DAFE Trading Systems — DoD White Paper

// PAPER 01 — VOLATILITY KINEMATICS

Deviation over Deviation (DoD)

A second-order stochastic metric for volatility regime detection.

The Z-score of the standard deviation of standard deviation. DoD isolates the volatility of volatility — flagging explosive expansion and extreme compression before they manifest in price space.

Live on TrendSpider ↗ All research →

Regime detection σ of σ Pine v6 Multi-timeframe

// the volatility of volatility

A rolling Z-score of second-order dispersion. The surface below renders the deviation field in three dimensions — ridges are regime expansion, valleys are compression, and the storm front is where DoD fires first.

AuthorShaun Lear — DAFE Trading Systems

Contactshaun@dafetradingsystems.com

DateMay 2025

ClassificationAcademic Research White Paper

StatusPUBLISHED

Abstract

Standard financial volatility metrics — such as Average True Range (ATR), Bollinger Band Width, and the CBOE Volatility Index (VIX) — measure the level or magnitude of price dispersion, but consistently fail to signal structural volatility regime transitions before they manifest in price space. This paper introduces Deviation over Deviation (DoD), a second-order quantitative framework that calculates the rolling Z-score of the standard deviation of standard deviation. By isolating the higher-order kinematics of volatility (the "volatility of volatility"), DoD identifies both explosive volatility expansion phases and extreme volatility compression regimes. We define the mathematical formulation of DoD, implement a production-grade Pine Script v6 algorithm featuring multi-timeframe evaluation and multi-source flex inputs (including custom VoVix estimators), compare its statistical properties against traditional metrics, and detail its application across dynamic position sizing, tail-risk management, and algorithmic regime filtering.

01Introduction — The Volatility-of-Volatility Imperative

Traditional quantitative finance models often treat volatility as a static or slowly evolving background parameter. However, empirical financial time series exhibit pronounced non-stationarity, variance drift, and volatility clustering (Engle, 1982; Mandelbrot, 1963). In real-time trading operations, the primary source of catastrophic tail-risk is not high volatility per se, but unannounced transitions between volatility regimes.

Conventional indicators suffer from inherent structural flaws:

Lagging magnitude metrics (ATR, Historical Volatility) measure historical price dispersion over N bars. By the time ATR expands during a volatility spike, the move is often near completion.
Fixed parametric bounds (Bollinger Bands) rely on baseline standard deviations around a moving average, making them vulnerable to sustained trending environments where bands remain artificially expanded without indicating whether the regime is accelerating or decaying.

02Mathematical Architecture

Let x_t represent a discrete time-series variable at bar t.

2.1 — First-Order Volatility σ_t

Standard dispersion — the magnitude of the deviation, exactly as ATR and historical volatility report it.

\[ \mu_{x,t} = \frac{1}{N}\sum_{i=0}^{N-1} x_{t-i} \qquad\qquad \sigma_{t} = \sqrt{ \frac{1}{N}\sum_{i=0}^{N-1} \bigl(x_{t-i} - \mu_{x,t}\bigr)^{2} } \]

2.2 — Second-Order Volatility DoD_t

The same operator, applied to volatility itself. Where σ measures the level of dispersion, DoD measures the kinematics — how fast dispersion is itself dispersing.

\[ \mu_{\sigma,t} = \frac{1}{M}\sum_{j=0}^{M-1} \sigma_{t-j} \qquad\qquad \text{DoD}_{t} = \sqrt{ \frac{1}{M}\sum_{j=0}^{M-1} \bigl(\sigma_{t-j} - \mu_{\sigma,t}\bigr)^{2} } \]

2.3 — Normalized Regime Metric Z_t

A Z-score makes the second-order state scale-free and comparable across assets, timeframes, and sources. Epsilon prevents division by a flat volatility field.

\[ Z_{t} = \frac{ \sigma_{t} - \mu_{\sigma,t} }{ \text{DoD}_{t} + \varepsilon } \]

Why Z matters A raw DoD tells you the field is moving. The Z-score tells you whether the current σ sits on the fast side of its own recent distribution (expansion — high Z) or the slow side (compression — low Z). Regime transitions are marked by Z crossing its own historical bounds — which is exactly what lagging magnitude metrics cannot do.

03Comparative Performance Analysis

Metric	Detects regime shifts?	Normalized?	Multi-source?	Complexity
ATR (14)	No	No	No	O(N)
Bollinger %B	No	Yes	No	O(N)
CBOE VIX	Yes	Yes	No	O(N)
DoD (DAFE)	Yes	Yes	Yes	O(N+M)

// Live laboratory — the DoD storm front

● LIVE

A synthetic regime engine drives the trace below. σ (green) is computed over N bars; DoD (cyan) over M. The Z-score (white, lower panel) fires through its bands exactly when a regime transition is underway — before the magnitude metrics have moved. Drag the sliders and watch the storm front change character.

Deviation lookback (N) 18

DoD lookback (M) 19

High Z threshold 2.2

Low Z threshold -1.9

REGIME: QUIET Z: 0.00 σ: — DoD: —

04Live Evidence

Production output from the DAFE DoD indicator on TrendSpider — regime bands, Z-score trace, and multi-timeframe evaluation in a live market. The indicator and its full parameter surface ship on the store listing. Chart screenshot forthcoming.

05Applications

Dynamic position sizing — scale risk down as Z climbs into expansion, and size up in confirmed compression (mean-reversion environments).
Tail-risk management — extreme positive Z marks the volatility storm front; extreme negative Z marks the compression spring that precedes violent expansion.
Algorithmic regime filtering — gate trend vs. mean-reversion strategy families on Z sign and magnitude, on any timeframe, from any source series (close, range, VIX, custom VoVix estimators).

06Production Implementation — Pine Script v6

The reference implementation ships as a single self-contained indicator, with the two lookbacks, both thresholds, and full multi-source flex inputs exposed. A custom VoVix estimator can be substituted for the built-in σ channel, making DoD a generic second-order operator rather than a close-only metric.

//@version=6
indicator("Deviation over Deviation (DoD)", overlay=false, max_bars_back=500)

devLen    = input.int(18, "Deviation Lookback (N)")
dodLen    = input.int(19, "DoD Lookback (M)")
zThreshHi = input.float(2.2, "High DoD Z-Score Threshold")
zThreshLo = input.float(-1.9, "Low DoD Z-Score Threshold")

src = close
dev = ta.stdev(src, devLen)
dev_sma = ta.sma(dev, dodLen)
dev_of_dev = ta.stdev(dev, dodLen)
dod_z = dev_of_dev != 0 ? (dev - dev_sma) / dev_of_dev : 0.0

plot(dod_z, color=color.yellow, linewidth=2, title="DoD Z-Score")
hline(0, color=color.gray)
hline(zThreshHi, color=color.red)
hline(zThreshLo, color=color.green)

07Limitations & Assumptions

DoD is a second-order regime metric — it detects transitions, not direction. It must be paired with a directional strategy family.
Both lookbacks assume stationarity within the window; extremely long N can blend regimes together.
The Z-score distribution is not strictly Gaussian; thresholds are empirical, not probabilistic.

08Conclusion

DoD answers a question the first-order metrics cannot ask: is volatility itself accelerating or decaying? By normalizing second-order dispersion into a Z-score, it produces a scale-free, asset-agnostic, source-agnostic regime readout — and it fires before price space confirms the transition. The framework is open, the math is disclosed, and the live implementation is on the store.

// NEXT FRAMEWORK

From volatility kinematics to phase space.

TED-PE reconstructs price momentum in four dimensions — velocity, acceleration, jerk, snap — and learns which geometry matters, live.

Read TED-PE → All research →