Bắt đầu

Jamallo

Normalized Candles RSI [Jamallo]🔹 Intro The Normalized Candles RSI is a quantitative momentum and volatility indicator designed to transform raw, trending price action into a stationary 0-100 oscillator. By passing price data through a Fractional Differencing engine and a dynamic Z-Score normalization layer, it strips away market drift and presents pure, bounded candlestick momentum. This stationary price structure is then directly overlaid with a traditional RSI and Divergence engine for highly confluent signals. 🔹 Break down Fractional Differencing (FFD): Transforms non-stationary price data into a stationary series while retaining deep historical memory (controlled by the "d" parameter). This prevents the data loss associated with standard period-to-period differencing. Normalization Engine: Applies a rolling Z-Score to the differenced data, which is then passed through a Sigmoid compression function. This mathematically forces the resulting candlesticks into a strict 0-100 range without repainting. RSI & Divergence Engine: Overlays a classic RSI with optional smoothing moving averages and Bollinger Bands. Built-in regular bullish and bearish divergence detection algorithms automatically scan for momentum exhaustion relative to the price. 🔹 Visual Guide: Indicator Anatomy The true power of this indicator lies in its visual mapping. As shown in the snapshot, the indicator pane perfectly merges the structure of raw price action directly with a classic RSI oscillator. 1:1 Structural Mapping: Notice how the normalized candles in the middle pane mirror the raw price action in the main chart exactly (highlighted by the "1:1 with overlay candle sticks" annotation). You get the exact same highs, lows, and structural candlestick patterns, but stripped of drift and mathematically squashed into a stationary box. Normalized Candles + Standard RSI: If you look at a standard RSI (bottom pane), you only see a floating line. By overlaying the normalized candles (middle pane), you can see exactly how the physical candlestick bodies and wicks behave relative to the RSI's 0-100 bounds and its smoothing moving average (the yellow line). Actionable Context: When a localized top or bottom forms, you aren't just guessing based on a purple line peaking. You can see the actual candlestick structure exhausting itself, forming wicks and reversal patterns right against the upper (70+) or lower (30-) overextended boundaries. 🔹 Settings Parameters - Diff Order (d) (0.01 - 0.99): The fractional differencing order. Higher values make the series more stationary but remove more historical memory. - Z-Score Lookback: The rolling window used to calculate the mean and standard deviation for the 0-100 normalization engine. - Sigmoid Compression Scale: Controls the elasticity of the 0-100 bounds. A higher value results in softer compression at the extremes, while a lower value makes it compress faster. - RSI Settings & Smoothing: Standard parameters for the RSI length, smoothing moving averages (SMA, EMA, VWMA, etc.), Bollinger Bands, and the divergence lookback constraints. 🔹 Credits & Acknowledgements The core RSI and Divergence engine utilized in this script is adapted from TradingView's built-in Relative Strength Index indicator code.
Chỉ báo Pine Script®
của Jamallo22
6
Fractional CUSUM Regime Filter [Jamallo]🔹 Intro The Fractional CUSUM Regime Filter is a quantitative trend and volatility channel indicator designed to identify structural market regimes with high stability. Unlike traditional filters that suffer from lag or whipsaw signals, this indicator applies a dual-layered preprocessor combining Lopez de Prado’s Fixed-Window Fractional Differentiation (FFD) with a classic Cumulative Sum (CUSUM) statistical trigger. 🔹 Break down Fractional Differentiation (FFD) : Stationarizes pricing input while retaining long-term historical memory (controlled by the "d" parameter). Classic CUSUM Filter : Accumulates deviations from a rolling baseline relative to current volatility. When cumulative deviation exceeds the threshold multiplier (h), a new regime change is triggered, and the baseline steps dynamically to the new price level. Vol-Adjusted Percentile Bands : Linear interpolation percentiles of price deviation relative to the stepped baseline. The distances are "locked" and only update on CUSUM regime triggers to prevent wobbly bands. 🔹 Visual Guide: Indicator Anatomy Here we have the structure of the indicator, including the CUSUM baseline, the 68% inner percentile band, and the 95% outer percentile band. How to use: Mean Reversion (Pullbacks to the Mean) The channel boundaries represent statistical extremes. When the price is pushed outside the bands, it is mathematically overextended and highly likely to revert back to the CUSUM baseline. Long Setup (Pullbacks in Bullish Trend): During a green CUSUM uptrend, watch for price to pull back to the lower bands ("micro pullback") and enter as it heads back up to the baseline mean. Short Setup (Pullbacks in Bearish Trend): During a red CUSUM downtrend, watch for price to rally into the upper bands ("micro pullback") and enter as it reverts down to the baseline mean. 🔹 How to use: Trend Following & Risk Management The indicator is designed to capture sustained macro trends while providing clear risk parameters. Entering on Breakthroughs : Enter when the BUY/SELL signal flags appear (indicating a new CUSUM regime shift). Stop Loss Placement : Place stop loss orders just below the opposite outer band or below the stepped baseline. Letting Winners Run : Ride the trend as long as the CUSUM baseline maintains its colored regime state (Green for Long, Red for Short). 🔹 Settings Parameters d (0.01 - 0.99) : Differentiation order. Lower values retain more historical memory; higher values approach first-difference. CUSUM Drift & Threshold : Controls baseline sensitivity to regime changes. Percentile Lookback & Targets : Set the statistical width of the inner and outer boundaries.
Chỉ báo Pine Script®
của Jamallo22
Machine Learning: seMLP Q-Wavelet RL Engine [Jamallo]Author Note: I always get asked: "How can I build a Machine Learning or Artificial Intelligence trading system?" I created the study "Machine Learning: seMLP Q-Wavelet RL Engine" to showcase exactly how it can be done in a beginner-friendly manner. We will break down exactly how this AI thinks in plain English, and then show you exactly how the Pine Script code executes it step-by-step. Introduction: The Institutional Approach to Algorithmic Trading Most retail and algorithmic traders spend years searching for the "holy grail" by combining static indicators and hard-coded `IF/THEN` rule sets. They are often unaware that institutional quant desks abandoned those basic, curve-fitted patterns decades ago. Standard algorithmic analysis fails because financial markets are inherently chaotic—a hardcoded strategy that works perfectly in a backtest will systematically break down during a live regime shift. To acquire a true institutional edge, algorithmic strategies cannot rely on rigid, backwards-looking formulas; they require a system that adapts dynamically in real-time. This script brings that quantitative firepower directly to your chart by constructing a live Self-Teaching AI . Dynamic Filtering : It uses advanced frequency mathematics (Wavelets) to separate random market noise from true institutional momentum footprints with near-zero lag. Artificial Brain : It feeds that data into a neural network—a living matrix of artificial "neurons" that continuously analyze and execute decisions. Self-Correction : Most importantly, it executes Reinforcement Learning. If a trade fails, the AI actively calculates the error and mathematically rewires its own brain, ensuring it constantly evolves to survive changing market conditions. Ultimately, this serves as a foundational study showing you exactly how to break away from basic scripting and get started in true Quantitative Algorithmic Trading. 1. The Core Architecture Loop Here is the high-level flow of how the AI thinks on every single candle: The Invisible "Burn-In" Phase Because the AI starts with a completely randomized, "empty" brain, it will make terrible decisions on the very first few candles. To prevent it from acting prematurely on live data, the script executes an aggressive Burn-In Phase (e.g., the first 300 bars of the chart). During this period, the indicator is completely invisible. It aggressively executes hundreds of "mock trades" in the background, tracking virtual PnL, taking massive risks, and rapidly rewiring its brain without showing a single signal on your screen. Once the 300 bars are up, the burn-in phase ends. The AI stops acting recklessly and officially enters "Live Trading" mode with a fully trained, highly-intelligent brain. SECTIONS 2 & 3: Setting Up the Brain Conceptual Overview Imagine the brain as a massive team of financial analysts. We have 16 junior analysts looking at chart data. They report their findings up to 12 senior analysts. The seniors report to 6 directors. The 6 directors send their final opinions to 3 executives representing the 3 possible actions: `BUY, SELL, HOLD`. This is called a 16 → 12 → 6 → 3 network structure. Before we hand the price data to the junior analysts, we Normalize it (Z-Score). This just means "leveling the playing field" so a massive $500 candle wick doesn't break the analysts' math compared to a tiny $1 movement. The Code Breakdown // Section 2: Brain Size Constants int NI = 16 // 16 Inputs (Junior analysts) int NH1 = 12 // 12 Hidden layer 1 nodes int NH2 = 6 // 6 Hidden layer 2 nodes int NO = 3 // 3 Outputs // Section 3: Normalization Helper norm(series float x, simple int win) => float mu = ta.sma(x, win) float sg = ta.stdev(x, win) float sf = nz(sg) < 1e-10 ? 1.0 : sg float res = (x - nz(mu, x)) / sf // Levels out the price data na(res) ? 0.0 : res SECTIONS 4 & 5: Giving the AI "Memory" Conceptual Overview By default, TradingView indicators suffer from permanent amnesia! Every time a new candle paints, TradingView completely deletes its short-term memory and forgets what happened on the last candle. If we are building an AI for trading that needs to "learn", it must be able to remember its past mathematical mistakes. To force TradingView to remember, we use special variables called `var` to create "Persistent Memory Matrices" where the AI for trading stores its brain's wiring throughout the entire chart history. The Code Breakdown // Using 'var' locks the memory so it never resets when a new candle paints var matrix W1 = matrix.new(NI, NH1, 0.0) // The connections between neurons var matrix W2 = matrix.new(NH1, NH2, 0.0) ... var int pos = 0 // The AI remembers its current position: Long (1), Short (-1), or Flat (0) SECTION 6: Seeing the Market (Wavelets) Conceptual Overview If you use a Moving Average, it always "lags" behind the real price. By the time the Moving Average crosses to tell you to buy, the massive breakout has already happened. To fix this, we teach the AI for trading to see using Haar Wavelets . A Wavelet is a piece of advanced math that splits the price candle with minimal lag into two things: The Detail (D) : The immediate, rapid volatility chop. The Smooth (V) : The true underlying smooth momentum. By looking at the detail and momentum completely separately, the AI for trading can react to shifts with minimal lag. The Code Breakdown // We take standard features like Open, Close, and Volume: float f0 = open float f1 = close... // We break them into Wavelets using simple math combinations: float v1_0 = (f0 + nz(f0 , f0)) / 2.0 // Smooth momentum float d1_0 = (f0 - nz(f0 , f0)) / 2.0 // Instant volatility detail ... // We pack all 16 traits into the 'feat' array to feed the AI for trading's Brain feat.set(0, norm(d1_0, i_normWin)) feat.set(14, float(pos)) // Tells the brain its current trade position feat.set(15, norm(portRet, i_normWin)) // Tells the brain its current open trade return SECTION 7: How the Brain Thinks (seMLP) Conceptual Overview An "MLP" is just a standard Neural Network (a massive web of variables that pass data to each other). The problem is that if you give TradingView an insanely massive web of math equations, it will crash and throw a compiler timeout error. So, we use a Self-evolving MLP (seMLP) . The AI pushes the Wavelet data through its network dynamically. To prevent "dead zones" where a neuron just stops firing in a flat market, it uses a formula called LeakyReLU . It basically acts as a gatekeeper that tells the neuron: "If this signal is incredibly weak, shrink it down to 1%, but don't explicitly delete it." The Code Breakdown // The data enters Hidden Layer 1 (h1) array h1 = array.new(NH1, 0.0) for j = 0 to NH1 - 1 float s = B1.get(j) // The inner brain loops through all 16 incoming inputs for i = 0 to NI - 1 s += feat.get(i) * W1.get(i, j) // LeakyReLU Formula: f(x) = x if x > 0 else 0.01 * x // If the signal 's' is positive, keep it. If 's' is negative, shrink to 1% h1.set(j, s > 0 ? s : 0.01 * s) SECTION 8: Taking Action (Exploration vs Exploitation) Conceptual Overview How does the AI actually press the BUY or SELL button? It calculates a "Confidence Score" (called a Q-Value) for all three options— Buy, Sell, and Hold. The highest score wins and executes the trade. However, during its invisible "Burn-In Period", the AI uses a variable called Epsilon . Think of Epsilon as a dice roll. Sometimes, instead of making the smartest, highest-scoring choice, the AI will randomly pick a completely stupid trade just to "experiment" and see if a hidden market pattern exists! This is conceptually how AI for trading discovers new, out-of-the-box strategies. As training goes on, Epsilon gets smaller, and the AI stops experimenting. The Code Breakdown // Calculate Epsilon: Start at a high 50% and slowly decay to 5% over time float epsilon = bar_index <= i_burnIn ? math.max(0.05, i_epsStart_val * ...) // Roll the dice. If the random number is less than epsilon, we experiment randomly! bool explore = math.random(0.0, 1.0) < epsilon // Find the AI for trading's highest confidence choice: Q(0) = Buy, Q(1) = Sell, Q(2) = Hold if Q.get(1) > bestQ // If Sell confidence is higher than current best (Buy)... bestQ := Q.get(1) bestAct := 1 if Q.get(2) > bestQ // If Hold is even higher... bestQ := Q.get(2) bestAct := 2 // Execute the final action int act = explore ? math.min(int(math.floor(math.random(0.0, 2.999))), 2) : qArg SECTION 9: Training with Rewards (Reinforcement Learning) Conceptual Overview This is the heart of Machine Learning. It functions exactly like training a pet. If the AI makes a winning trade that generates cash, we give it a mathematical "treat" (a positive reward). If the AI loses money, we hit it with a brutal negative reward. Over time, the AI autonomously refines its neural weights exclusively to collect the maximum amount of "treats". The Code Breakdown // Calculate how much money the candle moved float cRet = nz((close - close ) / close , 0.0) // The Reward (R) is a combination of three factors: // 1. PnL (rPn) - Did we make raw cash profit? // 2. Trail (rTn) - Did we efficiently track the trend? // 3. Lee (rLee) - A shaping bonus for correct directional positioning. float R = i_alphaT * rTn + i_alphaP * rPn + 0.1 * rLee SECTION 10: Learning from Mistakes (Backpropagation) Conceptual Overview If the AI's trade failed, how does it adjust its internal logic? It uses a process called Backpropagation . It looks at the Reward it just received, realizes it was horribly wrong, and calculates the "Error Margin" (How far off my prediction was I?). It then mathematically rewrites all of the internal connections `(W1, W2, W3)` in reverse, editing them to be slightly smarter for the next candle! Because updating a massive brain on every single micro-tick causes chaotic glitches, we "Accumulate" the errors in a batch over several candles, and then update the brain smoothly with the batch average. The Code Breakdown // Compare the Target Reward vs what the Brain actually Predicted (Temporal Difference Error) float tgt = R + i_gamma * max_qt float td = tgt - pOut.get(prevAct) // Accumulate the backwards gradients over multiple bars so we don't glitch for j = 0 to NO - 1 gB3_acc.set(j, gB3_acc.get(j) + g3.get(j)) accumCount += 1 // Once 'i_accumSteps' bars have passed, we apply the compiled batch update to 'Rewire' the Brain weights! if accumCount >= i_accumSteps for i = 0 to NH2 - 1 for j = 0 to NO - 1 float dw = gW3_acc.get(i, j) * sc W3.set(i, j, W3.get(i, j) + clr * dw - clr * i_l2 * W3.get(i, j)) SECTION 11: Link Pruning (Making the Brain Faster) Conceptual Overview Stage 1: The Initial Brain (Complex & Slow) Stage 2: The Pruning Decision Stage 3: The Optimized AI for trading (Sleek & Fast) As the brain learns, some of the mathematical connections become totally useless. Having a giant Tradingview indicator calculate hundreds of useless math connections will trigger a calculation timeout. At a specific point in training length (defaulting to the end of the 300-bar burn-in period), the script literally pauses and deletes (zeroes out) the weakest neural links. TradingView skips over calculations containing plain zeroes, making your indicator insanely fast and completely lag-proof. The Code Breakdown if bar_index == i_pruneBar and not pruned // Evaluate every single connection weight... // Find the bottom weakest percentage (i_prunePct) float thr = absW.get(pidx) // Explicitly set the weakest weights to Zero! for i = 0 to NI - 1 for j = 0 to NH1 - 1 if math.abs(W1.get(i, j)) <= thr W1.set(i, j, 0.0) // Permanent pruning: weak link removed Important Disclaimer This indicator is published strictly for educational and research purposes. It is a conceptual showcase proving that advanced Deep Reinforcement Learning architectures generally reserved for Python/TensorFlow can be natively executed within the TradingView Pine Script environment. Due to Pine Script's structural time-series limitations—specifically the lack of a random-access historical buffer required for true experience replay—this is NOT intended for practical live trading. For production-grade deployment, it is highly recommended to port this mathematical framework to Python. References This indicator's mathematical engine was directly modeled and bridged from the following quantitative research papers: Lee et al. (2021) — " Learning to trade in financial time series using high-frequency through wavelet transformation and deep reinforcement learning " (Used for the MODWT Wavelet integration & State architecture). Tsantekidis et al. (2021) — " Price Trailing for Financial Trading using Deep Reinforcement Learning " (Used for the dynamic margin-trailing reward system). Seow et al. (2021) — " seMLP: Self-evolving Multi-layer Perceptron " (Used for the 16 → 12 → 6 → 3 sparse Neural Network structure and the automatic Link Pruning logic).
Chỉ báo Pine Script®
của Jamallo22
Thermal Candlestick Spectrogram [Jamallo]The Thermal Candlestick Spectrogram replaces flat, binary candlesticks with a heat-mapped view of price density. Every bar radiates outward from a white-hot core through layers of orange, red, and deep amber — the same way thermal imaging reveals intensity beneath a surface. The result is a chart that doesn't just show you what price did, it shows you how concentrated that move was. How It Works A 7-layer gradient stack is anchored to each candle's midpoint and expands symmetrically toward the edges of the body. Transparency increases with each outer layer, creating a natural falloff that mimics heat dissipation: White-yellow core — maximum density at the center of the body Orange-red mantle — transitional layers that reveal the spread of price action Dark outer boundary — a near-transparent edge that lets candles breathe without hard lines Fading wicks — a dual-layer vertical fade that tapers wicks naturally toward the High and Low extremes, reducing visual noise Anti-aliased price dot — a precision close marker with a subtle white halo for clean real-time tracking Design & Performance The entire effect runs on a fixed 9-plot architecture (7 body layers + 2 wick layers), keeping script weight low and render performance stable even alongside other indicators. No dynamic loops, no repainting — just a clean, deterministic stack that works across all timeframes and instruments. Visual Comparison Traditional flat candles vs. the Thermal Spectrogram — same data, entirely different depth.
Chỉ báo Pine Script®
của Jamallo22
2
Haar Wavelet RSI [Jamallo]Author's Note This is the oscillator companion to the Haar Wavelet Range Filter. Both indicators share the same MODWT Haar wavelet foundation, creating a synergistic effect when used together — trend and momentum derived from the same mathematical decomposition of price. Introduction The Haar Wavelet RSI is a momentum oscillator that replaces the standard price-difference input of a traditional RSI with wavelet detail coefficients derived from a Maximal Overlap Discrete Wavelet Transform (MODWT) with a Haar basis. Instead of measuring bar-to-bar price change, it measures momentum at a specific frequency scale — filtering out the noise above and below that scale before the RSI calculation even begins. How It Works MODWT Haar Wavelet Decomposition The script decomposes hl2 through up to 5 wavelet levels using a shift-invariant Haar transform. At each level, the detail coefficients capture the local price delta at that specific scale — Level 1 is 2-bar momentum, Level 3 is 8-bar momentum, Level 5 is 32-bar momentum. These detail coefficients become the input to the RSI in place of raw price change, meaning the oscillator is measuring directional energy at a single isolated frequency rather than the full noisy price series. Wavelet RSI Calculation The selected detail coefficient is split into its upward and downward components, smoothed with Wilder's RMA, and fed into the standard RSI formula. The math is identical to a conventional RSI — the only difference is what it is measuring. Because the detail coefficients are already frequency-isolated, the resulting RSI is inherently cleaner than applying RSI to raw price. Adaptive Deadband Step-Hold The wavelet RSI output is then passed through an adaptive deadband filter. The threshold is self-derived from the RSI's own recent volatility — a short rolling average of absolute RSI changes — scaled by the deadband multiplier. The RSI only updates when it departs from its held value by more than this threshold. Below the threshold it holds flat. This eliminates the continuous micro-oscillation that makes standard RSI difficult to read on noisy bars, producing a stepped line that moves with conviction or not at all. Closing This is not a standard RSI with cosmetic changes. The wavelet decomposition fundamentally changes what the oscillator is measuring — momentum at a defined frequency scale rather than raw bar-to-bar noise. The decomposition level is the most important setting: lower levels suit faster timeframes and shorter momentum cycles, higher levels filter more aggressively and are better suited to swing-level momentum reads.
Chỉ báo Pine Script®
của Jamallo22
2
Haar Wavelet Range Filter [Jamallo]Intro Haar Wavelet Range Filter Standard moving averages often struggle with a fundamental trade-off: they are either too fast (reacting to every bit of market noise) or too slow (lagging behind actual trend shifts). This indicator solves that problem by moving away from time-based averaging and into the frequency domain . By utilizing a Maximal Overlap Discrete Wavelet Transform (MODWT) with a Haar basis, this filter decomposes price action into specific "energy levels." It isolates the core trend from the high-frequency "chatter" (wicks and micro-jitters). The result is a highly responsive, shift-invariant trend line that stays flat during consolidation and moves decisively when a true volatility-backed shift occurs. How It Works The indicator processes price through a multi-stage signal processing pipeline: Stage 1 — MODWT Haar Decomposition Instead of a simple windowed average, the engine performs a wavelet decomposition. It splits the price into Scaling coefficients (the smooth trend) and Detail coefficients (the noise/volatility). Unlike standard DWT, this uses MODWT , making it shift-invariant. This means the filter doesn't "jump" or repaint when new bars arrive; the relationship between price and the wavelet remains constant over time. The Decomposition Level (1–5) dictates the scale of the trend, ranging from ultra-fast 2-bar captures to macro 32-bar structural cycles. Stage 2 — Adaptive Detail-Energy Deadband The "step-like" nature of the filter is driven by an automated threshold. The indicator measures the Detail Energy—essentially the absolute volatility of the noise at your chosen scale. Stage 3 — Non-Parametric Percentile Bands Rather than using standard deviation (which assumes price follows a normal "Bell Curve" distribution), this indicator uses Linear Interpolation Percentiles.
Chỉ báo Pine Script®
của Jamallo22
8
Iteratively Reweighted Least Squares Range Filter [Jamallo]Intro Iteratively Reweighted Least Squares Range Filter Most filters treat every bar equally. This one doesn't. The IRLS Range Filter uses a robust statistical engine to continuously evaluate how much each bar in the lookback window should be trusted — down-weighting noisy data and rejecting outliers entirely — before computing a consensus price level. The result is a smoother, more honest trend line that is far less susceptible to wicks, gaps, and volatile noise than conventional moving averages. How It Works The indicator runs in three stages: Stage 1 — IRLS Consensus Engine . Every bar in the rolling window is assigned one of three trust states based on how far its price deviates from the current consensus, measured in adaptive sigma units. High Trust bars (within Lo-Trust Band σ) receive full weight. Lo Trust bars receive half weight. Rejected bars receive none. A Schmitt-trigger hysteresis mechanism prevents bars from flickering between states on minor fluctuations. The weighted average of all contributing bars becomes the consensus price. This residual-based weighting scheme is adapted from the IRLS framework described in Li & Deng (2026), arXiv:2603.08158. Stage 2 — Deadband Quantizer . The continuous consensus is passed through a step filter. The output only moves when the consensus has shifted by more than one deadband unit (Step Size × σ). This eliminates micro-jitter and produces a clean staircase line with fewer false trend flips. Stage 2.5 — Sample-and-Hold Bands. The volatility channel is only updated when the filter takes a new step, so band width reflects conditions at the last decisive move rather than real-time noise. An EMA-style smoother (Band Smooth) prevents abrupt band jumps between steps. Reference Li, J., & Deng, F. (2026). Outlier-robust Autocovariance Least Square Estimation via Iteratively Reweighted Least Square. arXiv preprint arXiv:2603.08158v1.
Chỉ báo Pine Script®
của Jamallo22
Iteratively Reweighted Least Squares (IRLS) [Jamallo]Author's Note: This script is published as a unique mathematical contribution to TradingView's open-source public library. It is intended to introduce a novel application of a robust statistical method for the community and developers to study, adapt, and build upon, rather than to serve as a standalone, out-of-the-box trading strategy. Introduction Almost all moving averages and smoothing filters in technical analysis treat historical price data equally or apply a fixed mathematical decay (like an EMA). The problem? A massive, anomalous wick or a sudden volatility spike will inevitably drag the average away from the true underlying market consensus. Enter Iteratively Reweighted Least Squares (IRLS) . IRLS is a robust statistical method that calculates a "consensus" price by actively identifying and down-weighting outliers. Instead of letting a large wick distort the line, the algorithm assigns less weight to prices that deviate furthest from the current estimate. The result is a filter that cuts through noise, ignores price-distant spikes, and naturally locks onto the dominant, high-density price levels. How It Works The indicator uses the Hardy weight function to determine how heavily each historical candle influences the current estimate. On every bar, the algorithm checks the distance of each sample from the current consensus and iteratively refines the line until it converges on a robust mean. Epsilon — the outlier rejection scale — is derived dynamically from the average High–Low range, keeping the filter dimensionless and consistent across all instruments and timeframes. Parameters Window Size (N) : The rolling lookback window of historical samples the kernel considers. Larger values produce a smoother, slower-responding line. Sparsity (s/N) : The core behavioral control. Dictates the fraction of the window allowed to "vote" on the estimate. Low Sparsity (e.g., 0.1) : Only the 10% of samples closest to the current estimate participate. Produces a snappy, selective line that locks tightly onto the most dominant price cluster. High Sparsity (e.g., 1.0) : All samples participate, resulting in a smoother, more conventional robust mean. Gamma (ε scale) : Controls the strength of outlier rejection. Lower values enforce harsh, median-like rejection. Higher values soften the rejection toward a standard weighted mean. Iterations : The number of reweighting convergence passes per bar. 2–3 is sufficient for practical convergence. Potential Applications The Hardy IRLS filter provides a unique lens into market structure by shifting the focus from simple time-averaged prices to spatial price consensus. Because it rejects price-distant wicks and noise spikes by design, it can serve as a foundation for: Custom trailing stops Dynamic support and resistance trackers Baseline trend or regime filters Feel free to inspect the open-source code, experiment with extreme sparsity and gamma settings, and integrate the IRLS core into your own quantitative projects. References Li Shuang, "Sparse Representation of Hardy Function by Iteratively Reweighted Least Squares," 2020 International Symposium on Computer Engineering and Intelligent Communications (ISCEIC), IEEE, 2020. DOI: 10.1109/ISCEIC51027.2020.00020
Chỉ báo Pine Script®
của Jamallo22
Trend Sniper v2.5 - Haar Wavelet Edition [Jamallo]Author's Note: The previous Trend Sniper v2.5 was built around a 2-Pole Butterworth Super Smoother with Parkinson historical volatility driving the trailing stops. This edition replaces that entire core with a Maximal Overlap Discrete Wavelet Transform (MODWT) using Haar basis functions and a Vervoort ATR trailing stop . Where the Butterworth produced smooth, continuous curves, the Haar wavelet produces structural steps — only updating when true market movement exceeds the calculated noise energy floor. Same Trend Sniper framework, completely different engine under the hood. Intro Trend Sniper v2.5 - Haar Wavelet Edition the Maximal Overlap Discrete Wavelet Transform (MODWT) using Haar basis functions. The result is a structural, step-based signal line that ignores minor price fluctuations entirely and only updates when true market movement exceeds a dynamically calculated noise threshold. Combined with a Vervoort ATR trailing stop, dual SuperTrend envelopes, and KAMA midpoint, this indicator provides a complete trend-following framework built on signal processing principles rather than traditional moving average logic. Breakdown Haar Wavelet Signal Line (MODWT + Adaptive Deadband) The core of this indicator is a 5-level cascaded MODWT Haar wavelet decomposition applied to smoothed Heikin-Ashi price. At each level, price is split into a smooth coefficient (trend) and a detail coefficient (noise). The user selects a decomposition level (1–5), controlling the structural scale — from 2-bar micro-structure up to 32-bar macro-structure. An adaptive deadband then wraps the smooth coefficient: the average absolute detail energy is measured over a lookback window and scaled by a multiplier. The signal line only steps to a new value when the smooth coefficient moves beyond this noise-energy threshold. This creates the characteristic "staircase" behavior — flat holds during noise, clean steps on real moves. The signal line colors by its own step direction: bull when it steps up, bear when it steps down. Vervoort ATR Trailing Stop A Sylvain Vervoort-style trailing stop is anchored directly to the Haar wavelet signal line rather than raw price. The stop distance is calculated as ATR × multiplier. In an uptrend, the stop ratchets upward and never retreats; in a downtrend, it ratchets downward. The stop flips when price closes through it. Because it's anchored to the structural wavelet line instead of noisy price, it produces cleaner, more decisive flip points. The trailing stop uses independent CMO (Chande Momentum Oscillator) + deadband hysteresis for its coloring — it only changes color when confirmed momentum cleanly breaches the CMO threshold, preventing color flicker during consolidation. Dual SuperTrend Envelopes Two SuperTrend calculations using smoothed Heikin-Ashi ATR create the outer structure: Slow SuperTrend (default mult 9.0) — the wide envelope defining macro trend boundaries Fast SuperTrend (default mult 6.0) — combined with the signal line to create the Fast Trigger dots The Fast Trigger is the midpoint of the fast SuperTrend and the Haar signal, plotted as orange circles for quick visual reference of momentum alignment. KAMA Midpoint & 4-State Fill A Kaufman Adaptive Moving Average smooths the midpoint between the slow SuperTrend and the Haar signal. The fill between the KAMA midpoint and the signal line uses four distinct color states based on whether price is above/below the slow SuperTrend and whether the signal is above/below the fast trigger — providing an immediate visual read on trend alignment and momentum phase. End The wavelet decomposition reveals the true step-by-step nature of price movement by mathematically separating signal from noise — use it alongside your own risk management and confluence analysis. Trade the structure, not the noise.
Chỉ báo Pine Script®
của Jamallo22
1
Trend filter v2 [Jamallo]Intro Trend Filter v2 is the evolution of the original Trend Filter, replacing the Kalman Filter baseline and WMA-scaled bands with a kernel-smoothed Vervoort Stop and percentile-ranked deviation bands. It also draws conceptual lineage from the ENW Kernel Bands indicator, inheriting the composite kernel and dual-Vervoort architecture — but rethinking both what the kernel is applied to and how bands are sized. Breakdown A Line — Vervoort Stop A classic trailing stop built on Heikin Ashi smoothed price and ATR. Color-coded by a trend state machine: bullish when price is above a rising A line, bearish when below a falling one, neutral otherwise. This is the signal line. K Line — Kernel-Smoothed Vervoort Stop Unlike ENW where the composite kernel is applied directly to raw price, v2 feeds a second Vervoort Stop output into the kernel. This means the K line is trend-filtered before smoothing even begins, producing a slower, more structurally stable baseline that is less reactive to noise. Signal Line — Independent Vervoort Stop A third Vervoort Stop with its own ATR settings, used as an additional trend reference. Percentile Bands The key structural departure from both v1 and ENW. Instead of scaling bands by smoothed volatility × a fixed multiplier, v2 ranks upside and downside deviations from K separately using `percentile_nearest_rank` over a configurable lookback. END v2 replaces the Kalman with a kernel-smoothed Vervoort Stop and swaps fixed-deviation bands for percentile-ranked asymmetric bands — a fundamental rethink of both the baseline and the band sizing at every level.
Chỉ báo Pine Script®
của Jamallo22
2
Exponential Nadaraya Watson kernel regression [Jamallo](2025) Intro Nadaraya-Watson (N-W) kernel regression is a non-parametric smoothing technique that estimates the underlying trend of a price series without assuming any fixed model shape (like a straight line or curve). Unlike a simple moving average which weights bars equally, or an EMA which applies a fixed exponential decay, N-W regression derives its curve by computing a weighted average of all prices within a lookback window — where the weights are determined by a kernel function. The most common kernel used is the Gaussian kernel, which assigns weights in a bell-curve shape — prices closer to the center of the window receive higher weight, prices at the edges receive lower weight. Most N-W implementations use a pure symmetric Gaussian kernel, meaning every bar within the lookback window is weighted purely by its distance from the center, with no preference for recency. This indicator uses a hybrid Exponential Nadaraya-Watson (ENW) kernel that combines two weighting forces simultaneously: Gaussian spatial weight — bell-curve weighting centered on the window, same as standard N-W Exponential time-decay weight — progressively heavier weighting on recent bars, similar to how an EMA behaves Both weights are multiplied together for each bar, meaning a price bar must be both spatially central and recent to receive maximum influence. In practice this shifts the effective weight peak toward the recent end of the window, making the K Line more responsive to current price action than standard N-W. Breakdown K Line — ENW Kernel Regression The central baseline of the indicator. A smooth adaptive curve derived from the hybrid ENW kernel applied to closing prices. Bandwidth is controlled via the Kernel Length and Alpha inputs — higher alpha increases the recency bias, lower alpha brings behavior closer to standard N-W. Volatility Bands (Inner & Outer) Rather than using ATR or standard deviation for band width, this indicator measures the absolute deviation of price from the K Line and smooths that deviation through the same ENW kernel. This means bands are fully adaptive — they expand and contract organically based on how far price has been straying from the regression curve, not a fixed statistical formula. Inner and outer bands are independently scaled via deviation multipliers. A Line — Vervoort ATR Stop A trailing stop built on the HLC4 price source with an ATR-based loss distance. Flips direction on a close beyond the stop level. Serves as the primary trend bias line — when above the K Line the fill turns bullish, when below it turns bearish. Cross signals (triangles) are plotted whenever the A Line crosses the K Line, marking potential trend shifts. Signal Line — Vervoort ATR Stop (Secondary) A second independent Vervoort trailing stop running on its own ATR period and multiplier settings. Typically configured looser than the A Line — wider multiplier, longer or equal period — so it acts as a slower confirmation layer. Useful for filtering noise on the A Line crosses: an A Line cross that also aligns with the Signal Line's bias carries more weight than one that doesn't. ------- Nadaraya-Watson kernel regression concept — E. Nadaraya (1964), G.S. Watson (1964) Vervoort ATR Stop — Sylvain Vervoort
Chỉ báo Pine Script®
của Jamallo22
5
Kalman Regime [Jamallo](2025) Intro Kalman Regime filters price through two stages — a Gaussian kernel weighted average followed by an adaptive Kalman filter — to produce a smooth, noise-resistant baseline. ATR is used throughout to make the system volatility-aware. Breakdown The Gaussian kernel pre-smooths price using a bell-shaped weight profile that blends recency bias (recent bars matter more) with center localization (edge bars contribute less). This pre-smoothed value is then passed into an adaptive Kalman filter, which recursively estimates the "true" price state by balancing how much to trust the new measurement versus the prior state — with that balance dynamically scaled by current ATR. The resulting baseline drives regime detection through a two-tier state machine. A strong signal fires when price breaks the ATR envelope and the baseline slope confirms direction. A weaker signal fires when price crosses the baseline and slope exceeds an ATR-scaled threshold. This dual-gate structure reduces whipsaws without adding lag. END A two-stage price filter combining a Gaussian kernel pre-smoother with an adaptive Kalman filter, using ATR envelopes and slope confirmation for regime detection.
Chỉ báo Pine Script®
của Jamallo22
Fractal Retracement [Jamallo](2025) Intro FRAMA is a moving average that adapts its speed based on fractal geometry — specifically, the fractal dimension (D) of recent price action. When price is trending strongly (low fractal dimension), it moves fast. When price is choppy/ranging (high fractal dimension), it slows down. This makes it far more responsive than a standard EMA or SMA. Breakdown: The indicator wraps this with a continuous range logic layer: the filtered line = k only moves if price breaks beyond the FRAMA ± ATR-based range, creating a stepped/ratcheting effect that filters out noise. Two sets of bands are plotted around the filtered line, scaled by ATR multiplied by user-defined multipliers (tight at 0.5×, medium at 1.0×). They're smoothed with a short EMA to reduce jitter, and filled with gradient colors for visual clarity. Direction is simply determined by whether k is rising or falling, and colors everything green (uptrend) or pink/red (downtrend). END In short, it's a noise-filtered trend indicator useful for identifying trend direction, dynamic support/resistance , and gauging how far price has retraced from the trend baseline.
Chỉ báo Pine Script®
của Jamallo22
1
Hodrick-Prescott filter [Jamallo](2025) Intro The Hodrick-Prescott (HP) filter is a tool originally developed by economists to remove short-term noise from data and reveal the underlying long-term trend — think of it like a moving average, but mathematically smarter. Instead of just averaging recent prices, it solves for the smoothest possible trendline that still follows the data. Applied to price, it produces a cleaner baseline than most traditional moving averages. --- However it comes with 3 problems in practical day trading. Problem #1 Standard HP Repaints History (core issue) The standard HP filter is non-causal — the value at any point in time depends on future values CEPR, meaning past signals can change retroactively. How I solved it: Switching to a one-sided implementation fixes this — it only looks at past data, so history never changes once a bar closes. Problem #2 One-Sided HP Hooks at Turns Going one-sided introduces an endpoint artifact — near turning points the trend line briefly curves in the wrong direction before reversing, producing a false signal. How I solved it: Feeding the HP output into a range filter solves this. The line only moves when HP breaks outside a volatility band, so the false hook gets absorbed into a flat zone and never fires a signal. Problem #3 Fixed Band Would Be Too Rigid A fixed band would whipsaw in volatile markets and lag in calm ones. How I solved it: Instead the band is sized dynamically using a percentile of True Range, scaling automatically with current market conditions.
Chỉ báo Pine Script®
của Jamallo22
Trend Sniper v2.5 [Jamallo](2025) Intro Trend Sniper v2.5 is built around a novel-unique core construction — the Butterworth Stop : a ratcheting trailing stop anchored to a 2-Pole Butterworth Super Smoother and dynamically sized by Parkinson Volatility. While each algorithm used is individually well-documented, this specific combination is rare and not found in standard indicator libraries — most Butterworth implementations are simply plotted as trend lines, not used as the structural anchor of a volatility-adaptive stop mechanism. Break down: 2-Pole Butterworth Super Smoother (Core Trend Line) (smoothed price) The backbone of the indicator is a 2-Pole Butterworth Super Smoother, a concept brought into trading by engineer and author John Ehlers, first published in his 2004 book Cybernetic Analysis for Stocks and Futures In this indicator: The Butterworth filter runs on close with a user-defined period (default 20) and produces bw_trend — a clean, lag-minimized version of price. This smoothed value is not plotted directly. Instead, it acts as the anchor point for the trailing stop. The stop is placed at bw_trend ± stop_dist , meaning the stop hugs the filtered trend rather than raw price, making it far less susceptible to wick noise and erratic bars. Without this pre-filtering step, the trailing stop would oscillate erratically on volatile bars. Parkinson Volatility (Sizes the Stop Distance) "Developed by physicist Michael Parkinson in 1980, it estimates volatility using the natural logarithm of the high-to-low ratio each period, capturing intraday range rather than just close-to-close movement." In this indicator: Parkinson Volatility is calculated over a rolling window (default 50 bars) and multiplied by stop_mult (default 1.2) to produce stop_dist . When markets are volatile, stop_dist expands and the stop gives price more room. When markets are quiet, it contracts and the stop tightens. Butterworth Stop State Machine (The Main Line You Watch) This is the primary plotted line, It's a classic ratcheting trailing stop built on top of outputs from steps 1 and 2. In this indicator: The logic is a simple state machine with two modes — uptrend and downtrend. In an uptrend, the stop only moves up (never down), tracking at bw_trend − stop_dist . In a downtrend, it only moves down, tracking at bw_trend + stop_dist . When price closes on the wrong side of the stop, the state flips and the stop resets on the other side of the trend line. Because both the anchor bw_trend and the buffer stop_dist are noise-filtered and volatility-adjusted, the BW Stop flips far less often than a raw price-based trailing stop would, keeping you in trends longer. Chande Momentum Oscillator + Deadband (Colors the BW Stop Line) "The CMO measures momentum by calculating the difference between the sum of gains and the sum of losses over a specified period, dividing by their total to produce a normalized scale from -100 to +100 that accounts for both up and down days simultaneously." In this indicator: CMO is calculated on bw_stop itself (not raw price), over 14 bars by default. A deadband of ±10 is applied — the line only turns bull-colored when CMO exceeds +10, and bear-colored when it drops below −10. It does not change color in between. KAMA Midpoint (The Dynamic Midline and Fill Zone Anchor) "Introduced by Perry Kaufman in 1995, KAMA dynamically adjusts its smoothing based on the relative efficiency of price movement". In this indicator: Rather than applying KAMA to price directly, it's applied to the midpoint between the BW Stop and the Slow SuperTrend. This midpoint is a computed halfway value between the two lines. KAMA then smooths that midpoint adaptively, producing `kama_mid`. The fill colors you see on the chart are drawn between kama_mid and bw_stop — not between price and anything else. Dual SuperTrend (Signals and Fill State Logic) "Created by Olivier Seban, SuperTrend combines ATR-based volatility with trend detection into a single line that repositions dynamically as price confirms or breaks the current trend direction." In this indicator: Two SuperTrend lines run simultaneously. The Slow ST (9× ATR) is the macro trend reference. The Fast ST (6× ATR) is never plotted at full size — it's compressed halfway toward the BW Stop and shown as orange circles fast_shortened Putting It All Together The flow through the indicator is linear: Butterworth smooths price → Parkinson sizes the stop → the state machine builds the stop line → CMO colors it → KAMA anchors the midline fill → the dual SuperTrend fires signals and shades conviction. END Every component feeds the next, meaning the final signals and visuals are the product of five sequential filters rather than any single calculation — which is what gives the indicator its resistance to false signals in noisy market conditions.
Chỉ báo Pine Script®
của Jamallo22
6
Trend filter [Jamallo](2025) This is a trend qualification and filtering tool. Breakdown: Signal Line — Vervoort ATR Trailing Stop The signal line is an implementation of the ATR Trailing Stop developed by Sylvain Vervoort, first published in Technical Analysis of Stocks & Commodities, Volume 27, Issue 6 (June 2009), in his article "Average True Range Trailing Stops" . The stop uses a modified ATR calculated on the average price (O+H+L+C)/4 rather than the plain close. Baseline — Kalman Filter The baseline is derived from the Kalman Filter, a recursive algorithm developed in 1960 by Rudolf E. Kálmán, a Hungarian-American engineer and mathematician. Originally developed for aerospace applications in the early 1960s — including guidance of the Apollo spacecraft. In this indicator, the Kalman Filter acts as a smooth, low-lag trend estimate. Rather than applying fixed weights to historical data like conventional moving averages, it dynamically adjusts its trust between the predicted trend and observed prices. Bands — Volatility Channels Two pairs of bands (inner and outer) are built around the Kalman baseline using a 200-bar WMA.
Chỉ báo Pine Script®
của Jamallo22
J - Trend Sniper v2(2024) A trend-following indicator that combines an ATR trailing stop with a smoothed price line to detect trend shifts and generate long/short signals. Calculates a slow ATR trailing stop (main trend boundary) and a fast ATR trailing stop, then plots a midpoint "Fast Line" between the two A smoothed signal line (a) tracks price using a volatility-weighted EMA — more reactive during high-volatility candles, smoother during low-volatility ones Long signal fires when the signal line crosses above the ATR trailing stop; Short when it crosses below
Chỉ báo Pine Script®
của Jamallo22
J - ATR Trend Channel(2022) A trend-following overlay that combines an ATR-based trailing stop with a smoothed price average to identify trend direction and generate signals. The trailing stop dynamically trails price using ATR volatility bands, tightening in trending conditions The smoothed line (a) blends open/close/high/low into a volatility-weighted average, acting as a trend filter Best used on trending markets to ride momentum and spot early reversals.
Chỉ báo Pine Script®
của Jamallo22
4