Framework

Capital Allocation Framework

How qualified decisions are prioritized for resource deployment under constraints.

v0.2·working·Updated July 2026

Overview

Qualification answers a binary question: has this idea earned the right to execute? Allocation answers a fundamentally different question: given everything that has earned the right, where should limited resources go? These are distinct architectural concerns that require separate infrastructure.

The Capital Allocation Framework makes visible the system that sits between qualification and execution. It governs how qualified opportunities are ranked, sized, diversified, and monitored as a portfolio — not as isolated decisions. The unit of analysis shifts from the individual strategy to the collection of strategies competing for the same pool of capital.

This framework is constraint-driven. Capital is finite. Attention is finite. Correlation is real. The allocation layer exists because qualification alone cannot produce a coherent portfolio — it can only produce a set of individually worthy candidates. Allocation imposes portfolio-level coherence on that set.

Design Principles

These are governing constraints, not guidelines. Every architectural decision in the framework must be traceable to one of these principles.

01

Qualification Is Not Allocation

A decision that has earned the right to execute has not yet earned a specific share of resources.

Qualification establishes eligibility. Allocation establishes priority. Conflating them produces portfolios where every qualified strategy receives equal capital regardless of fit, correlation, or environmental context. The separation forces explicit reasoning about resource distribution.

02

Portfolio-Level Thinking

Every allocation decision must be evaluated in the context of the existing portfolio, not in isolation.

A strategy that looks excellent in isolation may be redundant, correlated, or regime-mismatched when viewed alongside what is already deployed. Allocation intelligence requires awareness of the whole, not just the candidate.

03

Regime-Aware Sizing

The same qualified strategy deserves different resource allocation under different environmental conditions.

A trend-following strategy qualified during a trending regime should receive less capital allocation when the environment shifts to choppy conditions. Regime awareness prevents static allocation from becoming stale allocation.

04

Isolation Preserves Attribution

Each allocation must operate with independent capital, independent positions, and independent risk limits.

Shared capital pools make outcome attribution impossible. When strategies share resources, you cannot distinguish whether a result came from the strategy or from portfolio interaction effects. Isolation is not just a safety measure — it is an epistemic requirement.

System Map

Five layers from qualification output to active capital management. Qualified decisions flow downward through increasingly concrete resource commitment. Performance evidence flows upward, informing reallocation and rebalancing.

1
Qualification Assessment2 components
Evidence IntakeFit Scoring
2
Allocation Intelligence3 components
Composite RankerCorrelation AnalyzerDiversification Constraints
3
Exposure Management2 components
Position SizingLeverage Governor
4
Risk Guard2 components
Drawdown MonitorExposure Cap Enforcer
5
Rebalancing2 components
Regime Transition ResponseCapital Reallocation
Intelligence flows down|Evidence flows up

Layers & Components

Each layer has distinct responsibilities, inputs, outputs, and ownership. No component spans multiple layers.

1

Qualification Assessment

Receives qualified decisions and structures their evidence for allocation consideration. The boundary between earning the right to execute and competing for resources.

Evidence Intake

Transforms qualification output into allocation-relevant metrics

In: Qualification status, Evidence trail, Regime proofOut: Allocation-ready candidate profilesOwns: Mapping from qualification evidence to allocation dimensions

Fit Scoring

Evaluates how well a qualified strategy fits the current environment and portfolio

In: Strategy behavioral profile, Current regime, Existing portfolio compositionOut: Fit scores with regime-conditional breakdownsOwns: Fit score computation, materialized weekly
2

Allocation Intelligence

Ranks and prioritizes qualified candidates based on evidence strength, environmental fit, and portfolio-level considerations.

Composite Ranker

Produces advisory rankings across qualified candidates

In: Fit scores, Paper trading equity curves, Correlation matrixOut: Ranked candidate list with recommendation tiers (increase/hold/avoid)Owns: Ranking algorithm, Sharpe-based scoring, hard floors (negative Sharpe caps at hold)

Correlation Analyzer

Detects redundancy and concentration risk across candidates

In: Equity curve histories, Strategy parametersOut: Correlation pairs, Portfolio diversity scoreOwns: Pearson correlation computation on equity curves, high-correlation warnings

Diversification Constraints

Enforces structural diversity across the allocated portfolio

In: Candidate profiles, Current portfolio compositionOut: Diversity-adjusted candidate eligibilityOwns: Per-asset limits, per-family caps, per-timeframe caps, regime coverage targets
3

Exposure Management

Governs the actual resource commitment — position sizing, leverage limits, and aggregate exposure.

Position Sizing

Determines capital allocation per strategy within portfolio constraints

In: Allocation rank, Risk parameters, Available capitalOut: Capital allocation per strategyOwns: Max 20% portfolio per position, max 50% account exposure invariants

Leverage Governor

Constrains leverage within safety limits across the portfolio

In: Strategy spec leverage, Product mode, Aggregate exposureOut: Effective leverage per strategyOwns: 2x default leverage, region-aware leverage caps (US spot: 1x max)
4

Risk Guard

Circuit breakers and hard limits that override allocation decisions when safety thresholds are breached.

Drawdown Monitor

Detects and responds to per-strategy and portfolio-level drawdowns

In: Position state, Equity curves, HWMOut: Kill signals, Pause directivesOwns: Max drawdown thresholds, per-strategy circuit breakers

Exposure Cap Enforcer

Prevents aggregate exposure from exceeding system safety limits

In: All active positions, Margin utilizationOut: Rejection of new allocations when caps are breachedOwns: System-wide bot capacity, per-tier subscriber caps
5

Rebalancing

Responds to environmental and performance changes by adjusting the active portfolio over time.

Regime Transition Response

Adjusts allocation priorities when environmental conditions change

In: Regime transitions, Per-strategy transition survivabilityOut: Trailing stop tightening, Library standby transitionsOwns: Toxic transition table, per-strategy survivability multipliers

Capital Reallocation

Manages capital flow between strategies via top-up and reduce operations

In: Performance data, User directives, Available balanceOut: Capital flow events (deposit/withdrawal)Owns: Top-up and reduce gates, min $5 thresholds, no-open-position guard

States & Transitions

A decision moves through these states as it accumulates evidence. Each transition is gated — there are no free promotions.

Qualified
Eligible
Allocated
Active
Reduced
Removed

Qualified

Has earned the right to execute but has not yet been allocated resources.

Entry: Passes qualification gates with regime proof

Exits: Selected for allocation -> Eligible · Disqualified -> Removed

Eligible

Passes portfolio-level checks (diversity, correlation, fit). Ready for capital commitment.

Entry: Allocation intelligence ranks favorably, diversification constraints met

Exits: User commits capital -> Allocated · Regime shift reduces fit -> Qualified

Allocated

Capital committed and deployed. Actively trading with real resources.

Entry: Exchange connection verified, capital deposited, fee status clear

Exits: Performing within bounds -> Active · Drawdown breach -> Reduced

Active

Running in steady state. Subject to continuous monitoring and rebalancing.

Entry: Allocation stable, no circuit breaker triggers

Exits: Capital adjusted -> Reduced · Regime transition -> Paused · Kill switch -> Removed

Reduced

Capital partially withdrawn due to performance, rebalancing, or user directive.

Entry: Capital reduce executed, remaining allocation >= minimum

Exits: Capital topped up -> Active · Fully withdrawn -> Removed

Removed

Allocation terminated. Capital returned. Strategy may remain qualified for future reallocation.

Entry: Kill switch, full capital withdrawal, or disqualification

Exits: Re-allocated with new capital -> Allocated

FromToTriggerGuard
QualifiedEligibleAllocation intelligence evaluationFit score above threshold, diversification constraints met, no high correlation with existing portfolio
EligibleAllocatedUser promotion with capitalExchange connection, sufficient balance, fee status clear, bot capacity available
AllocatedActiveFirst tick evaluationCapital deposited, strategy spec loaded
ActiveReducedCapital reduce or rebalancingNo open position, remaining capital >= $5
ActiveRemovedKill switch or disqualificationOpen positions closed, capital returned
ReducedActiveCapital top-upNo open position, sufficient available balance

Boundary Rules

Boundaries define what crosses between layers and what does not. Every boundary has an explicit failure mode.

Qualification -> Allocation

Crosses

Qualification status, evidence trail, regime proof, behavioral profile, fit scores

Does Not Cross

Raw backtest trades, paper account internals, qualification computation details

Failure Mode

Unqualified strategy cannot enter allocation pipeline. No backdoor from generation to capital.

Allocation Intelligence -> Exposure Management

Crosses

Advisory ranking (increase/hold/avoid), candidate profile, portfolio diversity assessment

Does Not Cross

Binding position sizes or leverage decisions. Intelligence advises; exposure enforces.

Failure Mode

Missing intelligence -> conservative defaults. Allocation intelligence is advisory, never autonomous.

Exposure Management -> Risk Guard

Crosses

Proposed position sizing, aggregate exposure state, margin utilization

Does Not Cross

Risk guard does not see strategy logic, qualification evidence, or allocation rationale. It only sees exposure numbers.

Failure Mode

Risk guard breach -> immediate rejection or kill. No negotiation. Hard boundary.

Risk Guard -> Rebalancing

Crosses

Kill signals, pause directives, drawdown alerts

Does Not Cross

Risk guard does not suggest rebalancing targets. It only constrains.

Failure Mode

Rebalancing without risk guard input -> potential safety violation. Risk guard has veto power.

Operational Semantics

How this framework operates in practice — cadence, consumers, and staleness rules.

DimensionValueRationale
Fit score materializationWeekly (Monday 4am UTC)Fit scores are computationally expensive and change slowly. Weekly materialization balances freshness against cost.
Allocation intelligence evaluationEvery 6 hours (batch)Runs alongside qualification assessment. Computes Sharpe, correlation, diversity scores for all active strategies.
Exposure monitoringContinuous (per-tick)Every paper and live tick evaluates position state against exposure limits. 1-minute cadence.
Risk guard enforcementContinuous (per-tick)Circuit breakers evaluate on every tick. No batch delay for safety-critical checks.
Rebalancing cadenceEvent-driven (regime transitions) + daily (lifecycle evaluation)Regime transitions trigger immediate response. Daily lifecycle sweep handles gradualnchanges.
Capital flow operationsOn-demand (user-initiated)Top-up and reduce are user actions, not automated. System advises; user decides.
Correlation computationEvery 6 hours (M5 model)Portfolio correlation matrix recomputed in eval batch. Feeds diversity score and high-correlation warnings.

Knowledge Lineage

Related Research

See this framework implemented in Orqis

Research — Warren Labs | Orqis