ES FUTURES //
Decision
Support System
Deterministic algorithms react to price. THOM4S is a source-governed ES futures decision support system that reads the E-mini S&P 500 auction across every institutional horizon. A sealed Master Brain—backed by eleven audited Specialist Modules, automated Sierra Chart capture and a three-model AI review—decodes value, market condition, and rhythm before capital is ever placed at risk.
Python Did Not Fail at Calculation. Deterministic Rules Failed at Context. Rigid IF/THEN indicator stacks collapse across volatility regimes because market observations are relational and path-dependent. A Cumulative Delta divergence in the center of balance is noise; that same relationship at a multi-day Composite Value edge after a failed breakout is high-conviction evidence. We built this ES futures decision support system to solve that exact problem, applying a 70/20/5/5 priority order: Context first, then the Developing Day, then the Order-Flow Microscope and the Forensic Audit—a discipline of attention, not a numeric score. The machine calculates, organizes, and challenges; the human principal retains final risk authority and waits until the market tips its hand.
// The Source-Governed Operational Pipeline — Local-First Authority Chain
11 Specialist Modules //
Auction Market Theory Engine
Flattening the market into a single formula creates combinatorial explosion and curve-fitted fragility. The system applies auction market theory across eleven adversarially audited Specialist Modules governed by a sealed Master Brain: RTH Composite Market Profile (CVA) & Prior-Period VWAP Static Value to map historical agreement; Matched-Anchor Yearly, Quarterly, Monthly, and Weekly VWAPs to detect Other Timeframe (OTF) institutional sponsorship; Market Condition & Rhythm as the governing interpretive engine; 18:00 ET Developing Day (DVA) to test static hypotheses in real time; and the Order-Flow Microscope (Cumulative Delta and Renko Footprint) to inspect effort versus result only after price reaches a location that matters.
“Know where you care before you watch how they trade. A structural level is a question, not an answer. By separating static historical value from developing multi-horizon auctions, our modules determine whether a local intraday move is genuine regime migration or merely inventory pullback inside a dominant higher-timeframe campaign.”
The Second Set of Eyes //
AI Chart Review & Weekly Case Vault
Discretionary flexibility without an audit trail invites hindsight bias; unmonitored AI invites hallucination. We engineered a source-governed "Second Set of Eyes" paired with our Python B1 Journal Engine and projected into our read-only B2 trading risk dashboard. Every 30 minutes, Optic Nerve automatically captures a six-view Sierra Chart ACSIL cockpit pack. Gemini reads the charts, ChatGPT interprets them against the doctrine, and Claude checks that interpretation against the Master Brain—all under strict Observation, Interpretation, Participant-Horizon Inference, and Unknown separation—enforcing no-lookahead temporal cutoffs. Raw AI proposals, Master Brain discrepancy checks, and human corrections are preserved as distinct, immutable layers inside longitudinal Weekly Case Studies.
“The future machine must learn from what was known at the time, not from a story written after the outcome. In our vault, the Week is the primary case, the Day is a state transition tracking what changed since yesterday, and the Trade is an execution event—separating thesis quality from execution outcome with cryptographic precision.”
Capture Integrity //
Evidence You Can Audit
A decision record is only as trustworthy as the evidence beneath it. The Optic Nerve capture pipeline is engineered so every chart image in the journal can be traced, verified and trusted—creating a complete audit trail for trading decisions. Each component below passed independent adversarial audits, mutation testing and hash-bound evidence checks before it was relied on. When integrity cannot be guaranteed, the system stops rather than guesses.
- [CAPTURE]Optic Nerve ProducerCaptures the six-view Sierra Chart cockpit every 30 minutes from the enrolled charts, into its correct time slot. Missed slots are recorded as missed, never backfilled.
- [VERIFY]Optic Nerve ListenerChecks every capture the moment it arrives. Incomplete or malformed evidence is flagged and preserved—never passed through as complete.
- [GUARD]Optic Nerve Guard ServiceA dedicated Windows service admits only the authenticated capture program to write evidence. Every other writer is refused.
- [IDENTITY]EGC Launch ModelEvery component's identity is verified before it runs, so evidence cannot originate from an impostor process.
- [RECORD]Native Result LogbookEvery capture run is recorded—successes, failures and unknowns alike. A missing or interrupted capture can never pass as complete.
- [PROOF]ObserverSecurity behavior is verified on the actual trading host, not only in simulation.
// Fail-Safe Doctrine
Stop, never guess. If evidence integrity cannot be guaranteed—a clock fault, an unverified writer, an unfinished capture—Optic Nerve stops safely and alerts the operator instead of producing a record it cannot stand behind.
Capture runs beside the trading platform, never through it. A capture fault can halt evidence collection; it never touches order entry or risk.
“An unreadable chart is a known gap. A misattributed chart is a false record. We engineered to tolerate the first and to make the second impossible to pass unnoticed.”
SPEED IS A COMMODITY.
CONTEXT IS
THE EDGE.
// The Thom4s Doctrine — Riding Institutional Coattails
The quantitative industry has spent billions shaving microseconds off wire latency, only to deploy brittle signal bots that cannot tell whether a local sell-off is a genuine breakdown or a routine pullback into a dominant Weekly or Monthly institutional imbalance.
In the E-mini S&P 500, Real Money pensions, sovereign hedgers, asset managers, macro funds, and HFTs transact simultaneously across competing holding periods. Systems that only read local price action stand in front of higher-timeframe institutional flow—and get run over.
THOM4S was architected to solve this multi-horizon problem for serious capital. This is not a retail SaaS subscription—it is a bespoke decision support system where the auction doctrine and the C++/Python engineering were built by the same hand.
- Brittle IF/THEN indicator gates that break across volatility regimes
- Blind to multi-horizon conflict across Weekly, Monthly & Quarterly value
- Confuses raw price motion with structural value acceptance
- Black-box signals that erase failed hypotheses with hindsight
- Vulnerable to look-ahead contamination and ungrounded AI hallucinations
- Applies the 70/20/5/5 priority order—context before the microscope
- Separates Observation, Interpretation, OTF Inference & Unknowns
- Optic Nerve & Guard Service lock 6-view cockpit evidence in C++
- Multi-Model AI Review (Gemini read → ChatGPT interpretation → Claude Master Brain check)
- Human-Approved Python B1 Vault preserves immutable Weekly Cases
ES Futures Decision Support System Specifications
Engineered to institutional audit tolerances. Every layer separates technical verification, analytical doctrine fidelity, and human risk authority. This is not retail chartware—it is a cryptographically pinned, source-governed decision and evidence engine.
- [DOCTRINE]Master Brain v1.0 & 11 Specialist ModulesSealed multi-horizon ES auction constitution reconstructed across 11 adversarially audited domain modules for the CME Group E-mini S&P 500.
- [CAPTURE]Optic Nerve (ACSIL C++ / Python + Guard Service)Automated six-view Sierra Chart cockpit capture every 30 minutes, ON_TIMING_01 slot discipline, and Guard-authenticated namespace lock ownership.
- [VAULT]Python B1 Journal Engine (Local-First SQLite)Immutable versioned state transitions, computed correction manifests, bounded as-of temporal reads, and human-only approval gates.
- [AUDIT]SHA-256 Cryptographic ProvenanceEnd-to-end byte, line-feed, and SHA-256 receipt verification across source ledgers, every captured evidence pack, and test harnesses.
- [PROJECTION]B2 Deterministic Flight DeckRead-only offline HTML/JSON projection preserving candidate vs. approved records and literal Unknowns with zero active attack surface.
- [ORCHESTRATION]Multi-Model AI Review LoopGemini chart reading, independent ChatGPT interpretation and Claude verification against the Master Brain, backed by ASan/UBSan-hardened C++ and deterministic Python test suites.
“The lesson from failed automation is not to stop using code—it is to stop asking code to be the entire trader. I built this system so the machine enforces uncompromising contextual and forensic discipline, leaving human judgment free to strike only when the auction tips its hand.”