The methodology behind governed capital deployment
A full account of how the MSIIA Institutional Edition converts raw market structure into disciplined, regime-aware exposure — signal by signal, gate by gate, under non-discretionary oversight.
Methodological Posture
Structure decides before conviction is allowed to matter
The system is organized as a hierarchy of accountability. Market regime is classified first; signals are heard second; defense and portfolio governance decide, last, whether opportunity is permitted to become capital.
The MSIIA Institutional Edition is a multi-sector, multi-symbol allocation engine built on MetaTrader 5. It does not treat a trade signal as an instruction. A signal is a candidate that must survive regime context, a committee vote, a stack of qualification gates, exposure governance, and a live defense layer before any capital is released — and every one of those steps can veto the others.
This ordering is deliberate. It prevents performance from being explained after the fact and forces the system to behave the same way under stress as it does in calm markets. Where deeper documentation is shared after review, it continues this posture rather than departing from it.
Reading Guide
Parameter values on this page are configuration defaults from the Institutional Edition build. They describe how the system is governed — not a promise of outcome. Past performance is not indicative of future results.
Anchor 01
Regime first
Prevailing market structure is classified before any signal is permitted to influence exposure. Context sets the ceiling.
Anchor 02
Verification & defense
Confirmation and damage control are decision layers in their own right — not monitoring added after the trade is on.
Anchor 03
Portfolio governance
Aggregate concentration, correlation and drawdown decide how a qualified opportunity is finally sized and released.
Signal To Release
Eight governed stages, one direction of authority
The flow is unidirectional. Each stage must clear its own constraints before it is allowed to pass authority to the next. Nothing downstream can override a veto from upstream.
Market Ingest
Multi-sector symbol universe, macro state, microstructure and an event calendar are normalized against a market reference (US500).
Regime Classification · IRDM
The regime engine scores each symbol 0–100 and tags the operating context. High scores lock the symbol out before any signal is heard.
Neural Committee Vote
Independent AI committees score conviction. No single signal family controls the outcome; the composite must clear the confidence gate.
Signal Filtration
Structural, volatility, breadth, cost and predictability gates run in series. A composite score below threshold rejects the candidate outright.
Exposure Governor
Regime-conditional sizing under the active capital tier. Capital is released, clipped or deferred against portfolio balance and equity ceilings.
Execution & Lifecycle
Order flow is governed by cadence discipline, smart breakeven, trailing/TP optimization and booster / straggler lifecycle logic.
Live Defense
Symbol damage scoring, collapse guard and crash-regime de-risking run continuously and can force-close or quarantine exposure.
Governed Release & Audit
Aggregate constraints are verified and every decision is time-stamped into an audit-ready trail for institutional due diligence.
Regime Detection Model
IRDM classifies the weather before it reads the map
The Institutional Regime Detection Model scores structural conditions per symbol and governs a strict state machine. A symbol in Hard Block cannot trade — regardless of how attractive its signal looks.
| Regime Parameter | Setting |
|---|---|
| Hard-block threshold | Score ≥ 70 |
| Soft-risk threshold | Score ≥ 50 |
| Pre-block confirmation | 10 bars ≥ soft |
| Hard-block confirmation | 5 bars in pre-block |
| Minimum lock duration | 90 calendar days |
| Recovery confirmation | 15 bars < soft |
| Trend filter | SMA 50 / SMA 200 |
| Volatility window | ATR 14 / ATR 60 |
| Structure lookback | 252-bar high · 60-bar DD |
| Replacement budget | 100% · max 3 / day |
Regime scoring blends short and long volatility (ATR 14 vs ATR 60), fast/slow trend alignment (SMA 50 vs SMA 200), momentum via RSI 14, distance from the 252-bar high and a rolling 60-bar drawdown. The result is a single 0–100 stress score per symbol.
To prevent thrashing, transitions are anti-flicker gated. A symbol must hold above the soft threshold for ten bars to enter Pre-Block, then five confirming bars to lock into Hard Block — where it stays for a minimum of ninety days. Recovery requires fifteen bars back below soft before authority is restored.
When a live position's symbol enters Hard Block, IRDM can close and replace it in a structurally healthy alternative, carrying up to 100% of the closed P&L value into the replacement and logging every substitution.
AI Core & Neural Committee
Conviction is a committee decision, never a single voice
Independent AI committees score every candidate. A learning optimizer adapts their weights over time, and a strict confidence gate blocks any trade whose composite score falls short.
ACommittee governance
Federated intelligence
Advanced AI committee engines evaluate each candidate in parallel under full institutional-mode governance logic. Authority is distributed so no single model dominates.
BML optimizer
Adaptive weighting
An online optimizer updates committee weights each cycle with a bounded learning rate and a performance-memory decay, persisting learned weights across runs.
CConfidence gate
Strict signal gating
Candidates scoring below the minimum AI confidence threshold are blocked before they ever reach the qualification stack. Gating is enforced, not advisory.
Why a committee
A single predictive model fails silently when its regime changes. Distributing the decision across independent committees — and continuously reweighting them by realized performance — means the system degrades gracefully rather than catastrophically when one signal family stops working.
Signal Filtration Framework
A candidate must clear every gate, then the composite
Qualification runs as a series of hard and soft gates. Hard gates reject outright; soft gates penalize the score. A final composite and a minimum count of positive components decide approval.
Every gate contributes to a normalized composite. Hard gates can reject a candidate on their own; soft gates apply calibrated penalty multipliers so that a symbol under mild stress is not banned, only made harder to approve.
Final approval requires the composite to clear its threshold and a minimum number of individual components to be positive, with a separate meta-score confirmation on top. Approved names are then capped at the portfolio level.
| Approval Rule | Threshold |
|---|---|
| Final composite score | ≥ 0.72 |
| Positive components required | ≥ 5 |
| Meta-score confirmation | Required |
| Approved symbol cap | 10 symbols |
| Cross-sector cap | 6 sectors |
| Adaptive reweighting | LR 0.03 · decay 0.95 |
Capital Allocation & Sizing
Deployment tiers set the ceiling, regime sets the release
Capital is deployed through named allocation tiers with explicit equity ceilings. The active baseline is a balanced growth engine; adaptive budgeting redistributes within the ceiling as conditions change.
Tier 01
Capital Preservation
Defensive deployment posture. Tightest per-position ceiling, prioritizing capital protection over participation.
Tier 03 · Active baseline
Growth Allocation Engine
Balanced growth engine — the standard deployment baseline. Participates through a disciplined per-position equity ceiling.
Tier 05
Expansion Mandate
Highest-participation tier, reserved for confirmed constructive regimes and available only under review.
Sizing
Lot floor
Minimum permissible lot.
Sizing
Lot ceiling
Maximum permissible lot.
Sizing
Step
Granularity between sizes.
Sizing
Concurrency
Approved-symbol cap.
Tier labels above illustrate the deployment ladder; the Institutional Edition ships with the Growth Allocation Engine as its documented baseline. Tier changes require formal access review and are not self-served.
Drawdown & Crash-Regime Defense
Loss limits are non-discretionary and staged
Drawdown governance escalates from individual-stock liquidation to global portfolio cut-off to emergency de-risking. Crash-regime defense adds a breadth-aware layer that can hard-block all new entries.
Portfolio protection is layered on top of per-symbol control. A crash-regime defense watches the whole book: floating drawdown, market breadth and drawdown acceleration together decide whether the system moves from Warning to full Crash posture.
In a confirmed crash the system hard-blocks new entries, disables recovery, replacement, booster and grid logic, and can force-close the weakest positions by live P&L.
| Crash Trigger | Warning | Crash |
|---|---|---|
| Portfolio floating DD | 3.0% | 5.0% |
| Force de-risk DD | — | 6.5% |
| Market breadth block | 55% | 70% |
| DD acceleration | 1.5% | 2.5% |
| ATR spike / gap-down | 2.5% / 2.0% | 4.0% / 4.0% |
| Force-close weakest | — | 2 positions |
Symbol Defense & Damage Scoring
Every symbol carries a live damage score
A weighted damage matrix scores each symbol continuously. As the score rises, the symbol escalates through six defensive states — each tightening entries, throttling capital, and eventually forcing exit.
Nine weighted damage inputs
Each input is exposure-weighted and normalized into a single score, so a large position under stress escalates faster than a token one. Live anomaly detection widens the score on abnormal range or spread events.
A correlation-cluster defense watches groups of related symbols together — if a cluster's aggregate stress crosses threshold, the whole cluster is defended, not just the worst name. State transitions are logged for audit, and only meaningful events are recorded to keep the trail clean.
Collapse Guard & Reallocation
When the market breaks, capital moves to strength
Collapse Guard detects systemic events across the tracked universe and reallocates adaptively — locking weak names to a minimum weight while boosting structurally strong assets, then waiting for a confirmed recovery.
Windows
Multi-horizon
Detection across three structural lookbacks.
Triggers
Stress thresholds
Drawdown, ATR spike and gap-down confirmation.
Breadth
Systemic call
Market-wide collapse when breadth crosses threshold.
Reallocate
To strength
Lock weak names; boost structurally strong assets only.
Recovery discipline
Reallocation locks for a defined period and only unwinds on a confirmed structural recovery — fifteen consecutive days back above the 200-day moving average. The system re-risks slowly and on evidence, never on a single green day.
Booster · Straggler · Grid Recovery
Positions are managed, not just opened and closed
Once live, every position runs a lifecycle. Winners can be amplified, laggards are triaged for recovery or exit, and a bounded grid engine handles staged re-engagement under strict exposure caps.
Engine · Booster
Momentum amplification
Confirmed winners are scaled while conviction holds, with an automatic exit when the AI score decays beyond tolerance.
Engine · Straggler
Underperformer recovery
Laggards below the confidence floor are prioritized for measured recovery or removed before they compound into damage.
Engine · Grid Recovery
Bounded escalation
Staged re-engagement runs under lifecycle exposure limits and position caps, and is disabled entirely for damaged symbols and crash regimes.
Volatility Adaptation & Telemetry
The system tightens itself as volatility rises
Trailing stops and take-profit targets adapt to live volatility, recovery accelerates for laggards, and institutional telemetry records the whole decision surface for diagnostics and audit.
Detect
Vol threshold
ATR-to-price ratio flags high-volatility regimes.
Trail
Tighten
Trailing stops compress under stress.
Target
Take-profit
TP targets pull in when volatility spikes.
Recover
Lag boost
Recovery rate lifts for lagging positions.
Telemetry & diagnostics
Smart-breakeven logic, take-profit optimization and volatility adaptation all report into an institutional telemetry layer. A runtime diagnostics and system-event console records state transitions, so the live decision surface can be reconstructed for review.
Governance, Compliance & Audit
Aligned to institutional standards, built to be audited
Every decision, signal and state transition is time-stamped and logged. The system is designed for institutional due diligence — informationally aligned to the frameworks allocators are held to.
Traceability
Full audit trail
Transaction and model traceability across the decision pathway. Every signal, state change and enforcement action is reconstructable.
Enforcement
Non-discretionary
Risk constraints execute without discretion. Human review is required only for access-tier changes — not for individual trades.
Access
Gated & account-bound
Licensed on a fixed five-year, non-transferable term with account-bound key validation. Parameters are shared only with reviewed allocators.
Infrastructure & Deployment
The operating envelope at a glance
Complete Subsystem Index
Thirty governed modules, end to end
Continue The Review
Methodology is one layer of the dossier
See how these subsystems assemble into the full stack, browse the governed systems, and review realized behavior across regimes.