus probability-weighted average loss), so a trade signal reflects payoff magnitude, not just direction frequency.

Dimension weights aren't fixed. A Bayesian Multi-Armed Bandit tracks which derivative has actually been predictive recently, with a forgetting factor so stale regimes get discounted. Two weighting modes are offered: Stochastic (samples from the posterior each bar — closer to true Thompson Sampling) and Deterministic (uses the posterior mean directly — steadier for live execution).

How To Use It

Base Indicator / Length / Filter Period set what gets differentiated. Search History Depth, Analog Matches, and Similarity Threshold control how the analog search behaves — larger search depth costs more compute but sees more regimes. Minimum Analog Count and Minimum EV Threshold gate how conservative signals are. Signal Timing toggles between Real-Time (fastest, can still change before the bar closes) and Confirmed (waits for the bar to close, one bar slower, never repaints). The dashboard reports live bandit weights, analog count, weighted win probability, average win/loss, and Expected Value, so you can see exactly why a signal did or didn't fire, not just that it did.

Known Limitations

d1-d4 are treated as independent axes and independent bandit arms, but they are mechanically correlated successive differences of one series — a Mahalanobis distance and correlated-arm bandit would correct this and are flagged as future work, not implemented here. Analog counts are frequently single digits; win rate and EV readouts from samples that small carry real variance and should be read as descriptive, not statistically conclusive. No out-of-sample or walk-forward validation has been performed. Transaction costs, slippage, and spread are not modeled in the EV or TP/SL calculations. This is a research framework, not a promise.

Whitepaper

https://www.dropbox.com/scl/fi/yhfoseeakehwbytsmr9nz/Bandit-Optimization-in-Higher-Order-Financial-Derivatives-Shaun-Lear-DskyzInvestments.zip?rlkey=ehqbs7qj5m8pvburdtmb9entx&st=0xv7ckue&dl=0

Structure over dogma. Kinematics over static thresholds. Evidence over conviction.
Dskyz | Velocity first. Acceleration decides.

// ═══════════════════════════════════════════════════════════════════════════════
//    THOMPSON-ENHANCED DERIVATIVE PATTERN ENGINE — TrendSpider
//    Recursive Derivative Cascade • Bayesian Thompson-Sampling Bandit •
//    Weighted-Euclidean Analog Search • EV-Gated Signals • Adaptive TP/SL
//    Created by Dskyz (DAFE) Trading Systems
//    © DAFE — Dskyz Advanced Financial Engineering
// ═══════════════════════════════════════════════════════════════════════════════
//  THEORETICAL FOUNDATION
//  Full derivation: "Non-Parametric Phase-Space Reconstruction and Bayesian
//  Bandit Optimization in Higher-Order Financial Derivatives: The TED-PE
//  Framework" — Shaun Lear (DskyzInvestments), July 2026. Section references
//  below point back to that paper; this file is the executable form of it.
//
//  • Kinematic cascade (Sec. 2) — a price oscillator is treated as the
//    displacement term of a physical system. Four successive finite
//    differences (d1 Velocity, d2 Acceleration, d3 Jerk, d4 Snap) expose the
//    structural fingerprint of a reversal — Snap and Jerk decay well before
//    price itself rolls over, which is the whole premise for reacting off
//    d1-d4 rather than the base oscillator's own overbought/oversold levels.
//  • Tompson causal Gaussian filter (Sec. 3.1) — repeated differentiation
//    amplifies high-frequency noise exponentially (Sec. 3, the Δⁿf(t)
//    binomial-difference identity), so the base indicator is smoothed with a
//    Gaussian-weighted kernel before differentiation. Per the paper's own
//    correction: this is causal, not zero-lag — expect ~(L−1)/2 bars of group
//    delay at the chosen Filter Period. That delay is inherent to the
//    method, not a porting artifact (see the real-time/confirmed note below
//    for the one piece of *additional*, avoidable lag this file controls).
//  • Rolling Z-score normalization (Sec. 3.2) — nD1-nD4 rescale each
//    derivative to standard deviations from its own rolling mean, which is
//    what makes d1-d4 comparable to each other and to the same indicator on
//    a different asset/timeframe.
//  • Weighted Euclidean analog search (Sec. 4) — the current 4D state vector
//    [nD1,nD2,nD3,nD4] is compared against up to `Search History Depth` bars
//    of prior states; matches are converted to a similarity score
//    1/(1+distance) and aggregated as a similarity-weighted ensemble, not a
//    plain top-C average or raw win rate.
//  • Expected Value gating (Sec. 4-5) — signals fire on probability-weighted
//    (avg win) − (probability-weighted avg loss), so payoff magnitude gates
//    the signal, not direction frequency alone.
//  • Bayesian bandit dimension weighting (Sec. 5) — normW1-normW4 come from a
//    Beta(α,β) posterior per derivative, decayed each bar by the forgetting
//    factor before the new observation is folded in. Stochastic mode draws a
//    Normal-moment-matched approximation of the Beta posterior (Box-Muller
//    via Math.random(), redrawn every bar) rather than an exact Beta
//    variate, per the paper's own correction that TrendSpider/Pine have no
//    native Beta-variate sampler; Deterministic mode uses the posterior mean
//    directly and is recommended for reproducible, repaint-free live use.
//  • Known, explicitly-flagged limitations carried over unchanged (Sec. 4-5,
//    7.5) — d1-d4 are treated as orthogonal axes and independent bandit arms
//    even though they're mechanically correlated successive differences of
//    one series (a Mahalanobis distance / correlated-arm bandit would
//    correct this — not implemented, flagged as future work in the paper
//    itself). Analog sample sizes are typically single digits, so win-rate
//    and EV readouts are high-variance and descriptive, not statistically
//    conclusive (Sec. 6). No out-of-sample or walk-forward validation has
//    been performed (Sec. 7, 7.5). Transaction costs, slippage, and spread
//    are not modeled anywhere in the EV or TP/SL math (Sec. 7.5, item 6).
// ══════════════════════════════════════════════════════════════════════════════
describe_indicator('Thompson-Enhanced Derivative Pattern Engine', 'price', { shortName: '⟡ TED-PE', mainColorInheritFrom: 'legend_anchor' });
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  INPUTS & CONFIGURATION                                                    ║
// ╚════════════════════════════════════════════════════════════════════════════╝
var i_baseType    = input.select('Base Indicator', 'CCI', ['RSI', 'MFI', 'CCI', 'OBV', 'CMF', 'ROC']);
var i_baseLen     = input.number('Indicator Length', 14, { min: 2 });
var i_banditMode  = input.select('Bandit Mode', 'Stochastic', ['Stochastic', 'Deterministic']);
var i_filterPer   = input.number('Filter Period', 9, { min: 3, max: 30 });
var i_normLen     = input.number('Normalization Lookback', 50, { min: 5 });
var i_decay       = input.number('Bandit Decay Factor', 0.999, { min: 0.90, max: 1.0, step: 0.001 });
var i_useFilter   = input.boolean('Apply Low-Pass Filter', true);
var i_useBandit   = input.boolean('Dynamic Bayesian Weighting', true);
var i_searchWin   = input.number('Search History Depth (Bars)', 500, { min: 100, max: 1000 });
var i_matchCount  = input.number('Analog Matches to Find', 5, { min: 2, max: 10 });
var i_matchThr    = input.number('Similarity Threshold', 0.75, { min: 0.1, max: 1.0, step: 0.05 });
var i_projFwd     = input.number('Projection Horizon (Bars)', 10, { min: 3, max: 50 });
var i_evThresh    = input.number('Minimum EV Threshold %', 0.10, { min: 0.0, max: 5.0, step: 0.05 });
var i_minMatches  = input.number('Minimum Analog Count', 2, { min: 1, max: 10 });
var i_signalMode  = input.select('Signal Timing', 'Real-Time', ['Real-Time', 'Confirmed (Bar Close)']);
var i_atrLen      = input.number('ATR Length', 14, { min: 1 });
var i_atrMult     = input.number('ATR Stop Multiplier', 2.0, { min: 0.1, step: 0.1 });
var i_showSignals = input.boolean('Show Signal Markers', true);
var i_showBars    = input.boolean('Color Candles by EV', true);
var i_showZones   = input.boolean('Show TP/SL Zone', true);
var i_glow        = input.boolean('Glow-Halo Signal Markers', true);
var i_theme       = input.select('Dashboard', 'Full', ['Full', 'Compact', 'Off']);
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  CORE HELPERS                                                              ║
// ╚════════════════════════════════════════════════════════════════════════════╝
var n = close.length;
function fillNull(len) {
    var arr = [];
    for (var i = 0; i < len; i++) arr.push(null);
    return arr;
}
function nz(v, fallback) {
    var fb = fallback === undefined ? 0 : fallback;
    if (v === null || v === undefined) return fb;
    if (typeof v === 'number' && isNaN(v)) return fb;
    return v;
}
function fmtNum(v, dec) {
    if (v === null || v === undefined || isNaN(v)) return '---';
    return v.toFixed(dec);
}
function smaArr(arr, len) {
    var m = arr.length;
    var out = fillNull(m);
    var runSum = 0, cnt = 0;
    for (var i = 0; i < m; i++) {
        var v = arr[i];
        if (v !== null && v !== undefined) { runSum += v; cnt++; }
        if (i >= len) {
            var old = arr[i - len];
            if (old !== null && old !== undefined) { runSum -= old; cnt--; }
        }
        if (i >= len - 1 && cnt > 0) out[i] = runSum / len;
    }
    return out;
}
function rmaArr(arr, len) {
    var m = arr.length;
    var out = fillNull(m);
    var smoothK = 1.0 / len;
    var prev = null;
    for (var i = 0; i < m; i++) {
        if (arr[i] === null || arr[i] === undefined) continue;
        prev = (prev === null) ? arr[i] : (smoothK * arr[i] + (1 - smoothK) * prev);
        out[i] = prev;
    }
    return out;
}
function diffArr(src) {
    var m = src.length;
    var out = fillNull(m);
    for (var i = 1; i < m; i++) {
        if (src[i] === null || src[i - 1] === null) continue;
        out[i] = src[i] - src[i - 1];
    }
    return out;
}
function zScoreArr(src, len) {
    var m = src.length;
    var out = fillNull(m);
    var meanArr = smaArr(src, len);
    for (var i = len - 1; i < m; i++) {
        if (src[i] === null || meanArr[i] === null) continue;
        var sumSq = 0, cnt = 0;
        for (var k = 0; k < len; k++) {
            var v = src[i - k];
            if (v === null || v === undefined) continue;
            sumSq += Math.pow(v - meanArr[i], 2);
            cnt++;
        }
        var sd = cnt > 0 ? Math.sqrt(sumSq / cnt) : 0;
        out[i] = sd === 0 ? 0 : (src[i] - meanArr[i]) / sd;
    }
    return out;
}
function barString(val, slots) {
    var s = slots || 10;
    var filled = Math.max(0, Math.min(s, Math.round(val)));
    var str = '';
    for (var i = 1; i <= s; i++) str += (i <= filled ? '█' : '░');
    return str;
}
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  BASE INDICATOR ENGINE                                                     ║
// ╚════════════════════════════════════════════════════════════════════════════╝
function rsiCalc(src, len) {
    var m = src.length;
    var gains = fillNull(m), losses = fillNull(m);
    gains[0] = 0; losses[0] = 0;
    for (var i = 1; i < m; i++) {
        var chg = src[i] - src[i - 1];
        gains[i] = chg > 0 ? chg : 0;
        losses[i] = chg < 0 ? -chg : 0;
    }
    var avgG = rmaArr(gains, len);
    var avgL = rmaArr(losses, len);
    var out = fillNull(m);
    for (var i = 0; i < m; i++) {
        if (avgG[i] === null || avgL[i] === null) continue;
        out[i] = avgL[i] === 0 ? 100 : (100 - (100 / (1 + avgG[i] / avgL[i])));
    }
    return out;
}
function trCalc() {
    var out = fillNull(n);
    for (var i = 0; i < n; i++) {
        if (i === 0) { out[i] = high[i] - low[i]; continue; }
        var a = high[i] - low[i];
        var b = Math.abs(high[i] - close[i - 1]);
        var c = Math.abs(low[i] - close[i - 1]);
        out[i] = Math.max(a, Math.max(b, c));
    }
    return out;
}
function atrCalc(len) {
    return rmaArr(trCalc(), len);
}
function mfiCalc(len) {
    var tp = fillNull(n), mf = fillNull(n), posMF = fillNull(n), negMF = fillNull(n);
    for (var i = 0; i < n; i++) {
        tp[i] = (high[i] + low[i] + close[i]) / 3.0;
        mf[i] = tp[i] * volume[i];
        if (i === 0) { posMF[i] = 0; negMF[i] = 0; continue; }
        if (tp[i] > tp[i - 1]) { posMF[i] = mf[i]; negMF[i] = 0; }
        else if (tp[i] < tp[i - 1]) { posMF[i] = 0; negMF[i] = mf[i]; }
        else { posMF[i] = 0; negMF[i] = 0; }
    }
    var posSum = smaArr(posMF, len);
    var negSum = smaArr(negMF, len);
    var out = fillNull(n);
    for (var i = 0; i < n; i++) {
        if (posSum[i] === null || negSum[i] === null) continue;
        var pS = posSum[i] * len, nS = negSum[i] * len;
        out[i] = nS === 0 ? 100 : (100 - (100 / (1 + pS / nS)));
    }
    return out;
}
function cciCalc(len) {
    var tp = fillNull(n);
    for (var i = 0; i < n; i++) tp[i] = (high[i] + low[i] + close[i]) / 3.0;
    var tpSma = smaArr(tp, len);
    var out = fillNull(n);
    for (var i = len - 1; i < n; i++) {
        if (tpSma[i] === null) continue;
        var devSum = 0;
        for (var k = 0; k < len; k++) devSum += Math.abs(tp[i - k] - tpSma[i]);
        var devMean = devSum / len;
        out[i] = devMean === 0 ? 0 : (tp[i] - tpSma[i]) / (0.015 * devMean);
    }
    return out;
}
function obvCalc() {
    var out = fillNull(n);
    var run = 0;
    for (var i = 0; i < n; i++) {
        if (i === 0) { run = volume[i]; }
        else if (close[i] > close[i - 1]) run += volume[i];
        else if (close[i] < close[i - 1]) run -= volume[i];
        out[i] = run;
    }
    return out;
}
function cmfCalc(len) {
    var mfv = fillNull(n);
    for (var i = 0; i < n; i++) {
        var hl = high[i] - low[i];
        mfv[i] = hl === 0 ? 0 : (((close[i] - low[i]) - (high[i] - close[i])) / hl) * volume[i];
    }
    var mfvSma = smaArr(mfv, len);
    var volSma = smaArr(volume, len);
    var out = fillNull(n);
    for (var i = 0; i < n; i++) {
        if (mfvSma[i] === null || volSma[i] === null || volSma[i] === 0) continue;
        out[i] = mfvSma[i] / volSma[i];
    }
    return out;
}
function rocCalc(len) {
    var out = fillNull(n);
    for (var i = len; i < n; i++) {
        out[i] = close[i - len] === 0 ? 0 : ((close[i] - close[i - len]) / close[i - len]) * 100.0;
    }
    return out;
}
var rawBase = fillNull(n);
if (i_baseType === 'RSI') rawBase = rsiCalc(close, i_baseLen);
else if (i_baseType === 'MFI') rawBase = mfiCalc(i_baseLen);
else if (i_baseType === 'CCI') rawBase = cciCalc(i_baseLen);
else if (i_baseType === 'OBV') rawBase = obvCalc();
else if (i_baseType === 'CMF') rawBase = cmfCalc(i_baseLen);
else rawBase = rocCalc(i_baseLen);
function lowPassFilter(src, length) {
    var clampLen = Math.max(length, 3);
    var weights = [];
    for (var k = 0; k < clampLen; k++) {
        weights.push(Math.exp(-Math.pow(k - (clampLen - 1) / 2.0, 2) / (2.0 * Math.pow(clampLen / 4.0, 2))));
    }
    var out = fillNull(n);
    for (var i = 0; i < n; i++) {
        if (src[i] === null) continue;
        var valSum = 0, wUsed = 0;
        for (var k = 0; k < clampLen; k++) {
            var idx = i - k;
            if (idx < 0 || src[idx] === null || src[idx] === undefined) continue;
            valSum += src[idx] * weights[k];
            wUsed += weights[k];
        }
        out[i] = wUsed === 0 ? src[i] : valSum / wUsed;
    }
    return out;
}
var baseInd = i_useFilter ? lowPassFilter(rawBase, i_filterPer) : rawBase;
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  DERIVATIVE ENGINE & STANDARD NORMALIZATION                                ║
// ╚════════════════════════════════════════════════════════════════════════════╝
var d1 = diffArr(baseInd);
var d2 = diffArr(d1);
var d3 = diffArr(d2);
var d4 = diffArr(d3);
var nD1 = zScoreArr(d1, i_normLen);
var nD2 = zScoreArr(d2, i_normLen);
var nD3 = zScoreArr(d3, i_normLen);
var nD4 = zScoreArr(d4, i_normLen);
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  BAYESIAN BANDIT (THOMPSON SAMPLING DYNAMIC WEIGHTING)                     ║                               
// ╚════════════════════════════════════════════════════════════════════════════╝
function mulberry32(seed) {
    var s = seed >>> 0;
    return function () {
        s = (s + 0x6D2B79F5) | 0;
        var t = s;
        t = Math.imul(t ^ (t >>> 15), t | 1);
        t = (t + Math.imul(t ^ (t >>> 7), t | 61)) ^ t;
        return ((t ^ (t >>> 14)) >>> 0) / 4294967296;
    };
}
function betaNormalSample(a, b, rngNext) {
    var mu = a / (a + b);
    var sig2 = (a * b) / (Math.pow(a + b, 2) * (a + b + 1.0));
    var sig = Math.sqrt(Math.max(sig2, 0.0));
    var u1 = Math.max(Math.min(rngNext(), 0.999999), 0.000001);
    var u2 = Math.max(Math.min(rngNext(), 0.999999), 0.000001);
    var zz = Math.sqrt(-2.0 * Math.log(u1)) * Math.cos(2.0 * Math.PI * u2);
    return Math.max(Math.min(mu + zz * sig, 1.0), 0.0);
}
var normW1 = fillNull(n), normW2 = fillNull(n), normW3 = fillNull(n), normW4 = fillNull(n);
var bd1A = 1.0, bd1B = 1.0, bd2A = 1.0, bd2B = 1.0, bd3A = 1.0, bd3B = 1.0, bd4A = 1.0, bd4B = 1.0;
var bw1 = 1.0, bw2 = 1.0, bw3 = 1.0, bw4 = 1.0;
for (var i = 0; i < n; i++) {
    if (i_useBandit) {
        if (i > i_projFwd) {
            var lb = i - i_projFwd;
            if (close[lb] !== null && close[lb] !== 0) {
                var priceChange = close[i] - close[lb];
                var priceDir = priceChange > 0 ? 1 : (priceChange < 0 ? -1 : 0);
                if (priceDir !== 0 && nD1[lb] !== null && nD2[lb] !== null && nD3[lb] !== null && nD4[lb] !== null) {
                    bd1A *= i_decay; bd1B *= i_decay;
                    bd2A *= i_decay; bd2B *= i_decay;
                    bd3A *= i_decay; bd3B *= i_decay;
                    bd4A *= i_decay; bd4B *= i_decay;
                    if ((nD1[lb] > 0 && priceDir === 1) || (nD1[lb] < 0 && priceDir === -1)) bd1A += 1;
                    if ((nD1[lb] > 0 && priceDir === -1) || (nD1[lb] < 0 && priceDir === 1)) bd1B += 1;
                    if ((nD2[lb] > 0 && priceDir === 1) || (nD2[lb] < 0 && priceDir === -1)) bd2A += 1;
                    if ((nD2[lb] > 0 && priceDir === -1) || (nD2[lb] < 0 && priceDir === 1)) bd2B += 1;
                    if ((nD3[lb] > 0 && priceDir === 1) || (nD3[lb] < 0 && priceDir === -1)) bd3A += 1;
                    if ((nD3[lb] > 0 && priceDir === -1) || (nD3[lb] < 0 && priceDir === 1)) bd3B += 1;
                    if ((nD4[lb] > 0 && priceDir === 1) || (nD4[lb] < 0 && priceDir === -1)) bd4A += 1;
                    if ((nD4[lb] > 0 && priceDir === -1) || (nD4[lb] < 0 && priceDir === 1)) bd4B += 1;
                }
            }
        }
        if (i_banditMode === 'Stochastic') {
            var rngNext = mulberry32(i + 1);
            bw1 = betaNormalSample(bd1A, bd1B, rngNext);
            bw2 = betaNormalSample(bd2A, bd2B, rngNext);
            bw3 = betaNormalSample(bd3A, bd3B, rngNext);
            bw4 = betaNormalSample(bd4A, bd4B, rngNext);
        } else {
            bw1 = bd1A / (bd1A + bd1B);
            bw2 = bd2A / (bd2A + bd2B);
            bw3 = bd3A / (bd3A + bd3B);
            bw4 = bd4A / (bd4A + bd4B);
        }
    } else {
        bw1 = 1.0; bw2 = 1.0; bw3 = 1.0; bw4 = 1.0;
    }
    var totalW = bw1 + bw2 + bw3 + bw4;
    normW1[i] = (bw1 / totalW) * 4.0;
    normW2[i] = (bw2 / totalW) * 4.0;
    normW3[i] = (bw3 / totalW) * 4.0;
    normW4[i] = (bw4 / totalW) * 4.0;
}
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  6. PATTERN RECOGNITION — WEIGHTED EUCLIDEAN ANALOG SEARCH                 ║            
// ╚════════════════════════════════════════════════════════════════════════════╝
var atrVals = atrCalc(i_atrLen);
var actualMatchesArr    = fillNull(n);
var weightedWinProbArr  = fillNull(n);
var avgWinArr           = fillNull(n);
var avgLossArr          = fillNull(n);
var expectedValueArr    = fillNull(n);
var consensusArr        = fillNull(n);
var winRateArr          = fillNull(n);
var avgWinAtrArr        = fillNull(n);
var avgLossAtrArr       = fillNull(n);
var searchStart = i_searchWin + i_projFwd + i_normLen + 5;
for (var i = searchStart; i < n; i++) {
    if (nD1[i] === null || nD2[i] === null || nD3[i] === null || nD4[i] === null) continue;
    var simArr = [];
    var retArr = [];
    var retArrAtr = [];
    var lowBound = Math.max(0, i - i_searchWin);
    var highBound = i - i_projFwd;
    for (var j = lowBound; j <= highBound; j++) {
        if (nD1[j] === null || nD2[j] === null || nD3[j] === null || nD4[j] === null) continue;
        var sq = 0;
        sq += normW1[i] * Math.pow(nD1[i] - nD1[j], 2);
        sq += normW2[i] * Math.pow(nD2[i] - nD2[j], 2);
        sq += normW3[i] * Math.pow(nD3[i] - nD3[j], 2);
        sq += normW4[i] * Math.pow(nD4[i] - nD4[j], 2);
        var dist = Math.sqrt(sq);
        var sim = 1.0 / (1.0 + dist);
        if (sim >= i_matchThr) {
            var fwdIdx = j + i_projFwd;
            var atrAtJ = nz(atrVals[j], 0);
            if (fwdIdx < n && close[j] !== 0 && atrAtJ > 0) {
                simArr.push(sim);
                retArr.push((close[fwdIdx] - close[j]) / close[j]);
                retArrAtr.push((close[fwdIdx] - close[j]) / atrAtJ);
            }
        }
    }
    var mLen = simArr.length;
    for (var a1 = 0; a1 < mLen - 1; a1++) {
        var bestIdx = a1;
        for (var a2 = a1 + 1; a2 < mLen; a2++) {
            if (simArr[a2] > simArr[bestIdx]) bestIdx = a2;
        }
        if (bestIdx !== a1) {
            var ts = simArr[a1]; simArr[a1] = simArr[bestIdx]; simArr[bestIdx] = ts;
            var tr = retArr[a1]; retArr[a1] = retArr[bestIdx]; retArr[bestIdx] = tr;
            var tra = retArrAtr[a1]; retArrAtr[a1] = retArrAtr[bestIdx]; retArrAtr[bestIdx] = tra;
        }
    }
    var actualMatches = Math.min(mLen, i_matchCount);
    actualMatchesArr[i] = actualMatches;
    if (actualMatches > 0) {
        var sumW = 0, sumWRet = 0, sumWWin = 0, sumWUpRet = 0, sumWUpW = 0, sumWDownRet = 0, sumWDownW = 0, winCnt = 0;
        var sumWUpRetAtr = 0, sumWDownRetAtr = 0;
        for (var kk = 0; kk < actualMatches; kk++) {
            var sim2 = simArr[kk], ret2 = retArr[kk], ret2Atr = retArrAtr[kk];
            sumW += sim2;
            sumWRet += sim2 * ret2;
            if (ret2 > 0) { winCnt += 1; sumWWin += sim2; sumWUpRet += sim2 * ret2; sumWUpW += sim2; sumWUpRetAtr += sim2 * ret2Atr; }
            else { sumWDownRet += sim2 * Math.abs(ret2); sumWDownW += sim2; sumWDownRetAtr += sim2 * Math.abs(ret2Atr); }
        }
        var wWinProb = sumW > 0 ? (sumWWin / sumW) : 0;
        var aWin = sumWUpW > 0 ? (sumWUpRet / sumWUpW) * 100.0 : 0;
        var aLoss = sumWDownW > 0 ? (sumWDownRet / sumWDownW) * 100.0 : 0;
        var ev = (wWinProb * aWin) - ((1.0 - wWinProb) * aLoss);
        weightedWinProbArr[i] = wWinProb;
        avgWinArr[i] = aWin;
        avgLossArr[i] = aLoss;
        expectedValueArr[i] = ev;
        consensusArr[i] = Math.max(Math.min(ev * 20.0, 100.0), -100.0);
        winRateArr[i] = (winCnt / actualMatches) * 100.0;
        avgWinAtrArr[i] = sumWUpW > 0 ? (sumWUpRetAtr / sumWUpW) : 0;
        avgLossAtrArr[i] = sumWDownW > 0 ? (sumWDownRetAtr / sumWDownW) : 0;
    } else {
        weightedWinProbArr[i] = 0; avgWinArr[i] = 0; avgLossArr[i] = 0;
        expectedValueArr[i] = 0; consensusArr[i] = 0; winRateArr[i] = 0;
        avgWinAtrArr[i] = 0; avgLossAtrArr[i] = 0;
    }
}
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  SIGNAL GENERATION — EXPECTED VALUE GATED                                  ║
// ╚════════════════════════════════════════════════════════════════════════════╝
var buyStateArr = fillNull(n), sellStateArr = fillNull(n);
for (var i = 0; i < n; i++) {
    var am = nz(actualMatchesArr[i], 0);
    var ev = nz(expectedValueArr[i], 0);
    var wp = nz(weightedWinProbArr[i], 0);
    buyStateArr[i] = (am >= i_minMatches && ev >= i_evThresh && wp > 0.5) ? 1 : 0;
    sellStateArr[i] = (am >= i_minMatches && ev <= -i_evThresh && wp < 0.5) ? 1 : 0;
}
var finalBuyArr = fillNull(n), finalSellArr = fillNull(n);
for (var i = 1; i < n; i++) {
    finalBuyArr[i] = (i_showSignals && buyStateArr[i] === 1 && buyStateArr[i - 1] === 0);
    finalSellArr[i] = (i_showSignals && sellStateArr[i] === 1 && sellStateArr[i - 1] === 0);
}
if (i_signalMode === 'Confirmed (Bar Close)' && n > 0) {
    finalBuyArr[n - 1] = false;
    finalSellArr[n - 1] = false;
}
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  DYNAMIC TP / SL ENGINE                                                    ║
// ╚════════════════════════════════════════════════════════════════════════════╝
var tpLineArr = fillNull(n), slLineArr = fillNull(n);
var tradeDir = 0, activeTp = null, activeSl = null, barsInTrade = 0;
var tradeStatus = 'Flat';
for (var i = 0; i < n; i++) {
    if (tradeDir !== 0) {
        barsInTrade += 1;
        if (tradeDir === 1 && high[i] >= activeTp) { tradeStatus = 'TP Hit (Long)'; tradeDir = 0; }
        else if (tradeDir === 1 && low[i] <= activeSl) { tradeStatus = 'SL Hit (Long)'; tradeDir = 0; }
        else if (tradeDir === -1 && low[i] <= activeTp) { tradeStatus = 'TP Hit (Short)'; tradeDir = 0; }
        else if (tradeDir === -1 && high[i] >= activeSl) { tradeStatus = 'SL Hit (Short)'; tradeDir = 0; }
        else if (barsInTrade >= i_projFwd) { tradeStatus = 'Time Exit'; tradeDir = 0; }
    }
    if (tradeDir === 0) { activeTp = null; activeSl = null; }
    if (finalBuyArr[i]) {
        activeTp = close[i] + (nz(avgWinAtrArr[i], 0) * nz(atrVals[i], 0));
        activeSl = close[i] - (nz(atrVals[i], 0) * i_atrMult);
        tradeDir = 1; barsInTrade = 0; tradeStatus = 'Long Active';
    } else if (finalSellArr[i]) {
        activeTp = close[i] - (nz(avgLossAtrArr[i], 0) * nz(atrVals[i], 0));
        activeSl = close[i] + (nz(atrVals[i], 0) * i_atrMult);
        tradeDir = -1; barsInTrade = 0; tradeStatus = 'Short Active';
    }
    tpLineArr[i] = i_showZones ? activeTp : null;
    slLineArr[i] = i_showZones ? activeSl : null;
}
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  9. VISUAL LAYER                                                           ║
// ╚════════════════════════════════════════════════════════════════════════════╝
var TH = {
    bull:    '#00FF88',
    bear:    '#FF3C00',
    bullDim: 'rgba(0,255,136,0.35)',
    bearDim: 'rgba(255,60,0,0.35)',
    glowBull:'rgba(0,255,136,0.16)',
    glowBear:'rgba(255,60,0,0.16)',
    zoneBull:'rgba(0,255,136,0.08)',
    zoneBear:'rgba(255,60,0,0.08)',
    gold:    '#FFD700',
    cyan:    '#00F5E9',
    d1:      '#00FF88',
    d2:      '#00D9FF',
    d3:      '#FFA500',
    d4:      '#FF3C60',
    panel:   '#0A0A14',
    border:  '#333344',
    a1:      '#00F5E9',
    a2:      '#9D4EDD',
    txt:     '#E2F1FF',
    dim:     '#8888AA'
};
var tiny = library('tinycolor2');
function rgba(h, a) { return tiny(h).setAlpha(a).toRgbString(); }
var glowBuyOuter = fillNull(n), glowBuyInner = fillNull(n), markerBuy = fillNull(n);
var glowSellOuter = fillNull(n), glowSellInner = fillNull(n), markerSell = fillNull(n);
for (var i = 0; i < n; i++) {
    if (finalBuyArr[i]) {
        if (i_glow) { glowBuyOuter[i] = '●'; glowBuyInner[i] = '●'; }
        markerBuy[i] = '▲';
    }
    if (finalSellArr[i]) {
        if (i_glow) { glowSellOuter[i] = '●'; glowSellInner[i] = '●'; }
        markerSell[i] = '▼';
    }
}
paint(glowBuyOuter,  { name: 'Buy Glow Outer',  style: 'labels_below', color: TH.glowBull, size: 'large' });
paint(glowBuyInner,  { name: 'Buy Glow Inner',  style: 'labels_below', color: TH.bullDim,  size: 'normal' });
paint(markerBuy,     { name: 'Buy Signal',      style: 'labels_below', color: TH.bull,     size: 'normal' });
paint(glowSellOuter, { name: 'Sell Glow Outer', style: 'labels_above', color: TH.glowBear, size: 'large' });
paint(glowSellInner, { name: 'Sell Glow Inner', style: 'labels_above', color: TH.bearDim,  size: 'normal' });
paint(markerSell,    { name: 'Sell Signal',     style: 'labels_above', color: TH.bear,     size: 'normal' });
var barColArr = fillNull(n);
for (var i = 0; i < n; i++) {
    if (!i_showBars) continue;
    var ev = expectedValueArr[i];
    if (ev === null) continue;
    if (ev >= i_evThresh) barColArr[i] = TH.bull;
    else if (ev <= -i_evThresh) barColArr[i] = TH.bear;
}
color_candles(barColArr, { name: 'EV Candle Color' });
paint(tpLineArr, { name: 'Take Profit', color: TH.bull, style: 'dotted', thickness: 2 });
paint(slLineArr, { name: 'Stop Loss',   color: TH.bear, style: 'dotted', thickness: 2 });
var bullZoneTop = fillNull(n), bullZoneBot = fillNull(n);
var bearZoneTop = fillNull(n), bearZoneBot = fillNull(n);
for (var i = 0; i < n; i++) {
    if (tpLineArr[i] === null || slLineArr[i] === null) continue;
    if (tpLineArr[i] > slLineArr[i]) { bullZoneTop[i] = tpLineArr[i]; bullZoneBot[i] = slLineArr[i]; }
    else { bearZoneTop[i] = slLineArr[i]; bearZoneBot[i] = tpLineArr[i]; }
}
color_cloud(bullZoneTop, bullZoneBot, TH.zoneBull, TH.zoneBull, 'Bull Zone Top', 'Bull Zone Bottom');
color_cloud(bearZoneTop, bearZoneBot, TH.zoneBear, TH.zoneBear, 'Bear Zone Top', 'Bear Zone Bottom');
var lastIdx = Math.max(n - 1, 0);
var tpLabelY = (n > 0 && tpLineArr[n - 1] !== null) ? tpLineArr[n - 1] : nz(close[lastIdx], 0);
var slLabelY = (n > 0 && slLineArr[n - 1] !== null) ? slLineArr[n - 1] : nz(close[lastIdx], 0);
var tpLabelText = (n > 0 && tpLineArr[n - 1] !== null) ? ('TP ' + fmtNum(tpLineArr[n - 1], 5)) : 'TP —';
var slLabelText = (n > 0 && slLineArr[n - 1] !== null) ? ('SL ' + fmtNum(slLineArr[n - 1], 5)) : 'SL —';
var tpLabelPts = fillNull(n); tpLabelPts[lastIdx] = tpLabelY;
var slLabelPts = fillNull(n); slLabelPts[lastIdx] = slLabelY;
var rTpLabel = paint(tpLabelPts, { name: 'TP Label Anchor', color: 'transparent', affects_scale: false, show_in_legend: false });
var rSlLabel = paint(slLabelPts, { name: 'SL Label Anchor', color: 'transparent', affects_scale: false, show_in_legend: false });
paint_label_at_line(rTpLabel, lastIdx, tpLabelText, { color: TH.bull, background_color: 'transparent', border_width: 0, vertical_align: 'middle' });
paint_label_at_line(rSlLabel, lastIdx, slLabelText, { color: TH.bear, background_color: 'transparent', border_width: 0, vertical_align: 'middle' });
register_signal(finalBuyArr, 'TED-PE Buy');
register_signal(finalSellArr, 'TED-PE Sell');
// ╔════════════════════════════════════════════════════════════════════════════╗
// ║  DASHBOARDS                                                                ║
// ╚════════════════════════════════════════════════════════════════════════════╝
var last = n - 1;
var l_nD1 = last >= 0 ? nD1[last] : null;
var l_nD2 = last >= 0 ? nD2[last] : null;
var l_nD3 = last >= 0 ? nD3[last] : null;
var l_nD4 = last >= 0 ? nD4[last] : null;
var l_w1 = last >= 0 ? normW1[last] : null;
var l_w2 = last >= 0 ? normW2[last] : null;
var l_w3 = last >= 0 ? normW3[last] : null;
var l_w4 = last >= 0 ? normW4[last] : null;
var l_matches = last >= 0 ? nz(actualMatchesArr[last], 0) : 0;
var l_winRate = last >= 0 ? nz(winRateArr[last], 0) : 0;
var l_wp = last >= 0 ? nz(weightedWinProbArr[last], 0) : 0;
var l_avgWin = last >= 0 ? nz(avgWinArr[last], 0) : 0;
var l_avgLoss = last >= 0 ? nz(avgLossArr[last], 0) : 0;
var l_ev = last >= 0 ? nz(expectedValueArr[last], 0) : 0;
var l_rr = l_avgLoss > 0 ? (l_avgWin / l_avgLoss) : null;
var e_d1 = Math.min(Math.abs(nz(l_nD1, 0)) * 2.5, 10.0);
var e_d2 = Math.min(Math.abs(nz(l_nD2, 0)) * 2.5, 10.0);
var e_d3 = Math.min(Math.abs(nz(l_nD3, 0)) * 2.5, 10.0);
var e_d4 = Math.min(Math.abs(nz(l_nD4, 0)) * 2.5, 10.0);
var totalEnergy = Math.abs(nz(l_nD1, 0)) + Math.abs(nz(l_nD2, 0)) + Math.abs(nz(l_nD3, 0)) + Math.abs(nz(l_nD4, 0));
var pctEnergy = Math.min((totalEnergy / 16.0) * 100.0, 100.0);
var energyColor = totalEnergy > 6.0 ? '#FF3C00' : totalEnergy > 4.0 ? '#FFA500' : totalEnergy > 2.0 ? '#9D4EDD' : '#666677';
function dirArrow(v) { return nz(v, 0) > 0 ? '▲ UP' : '▼ DN'; }
function dirColor(v) { return nz(v, 0) > 0 ? TH.bull : TH.bear; }
function mkCell(t, c, a, b, cs, bg) { return { text: String(t), color: c || TH.txt, textAlign: a || 'left', fontSize: 10, fontWeight: b ? 'bold' : 'normal', colspan: cs || 1, background: bg || TH.panel }; }
function mkSub(t, c, bg) { return { text: String(t), color: c || TH.a1, background: bg || rgba(TH.a1, 0.08), textAlign: 'center', fontSize: 9, fontWeight: 'bold', colspan: 2 }; }
var statsRows = [];
if (i_theme !== 'Off') {
    var statusColor = tradeDir === 1 ? TH.bull : (tradeDir === -1 ? TH.bear : TH.dim);
    statsRows.push({ cells: [ mkCell('⟡ TED-PE STATS', TH.a1, 'center', true, 2, TH.panel) ] });
    var barsNeeded = i_searchWin + i_projFwd + i_normLen + 5;
    var barsOk = n >= barsNeeded;
    statsRows.push({ cells: [ mkCell('Bars Loaded / Needed', TH.dim, 'left', false, 1), mkCell(n + ' / ' + barsNeeded, barsOk ? TH.bull : TH.bear, 'right', true, 1) ] });
    if (!barsOk) {
        statsRows.push({ cells: [ { text: '⚠ Not enough history — lower Search Depth', color: '#ffffff', background: rgba(TH.bear, 0.35), colspan: 2, textAlign: 'center', fontSize: 8.5, fontWeight: 'bold' } ] });
    }
    statsRows.push({ cells: [ mkCell('d1 Velocity', TH.dim, 'left', false, 1), mkCell(fmtNum(l_w1, 2) + 'x', TH.d1, 'right', true, 1) ] });
    statsRows.push({ cells: [ mkCell('d2 Accel', TH.dim, 'left', false, 1), mkCell(fmtNum(l_w2, 2) + 'x', TH.d2, 'right', true, 1) ] });
    if (i_theme === 'Full') {
        statsRows.push({ cells: [ mkCell('d3 Jerk', TH.dim, 'left', false, 1), mkCell(fmtNum(l_w3, 2) + 'x', TH.d3, 'right', true, 1) ] });
        statsRows.push({ cells: [ mkCell('d4 Snap', TH.dim, 'left', false, 1), mkCell(fmtNum(l_w4, 2) + 'x', TH.d4, 'right', true, 1) ] });
    }
    statsRows.push({ cells: [ mkSub('» ANALOG SEARCH') ] });
    statsRows.push({ cells: [ mkCell('Analogs / Win%', TH.dim, 'left', false, 1), mkCell(l_matches + ' / ' + fmtNum(l_winRate, 1) + '%', TH.txt, 'right', true, 1) ] });
    statsRows.push({ cells: [ mkCell('Weighted WinProb', TH.dim, 'left', false, 1), mkCell(fmtNum(l_wp * 100.0, 1) + '%', l_wp > 0.5 ? TH.bull : TH.bear, 'right', true, 1) ] });
    if (i_theme === 'Full') {
        statsRows.push({ cells: [ mkCell('Avg Win / Loss', TH.dim, 'left', false, 1), mkCell(fmtNum(l_avgWin, 2) + '% / ' + fmtNum(l_avgLoss, 2) + '%', TH.txt, 'right', true, 1) ] });
        statsRows.push({ cells: [ mkCell('Reward:Risk', TH.dim, 'left', false, 1), mkCell(l_rr === null ? 'n/a' : fmtNum(l_rr, 2), TH.txt, 'right', true, 1) ] });
    }
    statsRows.push({ cells: [ mkCell('Expected Value', TH.dim, 'left', false, 1), mkCell(fmtNum(l_ev, 3) + '%', l_ev >= 0 ? TH.bull : TH.bear, 'right', true, 1) ] });
    statsRows.push({ cells: [ { text: tradeStatus, color: '#ffffff', background: statusColor, colspan: 2, textAlign: 'center', fontSize: 9, fontWeight: 'bold' } ] });
}
paint_overlay('TED-PE Stats Dashboard', { position: 'top_right', order: 'above_all' }, { background: rgba(TH.panel, 0.95), border: '1px solid ' + TH.border, borderRadius: 4, width: 200, rows: statsRows });
var energyRows = [];
if (i_theme === 'Full') {
    energyRows.push({ cells: [ mkCell('⚡ DERIVATIVE ENERGY', TH.a2, 'center', true, 2, TH.panel) ] });
    energyRows.push({ cells: [ mkCell('d1 ' + dirArrow(l_nD1), dirColor(l_nD1), 'left', false, 1), mkCell(barString(e_d1), dirColor(l_nD1), 'right', false, 1) ] });
    energyRows.push({ cells: [ mkCell('d2 ' + dirArrow(l_nD2), dirColor(l_nD2), 'left', false, 1), mkCell(barString(e_d2), dirColor(l_nD2), 'right', false, 1) ] });
    energyRows.push({ cells: [ mkCell('d3 ' + dirArrow(l_nD3), dirColor(l_nD3), 'left', false, 1), mkCell(barString(e_d3), dirColor(l_nD3), 'right', false, 1) ] });
    energyRows.push({ cells: [ mkCell('d4 ' + dirArrow(l_nD4), dirColor(l_nD4), 'left', false, 1), mkCell(barString(e_d4), dirColor(l_nD4), 'right', false, 1) ] });
    energyRows.push({ cells: [
        { text: 'Total Energy', color: '#ffffff', background: energyColor, textAlign: 'left', fontSize: 9, fontWeight: 'bold' },
        { text: fmtNum(totalEnergy, 2) + ' (' + fmtNum(pctEnergy, 1) + '%)', color: '#ffffff', background: energyColor, textAlign: 'right', fontSize: 9, fontWeight: 'bold' }
    ] });
}
paint_overlay('TED-PE Energy Dashboard', { position: 'bottom_right', order: 'above_all' }, { background: rgba(TH.panel, 0.95), border: '1px solid ' + TH.border, borderRadius: 4, width: 220, rows: energyRows });

Add to TrendSpider

Terms and Conditions

THE INFORMATION PROVIDED THROUGH ANY CONTENT (“CONTENT”) PUBLISHED ON OR PROVIDED THROUGH THE MARKETPLACE (“MARKETPLACE”) SUPPORTED BY TRENDSPIDER LLC (“TRENDSPIDER”) IS PROVIDED "AS IS" AND "AS AVAILABLE" BY THIRD PARTIES AND NOT BY TRENDSPIDER OR ANY OF ITS AFFILIATES. TRENDSPIDER DOES NOT WARRANT THE ACCURACY, COMPLETENESS OR TIMELINESS OF ANY MARKET INFORMATION, FEATURES, TOOLS OR OTHER ITEMS (“INFORMATION”) MADE AVAILABLE THROUGH ANY CONTENT. TRENDSPIDER EXPRESSLY DISCLAIMS ALL WARRANTIES, EITHER EXPRESS OR IMPLIED, PERTAINING TO THE MARKETPLACE, THE CONTENT OR ANY INFORMATION MADE AVAILABLE THROUGH ANY CONTENT, INCLUDING, WITHOUT LIMITIATION, IMPLIED WARRANTIES OF MERCHANTABILITY, ACCURACY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. TRENDSPIDER EXPRESSLY DISCLAIMS ALL LIABILITY FOR ERRORS OR OMISSIONS IN, OR THE MISUSE OR MISINTERPRETATION OF, ANY INFORMATION MADE AVAILABLE THROUGH ANY CONTENT.

YOU ASSUME THE ENTIRE RISK AS TO THE USE OF ANY INFORMATION MADE AVAILABLE THROUGH ANY CONTENT. UNDER NO CIRCUMSTANCES WILL TRENDSPIDER OR ITS AFFILIATES, AGENTS, OR EMPLOYEES BE LIABLE FOR ANY DAMAGES OF ANY KIND THAT RESULT FROM THE USE OF, OR THE INABILITY TO USE, ANY CONTENT, OR ANY RELAINCE ON ANY INFORMATION MADE AVAILABLE THROUGH ANY CONTENT, EVEN IF ANY SUCH PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.

The Content is for informational, educational and entertainment purposes only. Although Content may make available Information relating to investment approaches and opportunities to buy or sell securities, you should not construe any Information made available through any Content as legal, tax, investment, financial or other advice. Nothing contained in any Content constitutes a solicitation, recommendation, endorsement, or offer to buy or sell any securities or other financial instruments. Investing involves risk and you alone assume the sole responsibility of evaluating the merits and risks associated with the use of any Information made available through any Content before making any decisions. You agree not to hold TrendSpider or its affiliates, agents or employees liable for any possible claim for damages arising from any decision you make based on Information made available through any Content.

Some free Content may require a separate subscription to paid Content in order to fully function. TrendSpider makes no representation or warranty as to the level of functionality of any Content offered on the Marketplace and hereby expressly disclaims any such warranty.

Unless otherwise indicated in the Marketplace, TrendSpider is not responsible for billing or user entitlement matters. Any issues pertaining these matters should be directed to the publisher of the Content.

Cancel Accept and continue

Thompson-Enhanced Derivative Pattern Engine

Disclaimer

Report This Indicator

Thompson-Enhanced Derivative Pattern Engine

Traders also viewed