The RC-4 Lattice™ Architecture

1. Introduction

Most retail automated trading systems reduce entry logic to a single condition: a moving-average crossover, a single oscillator threshold, a single candlestick pattern. This is computationally simple but discards information a discretionary trader would naturally use — the fact that a setup is more trustworthy when several independent pieces of evidence agree with each other on the same bar.

RockBot Nexus was built on the opposite premise, and its own internal engineering vocabulary — introduced fully in Section 2 — was developed alongside the system specifically to give its operator ownable, precise language for describing what is otherwise a genuinely novel combination of well-understood technical-analysis primitives. No single signal is treated as sufficient justification for a trade on its own; every qualifying setup contributes a weighted vote to a running conviction score inside the ConfluenceStack™, and that score is what ultimately determines whether a trade is taken and how large it is sized.

2. Proprietary Terminology at a Glance

The following terms are introduced and used throughout this paper to name real, specific subsystems of RockBot Nexus. Each is defined precisely where it is first used in depth; this table is a quick-reference summary.

Table 1 — Proprietary terminology index. Full definitions also appear in Appendix A.

3. The RC-4 Lattice™

At the center of RockBot Nexus are four stochastic oscillators of increasing period length — collectively, the RC-4 Lattice™:

Table 2 — The RC-4 Lattice's four rotation bands and their functional roles.

A Quad Rotation Event is defined as all four RC-4 bands crossing into aligned extreme territory simultaneously — a rare, high-conviction condition treated as a standalone confirmation bonus wherever it appears in the signal library (Section 5.7).

4. ConfluenceStack™ Scoring Methodology

Every bar, RockBot Nexus computes two independent scalar values — a long ConfluenceStack™ score and a short ConfluenceStack™ score — as the sum of the confluence weights of every signal currently true in that direction. Table 3 lists the full weight schedule as implemented in the current production build.

Table 3 — ConfluenceStack™ weight schedule. Weights marked elsewhere in this paper as design-calibrated (not backtest-derived) remain user-adjustable in the strategy's own parameters.

4.1 ConvictionBoost™

Two context-dependent bonuses can stack onto an already-qualifying ConfluenceStack™ score, collectively referred to as ConvictionBoost™:

• Key-Level Proximity (+0.5) — price is within a fixed tick tolerance of the session's prior-day or current-day high/low, VWAP, or the 200-period moving average.
• Reversal Candle (+0.5) — a Hammer, Shooting Star, or Engulfing pattern coincides with the qualifying signal's own trigger bar, anchored to the correct bar for each signal type so an unrelated candle is never credited.

The final ConfluenceStack™ score is converted to a contract quantity via linear interpolation against a configurable ceiling, floored at one contract for any qualifying signal and capped at the user's Max Contracts setting.

5. Signal Library

The following seven detectors constitute the complete RockBot Nexus signal library, each independently evaluated every bar.

5.1 Trendline Touch

A live, tick-resolution check for price intersecting a trendline within a fixed tick tolerance. By default the system considers only user-drawn lines, so the operator's own read of chart structure governs this signal rather than a self-generated one. A cross-detection guard ensures the signal fires once per discrete touch event rather than continuously. Because this signal is exempt from the universal entry filters described in Section 6, its only independent safeguard against a stale or forgotten line is FreshnessGate™ (Section 8.1).

5.2 Strict Quad Divergence

The most rigorously gated signal in the library. Requires, simultaneously: fifteen consecutive strictly-decreasing (or increasing, for shorts) lows/highs, four further equal-or-extending bars, the 9-3 band rising through a defined mid-range window, the 14-3 band confirming below its midpoint, the 40-4 band rising, and the 60-10 band below a low extremity threshold. No single condition is sufficient.

5.3 Bull / Bear Flag

Detects a momentum pause within an established directional extension: the 60-10 band must have held at or beyond a user-configurable threshold for at least five consecutive bars, followed by a 9-3 cross through a fixed trigger level, an EMA-proximity-and-side filter, and a minimum-magnitude pre-trend filter distinguishing a genuine pullback from an already-reversing trend. These three conditions are permitted to occur within a shared bar window rather than requiring exact same-bar coincidence.

5.4 20 EMA Reversal

A mean-reversion setup: price within a fixed tick tolerance of the 20-period EMA while the 60-10 band sits in extreme territory. Implemented as a one-time cross-detection event to prevent continuous re-firing.

5.5 Full Lattice Alignment

A compound signal requiring a live trendline touch to coincide with all four RC-4 bands simultaneously in aligned extreme territory. Deliberately bypasses the system's standard entry filters (Section 6) on the basis that the compound condition is already more selective than any individual filter it would otherwise be subject to.

5.6 Price / Stochastic Divergence

A classical bullish/bearish divergence pattern implemented as an explicit multi-bar state machine: a qualifying first extreme, a bounce phase, and a retest phase in which price revisits the original extreme's body-price while the 9-3 band confirms a genuine directional turn. The price-touch condition and the oscillator turn-confirmation condition are permitted to resolve on different bars within the retest phase, correctly handling gradually-developing reversal shapes rather than only sharp ones.

5.7 Quad Rotation Divergence — The Apex Signal

The system's highest-weighted signal, and the only one referred to internally as the Apex Signal: identical in structure to Price/Stochastic Divergence, with the additional requirement that a genuine Quad Rotation Event (Section 3) was observed during the setup's development. The only signal eligible for the Apex Override (Section 9).

6. Entry Gating and the Bypass Doctrine

Two universal filters apply to signals 5.1–5.4: a Universal Stochastic Filter (no long entry unless the 9-3 band is below a configurable ceiling; the mirror applies to shorts) and a Higher-Timeframe Confirmation (a secondary, higher-timeframe close must be on the correct side of its own 20-period EMA).

Signals 5.5–5.7 are exempt from both — the Bypass Doctrine. This is deliberate: each of these three signals already encodes a stronger, more specific condition than either universal filter would independently enforce, and empirical testing showed the universal filters actively suppressing legitimate instances of these three patterns without a corresponding reduction in false positives.

OverextensionGate™ (Section 8) independently governs new entries with the same exception: Price/Stochastic Divergence and Quad Rotation Divergence are exempt, since their defining structural premise — a sustained low or high 60-10 reading — is the same condition the gate is designed to flag as overextended.

7. AdaptiveGuard™ — The Unified Exit Layer

An earlier design iteration maintained separate exit logic per signal family, each re-implementing a similar strength-detection and trailing concept at slightly different thresholds. This was consolidated into AdaptiveGuard™: one mechanism, applied identically to every open position regardless of entry signal, after threshold collisions between the per-signal implementations were found to render the more specific logic unreachable in practice.

AdaptiveGuard™ is checked first, before any other position-management logic, on every bar. Its trigger level (default 80/20 on the 9-3 band) is user-adjustable. At the moment of trigger, two conditions are evaluated before any exit executes: a Profit Cushion (the position is profitable by at least one multiple of its own original stop-loss distance) and Momentum Strength (the 60-10 band is beyond its own confirmation threshold and actively moving further in the position's favor bar-over-bar).

If both hold, the stop moves to breakeven and a persistent, volatility-adaptive trail arms. If either fails, AdaptiveGuard™ checks for a same-bar Bull/Bear Flag pattern in the opposing direction and attempts a full reversal before falling back to a flat exit. Once armed, the trail distance recalculates every bar:

This produces a trail that widens automatically during elevated volatility and tightens during calm conditions, rather than holding a fixed tick distance for the remainder of the trade.

7.1 The Ambiguity Resolution Rule

The Momentum Strength check above was originally implemented as two explicit branches at the moment 9-3 rolls back out of its extreme: the 60-10 band has genuinely recovered past its confirmation threshold (flat exit), or it remains below threshold and is still declining (arm the trail). A third state is structurally reachable and was left unhandled by either branch — 60-10 below its confirmation threshold, but rising rather than declining. This is not an edge case; a market cooling off from an extreme routinely passes through exactly this transitional state.

Under the original two-branch logic, a position reaching this state received no exit, no breakeven lock, and no armed trail — fully exposed with no protective mechanism engaged. The Ambiguity Resolution Rule closes this: both directions of AdaptiveGuard™ now carry an explicit default for this third state, resolving to a flat exit rather than inaction. An indeterminate reading is treated as insufficient evidence of continued strength, not as a reason to do nothing.

8. OverextensionGate™ and the Risk Control Suite

Table 4 — Either tier independently triggers a lockout on new entries; both are checked every bar.

Independent of AdaptiveGuard™ and OverextensionGate™, the following controls are available:

Table 5 — Independent risk controls, all optional and user-configurable.

8.1 FreshnessGate™

Trendline Touch is exempt from the universal entry filters by design (Section 6), so a genuine touch of a drawn line is always taken on its own merits. That design choice has a structural consequence: nothing in the signal's own logic imposes an expiration on the line itself. A line drawn days earlier — off-screen, on an unopened chart panel, or simply forgotten by the operator — remains fully live and continues to trigger real entries indefinitely.

FreshnessGate™ (on by default, five-day threshold) closes this gap independently of the other entry controls: any drawn line — Line, Extended Line, Horizontal Line, or Trend Channel boundary — older than the configured threshold is excluded from evaluation entirely, not merely deprioritized. Age is measured from the line's older anchor point, i.e. when it was originally drawn, rather than from the present relative to a more recently adjusted anchor.

9. ProfitLock™ — Fixed-Target Override

ProfitLock™ is an optional, off-by-default mechanism evaluated before every other check in the position-management sequence, including AdaptiveGuard™ itself. When enabled, it closes any open position the instant its unrealized profit reaches a fixed, user-configured dollar target, regardless of what any signal-specific or universal exit rule would otherwise decide.

This is presented separately from the risk-control suite in Section 8 because it is not a protective mechanism in the same sense — it is a deliberate behavioral override with a real cost. Because ProfitLock™ runs first and unconditionally, it will also curtail trades that AdaptiveGuard™'s own strength-detection logic would have allowed to run into a substantially larger ATR-trailed extension. The trade-off is certainty of a smaller, faster win over the possibility of a larger one; it ships disabled by default because that trade-off is a deliberate operator choice, not a universally correct default.

The target is evaluated once per bar against whole-position unrealized profit, not per contract and not tick-by-tick, so the realized exit on a fast-moving bar can land at or slightly beyond the configured target rather than at it precisely.

10. ContinuityGuard™ — Restart State Recovery

NinjaTrader fully reinitializes a strategy's internal state on every restart — a deliberate re-enable, a recompile, or a platform reconnect — regardless of whether a position happens to be open at that moment. Prior to ContinuityGuard™, the strategy's own tracking of the open position's entry price and originating signal type was unconditionally reset to flat-state defaults as part of that reinitialization, producing two simultaneous failures: a breakeven-lock exit call would submit a stop order at a price of zero, rejected outright by the broker, and the position's entry-type classification would reset to a null state that routes to no entry-type-specific exit rule at all, leaving it managed only by the loosest universal checks.

ContinuityGuard™ detects this condition on initialization: if a position is already open at restart, the strategy rebuilds its entry-price tracking directly from the live position's actual average price and assigns a conservative default entry-type classification, so the position remains under active AdaptiveGuard™ management from the moment of restart rather than falling through to an unmanaged state. Each recovery event is logged explicitly for operator verification. ContinuityGuard™ restores continuity of management going forward; it does not retroactively recover which of the seven signals originally opened the trade, since that information does not survive a full state reset.

11. Position Sizing and the Apex Override

Baseline position size is a direct, deterministic function of the ConfluenceStack™ score, floored at one contract and capped at the user's configured maximum. The Apex Override is the single documented exception: when the Apex Signal (Quad Rotation Divergence) coincides with both a live trendline touch and a reversal candle confirmed specifically at the divergence retest bar, the position sizes at the full configured maximum regardless of the computed score — on the basis that this specific four-condition intersection represents the highest empirical conviction the system is capable of expressing.

12. Discussion and Limitations

• Several ConfluenceStack™ weights (20 EMA Reversal, the divergence ConvictionBoost™ weight, the AdaptiveGuard™ ATR multiplier) were calibrated by relative conviction tier and adjusted iteratively based on observed trade outcomes, rather than derived from formal statistical optimization. All remain user-adjustable.
• The Bypass Doctrine means the three exempted signals carry structurally less entry-stage downside protection than the other four, by design. This is a trade-off between suppressing false positives and limiting worst-case entries, not a free improvement in either direction.
• Trendline-based signals depend on the operator's own charting discipline by default; signal quality for this subset is bounded by the quality of the underlying chart annotation, not purely by the code. FreshnessGate™ bounds the lifespan of that dependency but does not eliminate it.
• ProfitLock™, when enabled, systematically truncates the right tail of the AdaptiveGuard™ outcome distribution in exchange for a higher hit rate of smaller, faster wins. Neither profile is more correct in the abstract; the choice depends entirely on the operator's own objective.
• All performance characteristics described in this paper are architectural and behavioral. No win rate, risk-reward ratio, or return figure is claimed anywhere in this document, because none has been formally, independently validated at the time of writing.

13. Conclusion

RockBot Nexus formalizes a confluence-based approach to discretionary-style technical trading into the RC-4 Lattice™ and ConfluenceStack™ architecture: seven independently-evaluated signal types, a weighted scoring mechanism determining both whether and how large to trade, and AdaptiveGuard™ applied uniformly regardless of which signal opened a position. Version 1.2 extends this with FreshnessGate™, ProfitLock™, and ContinuityGuard™, each closing a specific, previously-unhandled state identified through direct trace-back against live trading behavior. Every threshold in the system is an explicit, named, user-adjustable value — a deliberate choice for auditability, at the acknowledged cost of the flexibility a fully discretionary trader retains to deviate from stated rules.

Appendix A — Glossary of RockBot Nexus Terminology

RC-4 Lattice™

The four-band rotation-oscillator framework (9-3, 14-3, 40-4, 60-10) underlying every signal in the system.

Quad Rotation Event

All four RC-4 bands crossing into aligned extreme territory simultaneously — the rare, high-conviction condition behind the system's highest-weighted signal.

ConfluenceStack™

The weighted scoring engine summing qualifying signal weights into a single directional conviction score each bar.

ConvictionBoost™

The stacking bonus system for key-level proximity and reversal-candle confirmation.

AdaptiveGuard™

The unified, signal-agnostic exit mechanism: strength-check, breakeven lock, and ATR-based persistent trailing stop.

OverextensionGate™

The two-tier entry lockout pausing new positions when a rotation band has been pinned at an extreme too long.

The Bypass Doctrine

The design rule that Full Lattice Alignment, Price/Stochastic Divergence, and Quad Rotation Divergence are exempt from the universal entry filters and OverextensionGate™, since their own structural premise already encodes an equivalent or stronger condition.

The Apex Signal

Quad Rotation Divergence — the highest-weighted, highest-conviction signal in the library.

The Apex Override

The single maximum-position-size condition: Apex Signal + live trendline touch + retest-bar reversal candle, together.

FreshnessGate™ — New in v1.2

The age-based exclusion filter (default five days) preventing a stale, forgotten drawn trendline from indefinitely triggering Trendline Touch. Measured from the line's older anchor point.

ProfitLock™ — New in v1.2

The optional, off-by-default fixed-dollar-target override. Evaluated before AdaptiveGuard™ and every other exit rule; closes the position the instant unrealized profit reaches the configured target.

ContinuityGuard™ — New in v1.2

The automatic state-recovery mechanism that rebuilds a position's entry-price and entry-type tracking from live position data after a platform restart, keeping the trade under active AdaptiveGuard™ management.

The Ambiguity Resolution Rule — New in v1.2

The AdaptiveGuard™ refinement ensuring that an indeterminate 60-10 reading at the moment of a 9-3 rollover — neither clearly recovered nor clearly still declining — resolves to a flat exit rather than no action.