If predicting prices were really possible, OpenAI would be doing that instead of building ChatGPT. Tesla wouldn’t bother with factories. Amazon wouldn’t deal with thousands of employees and customers. And that’s just my two cents.
Those companies service me, I don't serve them. I think how unfortunate they have employees, buildings, inventory, legal problems, logistics, insurance and zillions of other problems every day. I see these companies as not so smart or successful. I have myself and a few other close researchers that ask each other for review of everything we do at no cost to each other. The value we give each other is honesty in the development process to keep us grounded to whats real. None of us look at big business and think they are the smartest of the species? Really? Every trader needs peer review and a mentor otherwise you will never have a clue how to discern what truth in trading is. m
I believe that if a Holy Grail of trading systems exist it's - Jerk math and beyond. Below is my personal journey of math beyond velocity and obscure data structures. 1. Theoretical Foundations of Spatial Jerk in Financial Markets In physics, jerk is the third derivative of position with respect to time (d3xdt3 \frac{d^3x}{dt^3} dt3d3x), measuring how quickly acceleration (the rate of change of velocity) changes. In financial markets, price serves as the analog to position, so jerk quantifies the rate of change of price acceleration, capturing sudden shifts in market dynamics, such as abrupt momentum surges or reversals. For example, on a ES Mini futures chart, jerk highlights moments when the market’s rate of momentum change (acceleration) itself changes rapidly, often signaling high-volatility events like news-driven spikes or institutional order flows. The term spatial jerk, emphasized, suggests a jerk-like metric that incorporates spatial dimensions of market data, beyond the purely temporal sequence of price changes. In financial markets, "spatial" can refer to: Price Levels: Key zones like support/resistance, pivot points, or moving averages, where price movements gain significance based on their proximity to these levels. Market Breadth: The distribution of advancing versus declining stocks (e.g., NYSE Advance Issues), which reflects the "spatial" strength or weakness across the market, not just a single asset. Order Book Structure: The spatial arrangement of bid/ask prices and order sizes, where rapid changes in depth or spread can mimic jerk-like behavior. Geometric Patterns: Sharp changes in the shape of price action (e.g., trendline breaks or parabolic moves), treated as spatial features on a price-time chart. Intermarket Relationships: Spatial correlations between assets (e.g., ES Mini futures and correlated indices like Nasdaq), where jerk might reflect shifts in relative positioning. Mark Brown’s Unique Context: Mark Brown, as the creator of Oddball Systems, is notable for his mechanical trading program that generates signals based on the momentum of market breadth (Advance Issues data) rather than direct price movements. This approach is inherently "spatial" because it considers the collective behavior of a broad set of stocks (a spatial distribution) rather than a single price series. Brown’s use of spatial jerk likely involves applying a jerk-like metric to this breadth momentum, capturing rapid changes in the rate of market participation (e.g., how quickly the acceleration of advancing stocks changes). This contrasts with traditional jerk, which focuses on a single asset’s price, making Brown’s method unique in its market-wide, spatially distributed perspective. 2. Mark Brown’s Unique Usage of Spatial Jerk Mark Brown’s Oddball Systems, published in Active Trader Magazine (December 2000), is a mechanical trading program designed for E-mini S&P index futures. Unlike conventional systems that rely on price-based indicators (e.g., moving averages, RSI), Oddball generates signals from the momentum of NYSE Advance Issues data, which measures the net number of stocks advancing versus declining. This market breadth metric reflects the "spatial" participation of the broader market, as it aggregates the behavior of thousands of stocks, creating a distributed view of market strength or weakness. Hypothesized Spatial Jerk in Oddball Systems While no public source explicitly details Brown’s use of "spatial jerk," his focus on breadth momentum suggests a unique application of jerk-like concepts to a spatially distributed dataset. Here’s how Brown might derive and use spatial jerk: Breadth as Position: Instead of price, Brown treats the net advance-decline line (Advance Issues minus Decline Issues) as a position-like variable. This represents the market’s "position" in terms of collective bullishness or bearishness across stocks. Velocity: The first derivative of the advance-decline line, measuring the rate of change in market breadth (e.g., how quickly more stocks are advancing). Acceleration: The second derivative, measuring the rate of change of breadth velocity (e.g., how quickly the pace of advancing stocks accelerates). Spatial Jerk: The third derivative, measuring the rate of change of breadth acceleration. High spatial jerk would indicate a sudden shift in the rate of market participation, such as a rapid surge in advancing stocks (e.g., during a market rally) or a sharp drop (e.g., during a panic sell-off). Brown’s unique usage lies in: Spatial Breadth Focus: By applying jerk to market breadth rather than price, Brown captures the spatial dynamics of the entire NYSE, reflecting the collective behavior of thousands of stocks. This is distinct from price-based jerk, which is confined to a single asset (e.g., ES Mini futures). Momentum-Driven Signals: Oddball’s signals are generated from breadth momentum, suggesting that spatial jerk could trigger trades when the market’s acceleration of participation changes abruptly, indicating a high-probability move in the S&P 500. Mechanical Precision: As a mechanical system, Oddball likely uses predefined thresholds for jerk magnitude or signal logic (e.g., buy when spatial jerk exceeds a positive threshold, sell when negative), ensuring systematic, repeatable trades. Why This Is Unique Traditional jerk in financial markets focuses on price derivatives of a single asset, often ignoring the broader market context. Brown’s approach, by contrast: Leverages Market Breadth: Uses a spatially distributed metric (Advance Issues) to capture market-wide dynamics, making jerk a measure of collective momentum shifts. Avoids Price Noise: Breadth data is less noisy than price, as it aggregates many stocks, potentially making spatial jerk a more stable signal for trading index futures. Anticipates Price Moves: Rapid changes in breadth acceleration (spatial jerk) may precede price movements in the S&P 500, giving Brown’s system a predictive edge. 3. General Methods for Arriving at Spatial Jerk To contextualize Brown’s unique usage, I’ll outline general methods for deriving spatial jerk in financial markets, emphasizing how they might align with or differ from his approach. These methods combine temporal derivatives with spatial market features, ensuring a comprehensive exploration. Method 1: Discrete Differencing with Spatial Contextualization Concept: Compute jerk as the third derivative of a position-like variable (e.g., price or breadth) and adjust it for spatial market features like price levels or volatility. Steps: Select a position variable: For Brown: Net advance-decline line (Advance Issues minus Decline Issues). For general use: Price (e.g., ES Mini) or a composite metric (e.g., weighted index of prices). Compute velocity as the smoothed change in position per step. Brown: Rate of change of advance-decline line, possibly smoothed via regression or exponential decay. General: Price difference or regression slope over a lookback window (e.g., 20 bars). Compute acceleration as the change in velocity per bar. Compute jerk as the change in acceleration per bar. Add spatial context: Brown: Normalize jerk by the breadth’s volatility (e.g., standard deviation of advance-decline changes) or compare to historical breadth extremes (e.g., overbought/oversold zones). General: Normalize by market volatility weight by proximity to key price levels. Advantages: Simple to implement in trading platforms like MultiCharts. Leverages existing smoothing techniques to reduce noise. Brown’s use of breadth makes jerk spatially robust, capturing market-wide shifts. Challenges: Jerk is noisy, requiring careful smoothing (Brown likely uses exponential decay, similar to his momentum calculations). Intrabar repainting occurs with tick-based data, though fixed post-bar close. Brown’s Uniqueness: Using breadth instead of price shifts the spatial focus to market participation, potentially anticipating index moves before price confirms. Method 2: Polynomial Regression for Spatial Trajectories Concept: Fit a polynomial to the position variable and compute analytical derivatives, adjusting for spatial features like price zones. Steps: Fit a polynomial (e.g., cubic, P(t)=at3+bt2+ct+d P(t) = at^3 + bt^2 + ct + d P(t)=at3+bt2+ct+d) to the position variable over a lookback window. Brown: Fit to the advance-decline line. General: Fit to price or a breadth-weighted index. Compute derivatives: Velocity: V(t)=3at2+2bt+c V(t) = 3at^2 + 2bt + c V(t)=3at2+2bt+c. Acceleration: A(t)=6at+2b A(t) = 6at + 2b A(t)=6at+2b. Jerk: J(t)=6a J(t) = 6a J(t)=6a. Evaluate jerk at the current time. Contextualize spatially: Brown: Weight jerk by the breadth’s position relative to historical highs/lows (e.g., breadth percentiles). General: Scale jerk by distance to a moving average or volatility bands (e.g., Bollinger Bands). Advantages: Captures non-linear trends, potentially more accurate for complex market moves. Brown’s breadth focus could smooth polynomial fits, reducing overfitting. Challenges: Computationally intensive, less practical for real-time trading. Polynomial degree and lookback window require tuning to avoid artifacts. Brown’s system likely avoids this complexity, favoring simpler differencing. Brown’s Uniqueness: If applied, Brown might use breadth’s smoother trends to fit polynomials, but his mechanical approach suggests he prefers discrete methods for speed. Method 3: Order Book Spatial Jerk Concept: Derive jerk from the spatial distribution of bid/ask prices or order sizes, treating the order book as a spatial domain. Steps: Define a position metric, e.g., the order book’s center of mass: POB(t)=∑pp⋅OrderSize(p,t)∑pOrderSize(p,t) P_{\text{OB}}(t) = \frac{\sum_p p \cdot \text{OrderSize}(p, t)}{\sum_p \text{OrderSize}(p, t)} POB(t)=∑pOrderSize(p,t)∑pp⋅OrderSize(p,t) where p p p is a price level. Compute velocity, acceleration, and jerk: Velocity: Change in center of mass per bar. Acceleration: Change in velocity per bar. Jerk: Change in acceleration per bar. Add spatial context: Brown: Correlate order book jerk with breadth jerk, as breadth may reflect liquidity shifts. General: Normalize by order book spread or depth. Advantages: Captures spatial liquidity dynamics, highly relevant for futures trading. Brown’s breadth signals might proxy order book shifts indirectly. Challenges: Requires high-frequency order book data, not typically available in standard charting platforms. Brown’s Uniqueness: Brown’s focus on breadth momentum, achieving a similar spatial effect with more accessible NYSE data. Method 4: Geometric Spatial Jerk Concept: Measure jerk in the context of price or breadth trajectory curvature, treating the chart as a spatial manifold. Steps: Fit a curve (e.g., spline) to the position variable (price or breadth). Compute curvature: κ(t)=∣P¨(t)∣(1+P˙(t)2)3/2 \kappa(t) = \frac{|\ddot{P}(t)|}{(1 + \dot{P}(t)^2)^{3/2}} κ(t)=(1+P˙(t)2)3/2∣P¨(t)∣ where P˙=dPdt \dot{P} = \frac{dP}{dt} P˙=dtdP, P¨=d2Pdt2 \ddot{P} = \frac{d^2P}{dt^2} P¨=dt2d2P. Define spatial jerk as the rate of change of curvature: JSpatial(t)=dκdt J_{\text{Spatial}}(t) = \frac{d\kappa}{dt} JSpatial(t)=dtdκ Relate to market structure (e.g., breakouts near resistance). Brown: Apply to breadth trajectory, detecting sharp changes in market participation patterns. General: Apply to price, focusing on trendline breaks or parabolic moves. Advantages: Highlights spatial patterns like breakouts, aligning with visual chart analysis. Brown’s breadth focus could reveal market-wide pattern shifts. Challenges: Complex to compute in real-time; requires subjective pattern definitions. Brown’s mechanical system likely prioritizes simpler metrics. Brown’s Uniqueness: Breadth-based curvature changes could anticipate index breakouts, but Brown’s approach probably uses differencing for simplicity. Method 5: Multidimensional Spatial Jerk Concept: Compute jerk in a high-dimensional space of market variables (e.g., price, breadth, volume, volatility). Steps: Define a state vector, e.g., S(t)=[Price,Breadth,Volume,Volatility] \mathbf{S}(t) = [\text{Price}, \text{Breadth}, \text{Volume}, \text{Volatility}] S(t)=[Price,Breadth,Volume,Volatility]. Compute derivatives: V(t)=dSdt,A(t)=dVdt,J(t)=dAdt \mathbf{V}(t) = \frac{d\mathbf{S}}{dt}, \quad \mathbf{A}(t) = \frac{d\mathbf{V}}{dt}, \quad \mathbf{J}(t) = \frac{d\mathbf{A}}{dt} V(t)=dtdS,A(t)=dtdV,J(t)=dtdA Project jerk onto a scalar: JSpatial(t)=JPrice2+JBreadth2+JVolume2+JVolatility2 J_{\text{Spatial}}(t) = \sqrt{J_{\text{Price}}^2 + J_{\text{Breadth}}^2 + J_{\text{Volume}}^2 + J_{\text{Volatility}}^2} JSpatial(t)=JPrice2+JBreadth2+JVolume2+JVolatility2 Contextualize spatially: Brown: Weight breadth jerk heavily, reflecting Oddball’s focus. General: Adjust for correlations between assets or market regimes. Advantages: Captures complex spatial-temporal interactions. Brown’s breadth focus aligns with multidimensional analysis. Challenges: Requires sophisticated modeling, impractical for simple trading systems. Brown’s mechanical approach likely avoids this complexity. Brown’s Uniqueness: By focusing on breadth, Brown simplifies multidimensional analysis to a single, spatially rich metric. Practical Implementation in a Trading Environment To implement spatial jerk in a trading platform for ES Mini futures: Data Collection: Use price (close) and breadth data (e.g., NYSE Advance Issues). Brown accesses breadth via market data feeds. Smoothing: Apply exponential smoothing or regression to reduce noise, as Brown likely does for breadth momentum. Jerk Calculation: Compute discrete differences (velocity → acceleration → jerk), normalizing by volatility or breadth extremes. Signal Generation: Brown: Trigger buy/sell signals when breadth jerk exceeds thresholds, anticipating S&P 500 moves. General: Use jerk peaks near price levels or during volatility spikes. Debugging: Log jerk values and signals for validation, ensuring signals align with market events (e.g., breadth surges during rallies). Testing: Backtest on historical ES Mini data, focusing on high-jerk events (e.g., FOMC announcements). Financial Applications and Challenges Applications: Scalping: Brown’s spatial jerk, derived from breadth, could trigger short-term trades in E-mini futures during rapid market shifts. Reversal Detection: High jerk near price levels or breadth extremes may signal overbought/oversold conditions. Market Timing: Breadth jerk anticipates index moves, giving Brown’s system an edge in volatile sessions. Risk Management: High jerk warns of erratic conditions, prompting reduced exposure. Challenges: Noise: Jerk’s sensitivity requires robust smoothing, which Brown addresses via breadth’s aggregated nature. Data Access: Breadth data is less common than price data, but Brown’s system leverages NYSE feeds. Repainting: Intrabar updates affect real-time signals, though fixed post-bar close. Complexity: Spatial context (e.g., normalizing by volatility) adds computational overhead, but Brown’s mechanical approach keeps it manageable. Brown’s Unique Contribution Mark Brown’s use of spatial jerk is unique because it: Focuses on Breadth Momentum: By deriving jerk from NYSE Advance Issues, Brown captures market-wide participation shifts, a spatial perspective absent in price-based systems. Anticipates Price: Breadth jerk likely precedes S&P 500 price moves, leveraging the spatial distribution of stock movements. Simplifies Spatial Analysis: Uses a single, aggregated metric (breadth) to achieve spatial insights, avoiding complex multidimensional models. Mechanical Execution: Embeds jerk in a systematic, rule-based system (Oddball), ensuring consistent application. Potential Extensions Jounce: Fourth derivative (d4Pdt4 \frac{d^4P}{dt^4} dt4d4P), though noisy and less practical. Breadth-Price Hybrid: Combine breadth jerk with price jerk for confirmation. Intermarket Jerk: Apply jerk to correlated assets (e.g., Nasdaq futures), enhancing spatial context. Machine Learning: Use breadth jerk as a feature in predictive models, though Brown’s mechanical approach avoids such complexity. Conclusion Mark Brown’s unique usage of spatial jerk likely involves deriving a jerk-like metric from the momentum of NYSE Advance Issues, capturing rapid changes in market breadth’s acceleration to anticipate E-mini S&P 500 moves. General methods for spatial jerk include: Discrete Differencing: Compute jerk from price or breadth, normalized by volatility or price levels. Polynomial Regression: Fit curves for analytical derivatives, adjusted for spatial features. Order Book Analysis: Derive jerk from bid/ask dynamics, though Brown uses breadth instead. Geometric Analysis: Measure curvature changes, less likely in Brown’s system. Multidimensional Analysis: Compute jerk across market variables, simplified by Brown’s breadth focus. Brown’s approach stands out for its spatially distributed, breadth-based perspective, making spatial jerk a powerful tool for systematic trading. by Mark Brown Grok3 edited
Good grief that's a lot to read. Too technical, too Anal. .. In summary.... Trading is part art, Part science Occam's razor...the simplest explanation is usually the correct one, more or less Brown of Mark
Every time you take the derivative of a variable, the result is more noise than you started with. If you start with very noisy data and then fit a model with enough parameters, you can make it say anything you want. I can't see why you're adding jerk or multiple different parameters together. Even it made sense to do that I would expect some kind of normalization first.