DAFE Trading Systems — CFE White Paper

// PAPER 03 — MATERIAL STATE THEORY

CFE

Coercive Field Engine.

Support and resistance are not lines — they are latent barriers with fatigue, damage, and healing. CFE tracks each level's material state in real time and reports the probability that the next attack breaks it. The field is live behind this text.

L0–L4 layers Fatigue model Break probability A2C adapter

// live stress field · price particles

The glowing plane is a barrier under attack. The amber point is price; the cyan halo is the field of latent resistance surrounding it. When accumulated damage crosses the fatigue threshold, the plane fractures — that is the CFE signal.

AuthorShaun Lear (DskyzInvestments)
DateJuly 3, 2026
ClassificationAcademic Research White Paper
StatusPUBLISHED

Abstract

Conventional support and resistance are drawn as static lines — but markets are not static, and neither are the levels they respect. This paper introduces the Coercive Field Engine (CFE), a framework that treats support and resistance as latent barriers with real material state: mass, load, fatigue, asymmetric damage, and participation-clocked healing. Barriers are organized into an L0–L4 layered architecture and the engine reports a live break probability per level. Rather than fighting the market with fixed zones, CFE coerces the market to reveal its own structural truth — the engine's job is to make the hidden material state of every level visible before the market does. Optional coupling to a Deep RL (A2C) trading adapter allows agentic parameter selection; the paper discloses where that coupling is aspirational and where it is live.

01The Static Line Fallacy

A level that has been tested nine times is treated identically to a level tested once — by both indicator lines and most traders. But structure degrades with every attack. The market's respect for a level is a material property that accumulates wear. CFE makes the wear measurable.

The engine converts the market's level-respect into a physics of stress and fracture: every touch deposits energy into the barrier; every break releases it. Tracked correctly, the fatigue curve of a barrier is observable — and it leads the price action.

02Latent Barrier Framework

A latent barrier B is characterized by a statistical profile rather than a line:

\[ \mu_{B} = \overline{\text{price}}_{touches} \qquad\qquad \sigma_{B} = \sigma_{\text{price}}(touches) \]

The barrier's half-width defines its influence zone — attacks inside this band are counted as touches:

\[ w_{B} = \frac{\sigma_{B}}{2} \]

With band-based identification:

\[ \text{touch}(p, B) \;=\; \mathbb{1}\!\left[\, |p - \mu_{B}| \le \frac{w_{B}}{2} \,\right] \]

Because every touch updates the profile, a barrier's location is never static — it creeps with the structure that creates it.

03Mass, Load, and Fatigue

Barriers accumulate mass from interactions, the heavier a level is, the more material it has to absorb:

\[ M_{B} = M_{B}^{\text{init}} + \sum_{\text{hist. touches}} \exp\!\left( -\frac{d^{2}}{2\sigma_{B}^{2}} \right) \]

The load at time t sums the influence of recent price with a time-decay kernel, keeping the field alive for every bar:

\[ L_{t} = \sum_{i=0}^{T-1} \exp\!\left( -\frac{t - i}{\tau} \right) \cdot f\bigl(p_{i} - \mu_{B}\bigr) \]

Damage is what attacks actually do — and it is asymmetric: approaching a level causes less damage than leaving it, mirroring how resistance is more fragile when attacked from below than above:

\[ \Delta D = \begin{cases} \kappa_{+} \cdot \dfrac{L_{t}}{M_{B}} & \text{attacking upward} \\[6pt] \kappa_{-} \cdot \dfrac{L_{t}}{M_{B}} & \text{attacking downward} \end{cases} \]
Attention asymmetry in practice In live CFE on TrendSpider, resistance zones attacked from below carry the full attack weight, while pullbacks from above register only ~40% of their nominal load. The asymmetry is what lets the engine price the last successful breakout higher than the ten failed touches before it.

04Healing and the Fatigue Threshold

Barriers heal — but only during participation, the window where price is inside the barrier's half-width band. Time outside the band does not heal; it only preserves:

\[ D_{t} \leftarrow \max\!\left( 0,\; D_{t} - \lambda \cdot D_{t} \right) \qquad \text{while } |p_{t} - \mu_{B}| \le \tfrac{w_{B}}{2} \]

When accumulated damage crosses the fatigue threshold, the barrier's structural integrity fails and break probability goes non-linear:

\[ B_{\text{eff}} = \frac{D_{t}}{D_{\text{thr}}} \qquad\qquad P_{\text{break}} = \frac{ B_{\text{eff}} }{ B_{\text{eff}} + e^{-T\cdot B_{\text{eff}}} } \]
Margin of safety The engine only fires when the margin is positive — the filtered climb rate required to reach the barrier is still achievable given remaining damage headroom:
\[ P_{\text{break}} > \varepsilon \quad \text{AND} \quad \text{slope}_{\text{required}} < \text{slope}_{\text{available}} \]

05Layered Architecture — L0 to L4

The engine resolves structure at five scales simultaneously:

L0Bar — raw bar data with volume and price footprint.
L1Pattern Layer — momentum and pattern metrics derived from bars.
L2Context Layer — market regime, volatility clustering, trend context.
L3Barrier Layer — the latent barriers with live fatigue state.
L4Decision Layer — break probabilities, margin checks, signal emission.

Higher layers read lower layers; the decision layer never sees raw price directly.

06The Optional A2C Adapter

Honest disclosure The paper describes an optional Deep RL (A2C — Advantage Actor-Critic) trading adapter. The live TrendSpider implementation does not run full A2C training in Pine; what ships is a rule-based policy that mirrors the A2C action/observation interface for future agent integration. Full training pipelines are aspirational — the barrier physics above is the live part; the agent is the roadmap.

// Live laboratory — attack the barrier

● LIVE

Move your mouse inside the field to drive price at the amber barrier. Load accumulates with every close approach; damage grows and heals on a participation clock; when damage crosses the fatigue threshold, the barrier fractures and a new one forms. The engine's break probability is computed exactly as in the paper.

LOAD: 0.00 DAMAGE: 0.00 BREAK PROB: 0.0% TOUCHES: 0 STATUS: INTACT
barrier μ load field broken / reformed

07Live Evidence

Production output from the CFE indicator on TrendSpider — fatigue curves and break probabilities plotted beneath the same market every other tool describes with straight lines. Chart screenshot forthcoming.

08System Parameters

System ParameterEvaluated Value / Setting
Latent Barrier Half-Widthσ/2 of touch cluster
Initial MassMinit = 1.0
Load Decay Time Constantτ = 50 bars
Fatigue ThresholdDthr = 0.80
Damage Asymmetryκ₊ = 1.00 up · κ₋ = 0.40 down
Healing Rateλ = 0.006 per bar in participation band
Break Probability EmissionBeff / (Beff + e^(−T·Beff))
ArchitectureL0–L4 layered barrier model

09Falsification Gates — Where This Model Can Be Proven Wrong

  1. If break frequency is independent of measured damage (D vs. actual break rate shows no monotonic relationship), the fatigue model fails.
  2. If healing occurs outside the participation band, the participation clock is wrong.
  3. If break probability is not monotone in Beff over a large out-of-sample set, the probability mapping fails.
  4. If the L0–L4 layering adds no predictive value over a flat two-feature model, the architecture is decorative.

10Conclusion

Every level in every market is a material under stress. CFE makes that stress legible — then converts it into the one number a trader needs: will this hold, or will it break? The framework is shipped live, the physics is testable, and the failure conditions are public.

// FRAMEWORK SERIES

Three frameworks. One coherent thesis.

DoD measures how unusual the volatility itself is. TED-PE places the market in kinematic phase space. CFE gives structure a material state. Read the series in order, or drop into the lab.

// DAFE TRADING SYSTEMS

Intelligent trading infrastructure — engineered, not assembled. Research-driven systems for real market conditions.

// CONTACT

Shaun Lear (DskyzInvestments)
Shaun@dafetradingsystems.com
(713) 725-8728
Houston, TX


Important: Trading Risk Disclosure. Futures, options, forex, and digital asset trading involves substantial risk of loss and is not suitable for everyone. Investors may lose all or more than their initial investment, so only risk capital should be used. Past performance is not indicative of future results.

© 2026 DAFE Trading Systems LLC — engineered, not assembled.