CONFIDENTIAL — INTERNAL USE ONLY — KALAYA CAPITAL — NOT FOR DISTRIBUTION
KALAYA CAPITAL

Operating Memorandum

Document ReferenceKC-OM-v0.4
Version0.4  BASELINE
SupersedesKC-OM-v0.3 (30 May 2026) — view v0.3  |  v0.2  |  v0.1
ClassificationINTERNAL  Not for distribution
Date30 May 2026
Prepared ByClinton Ratanatray, Founder — Kalaya Capital
In Consultation WithMatt Breakwell, Partner — Kalaya Capital
PurposeTo document the operating philosophy, research methodology, portfolio construction framework and governance model of Kalaya Capital. This document is the Kalaya Capital Operating System.
v0.4 ChangesFirst baseline version. LLP Governance Development added (§1.2). Primary asset statement strengthened (§2). Institutional knowledge framework added (§11, §16, §26). Regime framework updated with Dalio attribution and research objective language (§18). Current phase and unproven capabilities added (§31, §32). July LLP Session rewritten as first LLP engagement (§33). Path to October added. Editorial pass throughout.
Part I
Identity & Philosophy

1.What is Kalaya Capital

Kalaya Capital is a systematic, directional, long/short trading operation. It identifies recurring behavioural patterns in financial markets, builds mechanical rules to exploit them, validates those rules rigorously, and deploys capital systematically.

What we are

  • Systematic — all entry, exit, and sizing decisions are rule-based
  • Long/short — no structural bias toward either direction
  • Multi-market — strategies operate across FX and selected asset classes
  • Portfolio-driven — individual strategy performance is secondary to portfolio robustness
  • Process-first — the research and validation process is the core asset

What we are not

  • Discretionary traders — no trades taken on human judgement at execution time
  • Value investors — no long-term fundamental positions
  • Arbitrage traders — no cross-instrument pricing inefficiency exploitation
  • Macro forecasters — no directional views on economic policy or geopolitics
  • Prediction machines — we exploit recurring behaviour; we do not predict direction

1.1 Corporate Structure

Kalaya operates through two entities with distinct responsibilities:

EntityPrimary Activities
Kalaya Capital Limited Research, strategy development, code development, intellectual property ownership
Kalaya Capital LLP Capital deployment, portfolio construction, capital allocation, member capital management

Research and IP are owned by the Limited company. Capital management is conducted through the LLP. These are distinct activities and are not interchangeable. This document applies across both entities.

1.2 LLP Governance Development NEW

The Kalaya Capital LLP governance framework is currently under development. The following components are being built in parallel with the research and execution infrastructure so that the LLP can operate in a structured, repeatable, and transparent manner from the point of launch:

These components are being developed progressively. Their current status is reflected in §31 (Current Phase).

2.Why Kalaya Exists

Kalaya exists to build and maintain a repeatable, evidence-driven process for discovering, validating, and allocating capital to systematic trading strategies.

The primary asset is the operating system, not any strategy.

Kalaya's primary asset is not any individual strategy, model, report, or codebase. Its primary asset is the integrated operating system that continuously produces, validates, deploys, monitors, and manages systematic investment strategies — and the accumulated institutional knowledge embedded within it.

2.1 The Problem We Are Solving

Most systematic trading operations fail not because their strategies are wrong, but because their process is wrong: strategies are over-fitted to historical data, deployed without rigorous validation, and retired too late — or too early — without an objective framework. Kalaya is built to avoid those failures. The primary output is not any individual strategy. It is a process that continuously produces, tests, and manages strategies.

2.2 The Objective

The objective is not to predict markets. It is robust portfolio construction — assembling independently evidenced strategies across markets, behaviours, and timeframes such that the portfolio maintains positive expectancy across a wide range of conditions, even when individual strategies fail.

2.3 Strategies Are Temporary. The Process Is Permanent.

We expect strategies to degrade, fail, or be superseded over time. The durability of this operation depends entirely on the quality of the process — not on the durability of any individual strategy. A business that depends on a single strategy surviving indefinitely is not a business. A business with a robust process for continuously producing and replacing strategies is.

2.4 What Success Looks Like

3.Core Principles

These principles govern how Kalaya thinks, researches, and makes decisions. They apply to every component of the business — not just strategy development.

Signal Over Noise
Simplicity over complexity. Clarity over activity. More rules, more parameters, more filters create more noise — not more edge. The clearest signal is usually the simplest. Any increase in complexity requires justification, not encouragement.
Process Over Prediction
The objective is a repeatable, disciplined process — not accurate forecasting. Markets are not predictable with precision. A rigorous process applied consistently across many opportunities produces durable results where individual predictions do not.
Evidence Over Opinion
Every allocation decision must be backed by documented, reproducible evidence. An idea that has not been tested is a hypothesis, not an edge. Capital follows evidence. It does not follow conviction.
Simplicity as a Constraint
Every strategy must be expressible in a small number of deterministic rules, executable without human interpretation, and fully testable on historical data. A strategy that requires a chart to understand is not ready.
Kaizen — Continuous Improvement
The process is never complete. Research, validation, execution, governance, reporting, and business processes are all continuously reviewed and improved. Every cycle is an opportunity to identify and implement a better approach. Improvements are captured, documented, and embedded — not lost.

4.Investment Philosophy

4.1 Core Belief

Markets change. Regimes shift. No single strategy produces durable edge across all market conditions in perpetuity. The objective is a repeatable, disciplined process for identifying, validating, deploying, and retiring edges as market behaviour evolves.

4.2 First-Principles Approach

Strategy development begins with a fundamental question: what market behaviour are we trying to exploit? This shifts the work away from curve-fitting and indicator layering toward testing whether well-documented behaviours — momentum continuation and mean reversion — produce statistically meaningful edge when expressed as simple, mechanical rules.

4.3 Portfolio Over Strategy

Portfolio construction is more important than any individual strategy. A single strategy may fail in a given regime. A well-constructed portfolio of uncorrelated strategies is expected to maintain positive expectancy across a wider range of conditions. Strategies are assessed not only on their individual metrics but on their portfolio contribution — their regime behaviour, failure conditions, and correlation to existing deployed strategies.

5.People & Sustainability

Kalaya is built to be durable — for the markets, for the process, and for the people involved in building it.

5.1 Long-Term Alignment

The business is structured to generate long-term returns that are meaningful to founders, partners, and future participants. Short-term performance pressure creates exactly the wrong incentives in a systematic trading business: it encourages over-fitting, premature deployment, and reluctance to retire failing strategies. Kalaya's model is deliberately long-cycle. Research takes time. Validation takes time. Genuine edge accumulates over time.

5.2 Equitable Participation

The benefits of Kalaya's research, capital growth, and business development should flow equitably to those contributing to them. This applies to founders, partners, and any future contributors who bring genuine value to the research, operations, or governance of the business.

5.3 Sustainability

A sustainable business is one that can be operated at high quality over the long term. Kalaya prioritises a research pace that maintains quality over quantity; an operational model that does not require continuous human intervention; documentation and systems that can be handed off or scaled without loss of quality; and a culture of honesty about what is working and what is not.

Part II
Research Framework

6.Alpha Extraction Framework

6.1 Primary Inputs

Kalaya strategies operate primarily using price, time, and volume — where available. The relative importance of each input varies by strategy. External data — economic releases, central bank communications, fundamental valuations — are not used as signal inputs. They may inform portfolio-level decisions, but they do not enter the signal generation process.

6.2 Strategy Characteristics

CharacteristicDescription
SystematicAll signals, entries, exits, and sizing are generated by code. No human judgement at the execution level.
DirectionalStrategies express a view on price direction — long or short. No hedging within a strategy.
Long / ShortBoth directions with equal mechanical rules. No structural bias.
Multi-marketStrategies are instrument-agnostic. A validated strategy is tested across the full universe, not tuned to a single market.

6.3 The Objective

To identify recurring changes in market behaviour — moments where the probability of a directional move is measurably better than random — and exploit them systematically, across multiple markets, with defined risk, at scale.

7.Market Behaviour Framework

Kalaya organises all research around four behavioural categories describing what the market is doing when a signal fires — not what the strategy is called. Multiple strategies can operate within the same category and compete for capital based on evidence.

CategoryBehaviour Being ExploitedExpected Failure Regime
Momentum / BreakoutDirectional persistence after a structural level break, driven by order flow imbalanceChoppy, low-volatility ranging; high-frequency false breaks
Mean ReversionPrice overextends from a statistical anchor and returns to itStrong trending markets; sustained momentum moves
Trend PullbackContinuation entry at a retracement within an established directional trendRanging markets; deep pullbacks that become full reversals
Reversal / ExhaustionTrend termination at structural extremes; directional flip after exhaustion of order flowStrong trends with no exhaustion; low-conviction reversal signals

Strategies compete for capital within their category based on cross-market evidence, regime fit, and correlation to existing portfolio positions. A category with no validated strategy receives no allocation.

8.Research Philosophy

8.1 Ideas Are Tested, Not Believed

A strategy idea is a hypothesis, not a conviction. The research process exists to test the hypothesis. All findings — positive and negative — are valid outputs. A rejected strategy produces as much value as an approved one: it eliminates an idea that should not receive capital and adds to the institutional knowledge base.

8.2 Strategy Requirements

Before entering the research pipeline, a strategy must be codified as unambiguous, deterministic rules; classified into a behavioural category with a declared expected failure regime; testable on historical data without human interpretation; and applicable to the full research universe without instrument-specific modifications.

8.3 Questions the Pipeline Must Answer

  1. Does the strategy produce positive net expectancy after costs?
  2. Is the edge consistent across multiple instruments?
  3. Is the edge stable across different time segments?
  4. Does the edge survive 1,000 Monte Carlo resampling simulations at p95?
  5. What are the conditions under which the edge is expected to disappear?

8.4 Portfolio Robustness Over Individual Performance

A strategy with high Sharpe on one instrument is not automatically valuable. A strategy with modest but consistent positive expectancy across eight instruments, with failure conditions that differ from the existing portfolio, is more valuable. Individual performance metrics are diagnostic. They do not determine approval.

9.Research Before Capital

Capital follows evidence.

No strategy receives allocation until it has completed the full validation process and passed all defined approval criteria.

Capital deployed to a strategy that has not passed validation is speculation, not investment. Every pound of capital deployed must be backed by a documented research and validation record. A strategy that looks good on a chart is not a validated strategy.

10.AI Assisted Research

AI tools are used across the Kalaya research and operational workflow in clearly bounded roles.

AI Assists With

  • Generating and refining strategy ideas
  • Writing, reviewing, and testing pipeline code
  • Drafting and refining research reports and operational documents
  • Summarising findings and generating structured outputs
  • Organising and maintaining research records and knowledge

AI Does Not

  • Allocate capital
  • Approve strategies for deployment
  • Assume accountability for any decision
  • Replace the validation process
  • Replace human governance and oversight
Responsibility remains human.

AI accelerates the process. Every deployment decision, risk limit, and governance review is a human responsibility. The accountability structure does not change because AI was involved in producing an output.

11.One Source of Truth & Institutional Knowledge UPDATED

Every research run, validation result, deployment decision, governance discussion and lesson learned contributes to the institutional knowledge base of Kalaya Capital.

This accumulated knowledge is a core asset. It must be systematically captured, version-controlled, and accessible — not fragmented across conversations, emails, or undocumented side processes.

11.1 Why This Matters

A systematic trading operation generates large volumes of research data, validation records, code changes, decision records, and operational logs. Without a single authoritative source, decisions get made on incomplete information. Strategies get re-researched unnecessarily. Hard-won lessons are lost. The operating model drifts from documented reality. Institutional knowledge evaporates when individuals step away.

11.2 What Forms the Knowledge Base

11.3 Principles

11.4 Architecture

The specific technical architecture for knowledge management — tooling, integration, search — remains under development. The principle is fixed: one source, multiple outputs, full version control, zero information lost to undocumented processes.

Part III
Process & Lifecycle

12.Kalaya Operating Model

The Kalaya operating model is a linear, sequential process. Capital does not advance until each stage is complete. No stage can be skipped.

1
Idea
A market behaviour hypothesis is formed, codified as deterministic rules, assigned to a behavioural category, and registered before any code is written.
2
Research
Signal logic runs through the full pipeline across all target instruments using IBKR data, with costs applied. Outputs: trades CSV, equity curve, monthly returns, approval metrics.
3
Validation
Walk-forward stability, parameter sensitivity, Monte Carlo stress testing. A strategy that only works at specific parameters is rejected as fragile.
4
Portfolio Candidate
Approved strategies enter portfolio assessment. Capital competition determines whether the strategy adds genuine diversification benefit over existing positions.
5
Capital Competition
Strategies compete for allocation based on evidence strength, regime fit, and portfolio correlation. A strong strategy in a well-covered category may receive less than a weaker strategy covering an uncorrelated regime.
6
Execution
Capital is deployed to approved strategies within the Risk Layer's position sizing and diversification framework. Paper execution precedes live deployment.
7
Monitoring & Governance
Live behaviour is compared against expected profile. Strategies are maintained, reduced, retired, or reintroduced based on observed evidence. The research pipeline runs in parallel to maintain a supply of candidates.

13.Research Pipeline

The Python-native research pipeline implements stages 1–4 of the operating model. Each stage produces outputs consumed by the next. All stages run without human intervention once triggered. (Source: quant-pipeline/)

DataIBKR or CSV
SignalsDeterministic
CostsSpread + slip
SimulationEquity curve
Monte Carlo1,000 paths
Walk-ForwardSegment test
ApprovalPASS / REJECT
ReportHTML publish

Data is sourced from Interactive Brokers via ib_insync (MIDPOINT for FX). The pipeline runs across all instruments in config/instruments.yaml in a single execution. A cross-instrument summary is produced for portfolio-level comparison.

14.Strategy Lifecycle

Every strategy moves through a defined lifecycle. The lifecycle governs how capital is allocated — and how it is withdrawn — as evidence accumulates.

Idea Queue
Hypothesis formed
Research
Pipeline running
Portfolio Candidate
Passed validation
Paper Trading
Infrastructure test
Live
Capital allocated
Rejected
Failed gate. Archived.
Monitor
Live vs expected profile
KEEP REDUCE REINTRODUCE RETIRE

The pipeline is permanent. Individual strategies are not. A retired strategy is archived with its full specification, validation record, and failure analysis — preserving the knowledge it represents.

15.Strategies Are Disposable

We expect strategies to degrade, fail, or be superseded over time.

Kalaya's structural edge is not any individual strategy. It is the ability to continuously discover, validate, deploy, monitor, and replace strategies as market behaviour evolves.

15.1 Why Edges Degrade

15.2 The Correct Response

ResponseWhen Appropriate
ReduceEarly degradation signal detected. Edge weakening but not confirmed absent.
SuspendDegradation confirmed. Capital paused while diagnosis is underway.
RetireRe-validation fails. Edge confirmed absent. Archived as institutional knowledge record.
ReplaceA new validated strategy covers the same or better behavioural role.
ReintroduceA retired strategy shows renewed edge in changed regime conditions. Must re-validate.

The correct response to degradation is never to repair the strategy by adding complexity. That increases fragility without addressing the underlying cause.

16.Continuous Improvement UPDATED

The process is never complete.

Kaizen applies to the entire organisation — not just strategy development. Every component of the business is subject to ongoing review and improvement. Every lesson learned strengthens the institutional knowledge base.

ComponentWhat is Reviewed
ResearchSignal detection quality, data sources, research efficiency. Which approaches produce candidates and which do not — captured as institutional knowledge.
ValidationAre gating thresholds correctly calibrated? Are the right metrics being used? Each cycle's validation decisions are recorded as decision records.
ExecutionInfrastructure reliability, fill quality, guard layer calibration, execution cost accuracy.
GovernanceAre governance reviews producing useful decisions? Are risk controls still appropriate? Are lessons from previous governance cycles being applied?
ReportingDoes reporting give the right information at the right time? Is the portal useful and navigable?
Business processesOperational efficiency, documentation quality, communication between founders and partners. Lessons from operational failures are documented.
Operating modelThis document. Updated with each version to reflect current reality — not historical aspiration. Each version is a decision record in itself.

Improvements are documented, version-controlled, and implemented as discrete changes. The operating model at any point in time reflects the accumulated learning from all previous cycles. This institutional memory is a proprietary asset that grows with every iteration.

Part IV
Portfolio & Risk

17.Diversification Philosophy

The objective of diversification is stable portfolio behaviour across a wide range of conditions — not risk reduction for its own sake. A well-diversified portfolio maintains positive expectancy even when significant parts of it are in drawdown.

Markets
FX, equities, commodities — no single-market dependence
Behaviour
All four categories represented — failure regimes spread
Direction
Long and short — no structural directional bias
Timeframe
Multiple holding periods — intraday through swing
Regime
Strategies with different expected failure regimes

Single-strategy dependence is a portfolio-level risk. A portfolio that generates most of its returns from one strategy is not a robust portfolio — it is a concentrated bet on that strategy continuing to work. Kalaya treats this as a risk to be actively managed.

18.Market Regime Framework UPDATED

Capital allocation should be informed by the current market environment. Kalaya uses a structured regime framework to form a view on prevailing conditions. This is not forecasting. It is structured market thinking that supports portfolio allocation conversations.

Attribution:

The Growth / Inflation two-axis framework used below is adapted from concepts popularised by Ray Dalio and Bridgewater Associates. Kalaya does not claim ownership of this framework. It is used as a practical organising structure for regime assessment, applied alongside Kalaya's own analysis.

18.1 The Two-Axis Framework

Market regime is assessed on two axes — Growth (Increasing / Decreasing) and Inflation (Increasing / Decreasing) — producing four quadrants:

Growth ↑ / Inflation ↑
Growth ↑ / Inflation ↓
Rising growth, rising inflation
Broadly associated with risk-on behaviour. Rate pressure may increase volatility. Strategy performance in this quadrant is a subject of ongoing research.
Rising growth, falling inflation
Broadly associated with supportive conditions for risk assets. Strategy performance in this quadrant is a subject of ongoing research.
Growth ↓ / Inflation ↑
Growth ↓ / Inflation ↓
Falling growth, rising inflation
Stagflationary environment. Historically challenging across asset classes. Strategy performance in this quadrant is a subject of ongoing research.
Falling growth, falling inflation
Risk-off or recessionary conditions. Defensive assets may benefit. Strategy performance in this quadrant is a subject of ongoing research.

18.2 How Kalaya Forms a Regime View

18.3 Regime and Strategy Research — A Research Objective

This capability is aspirational, not yet proven.

One objective of the Kalaya research programme is to determine, through empirical testing, which behavioural categories and strategies perform best in each regime quadrant. If reliable regime-strategy relationships are identified in the data, this evidence may eventually be used to inform dynamic allocation decisions. Until that evidence exists, regime awareness informs qualitative allocation thinking only — it does not drive quantitative rebalancing.

19.Capital Allocation Philosophy

Portfolio construction and capital allocation are distinct activities. Portfolio construction decides which strategies belong in the portfolio. Capital allocation determines how much capital each strategy receives.

19.1 The Allocation Flow

Idea
Research
Validation
Portfolio Candidate
Capital Competition
Portfolio Inclusion
Allocation Layer

19.2 Allocation Principles

Passing validation is a necessary condition for capital allocation. It is not a sufficient condition. A validated strategy that duplicates the regime exposure of an existing strategy, or adds concentration risk, may receive zero allocation until portfolio conditions change.

20.Portfolio Risk Layer

Between validation approval and live capital allocation sits the Portfolio Risk Layer. Every validated strategy receives capital through this layer — not directly.

20.1 Risk Layer Responsibilities

ResponsibilityDescription
Position sizingApplying the risk sizing formula to determine actual position size from a fixed cash risk amount and defined stop loss.
DiversificationEnsuring new strategy additions improve portfolio diversification rather than add correlated exposure.
Exposure managementMonitoring total long and short exposure. Preventing unintended directional bias at the portfolio level.
Concentration controlCapping the percentage of total risk allocated to any single strategy, behavioural category, or instrument.
Regime adjustmentAdjusting allocation based on regime view and expected strategy behaviour in that regime — where evidence supports it.

20.2 Risk Sizing Formula

Lot size = Risk (£) ÷ Stop Loss (in points)

Every position is sized from a fixed cash risk amount and a defined stop loss distance — always. This applies in backtesting, paper trading, and live deployment.

The specific live deployment parameters — maximum risk per trade, maximum portfolio risk, and regime adjustments — remain under development and will evolve as live experience is accumulated. The formula is fixed; the parameters are calibrated over time.

21.Validation Framework

Gating metrics determine whether a strategy is deployable. Diagnostic metrics describe the character of the edge. These two groups are strictly separate.

21.1 Approval Gate — Hard Gating Metrics

All four must pass. (Source: quant-pipeline/evaluation/approve.py)

MetricThresholdRationale
profit_factor_net≥ 1.20Gross profit exceeds gross loss by ≥20% after costs
trade_count≥ 50Minimum sample size for statistical reliability
expectancy_net_pct> 0.0%Mean return per trade positive after all costs
max_drawdown_net_pct> −30%Extreme guard only — unusable signal protection

21.2 Diagnostic Metrics

Computed and reported. Do not influence approval. Inform portfolio construction and monitoring: Sharpe ratio, Calmar ratio, payoff ratio, maximum consecutive losses.

21.3 Robustness Tests

Monte Carlo — 1,000 simulations resampling trade returns with replacement. Key outputs: p95 max drawdown, p5 final return, median outcome. Reported; not gated. (Source: quant-pipeline/montecarlo/mc.py)

Walk-Forward — data split chronologically into thirds and run independently. A strategy that passes on full data but fails individual segments is flagged as fragile. (Source: quant-pipeline/robustness/walkforward.py)

Monthly Consistency — high total return concentrated in a small number of months is flagged. Consistency across months indicates structural edge, not outlier events.

22.Portfolio Construction Framework

22.1 Current Model — Signal Quality

The current portfolio construction model evaluates strategies on positive net expectancy. No position sizing or leverage is applied in the research pipeline. Results are expressed as return-per-trade percentage, normalised for cross-instrument comparison.

22.2 Target Model — Regime Overlay

The target model adds a regime overlay that adjusts strategy allocation based on market state. This is architected but not yet implemented. Market state is classified on two axes: structure (trending, ranging, transitioning) and volatility (low, normal, high). Evidence linking regime conditions to specific strategy performance must be established before this capability is deployed.

22.3 Transaction Cost Model

All simulations deduct round-trip execution cost from every trade:
round_trip_cost_pct = 2 × (spread_pct + slippage_pct) + commission_pct
Per-instrument cost assumptions live in config/instruments.yaml.

Part V
Operations

23.Execution Architecture

23.1 Current Execution Stack

The current Kalaya execution stack is Python-native, IBKR-connected. Signals are generated by Python strategy modules. Orders are placed via IB Gateway using ib_insync. All execution logic, safety guards, and monitoring run in Python.

Historical note — MT5 / MQL5:

The Kalaya repository contains MQL5 Expert Advisors built for MetaTrader 5 (DR, FBO, BB_KL, 2B). These represent an earlier research phase and are not part of the current operating stack. The MQL5 code is retained as a research archive.

23.2 Execution Safety Architecture

Every order attempt passes through four mandatory guard layers before reaching the broker. (Source: quant-pipeline/execution/execution_guard.py)

1
Kill Switch
Reads config/kill_switch.yaml on every call. If active: true — all orders blocked immediately. Takes effect within milliseconds of file save.
2
Pre-Trade Guard
Validates: account_type = paper · symbol in allowed_symbols · quantity within max_order_quantity · order type MKT or LMT · crypto blocked.
3
Risk Controls
Validates: IBKR connection alive · daily order count < 5 · open orders < 2 · rejected orders today < 3 · daily loss within −2.0% limit.
4
Paper Account Guard
All managed accounts verified to begin with "DU" (IBKR paper prefix). Any non-"DU" account triggers immediate disconnect. Live execution is hard-blocked at infrastructure level.

24.Human Intervention Layer

Execution is systematic. Allocation is supervised.

Humans do not intervene in trade execution. Humans intervene through review, allocation, risk decisions, and governance.

At the individual trade level — signal generation, order placement, position sizing, stop setting, exit execution — everything is automated. Human intervention is reserved for:

Intervention PointNature
Portfolio ConstructionWhich validated strategies enter the live portfolio. Initial allocation weights.
Capital AllocationAdjusting allocation percentages. Responding to equity changes and regime shifts.
Strategy DispositionRetain, reduce, retire, or replace decisions based on observed behaviour. Each decision is recorded.
Governance ReviewReviewing system health, monitoring outputs, risk controls.
Emergency OverrideKill switch activation. Immediate halt of all execution.

25.Strategy Behaviour Monitoring

Every deployed strategy has an expected behavioural profile derived from its validation run. Live behaviour is compared against this profile continuously. Material deviations trigger a structured review — the objective is early identification of degradation before it causes portfolio damage.

25.1 Review Triggers

A strategy enters enhanced monitoring when:

Degradation vs normal variance:

A strategy in a temporary drawdown within its expected envelope is behaving normally. Over-reacting to normal variance by retiring strategies too early is as damaging as under-reacting to genuine degradation. The monitoring framework must distinguish between the two. All review decisions are logged as decision records.

26.Data & Institutional Knowledge as an Asset UPDATED

Every research run, validation result, deployment decision, governance discussion and lesson learned contributes to the institutional knowledge base of Kalaya Capital.

This accumulated knowledge forms a long-term proprietary asset that compounds in value over time. Its quality determines the quality of every future allocation decision.

26.1 What Forms the Asset

Data TypeLocationValue
Historical price dataquant-pipeline/data/raw/Foundation for all research
Validation runsquant-pipeline/outputs/Evidence base for approval decisions
Approval recordsoutputs/<instrument>/approval.jsonDocumented evidence trail
Execution logsoutputs/executions/execution_log.csvGovernance audit trail
Research portal reportskalaya-research-portal/research/Published research history
Strategy metadata registryquant-pipeline/strategy/metadata.pyStrategy specification and history
Operating model versionskalaya-research-portal/memorandum/Decision record of business evolution

26.2 Why Rejection Records Are as Valuable as Approval Records

A rejected strategy, with its full evidence record, prevents that idea from being re-researched without new evidence. Each rejected hypothesis narrows the search space and reduces wasted future effort. Lessons from failures — what structural weaknesses were detected, what conditions caused failure — are retained as institutional knowledge and applied to future research design.

26.3 Long-Term Compounding

The value of this knowledge base compounds over time. An operation that has been running for five years, with disciplined record-keeping, has a research edge that cannot be replicated quickly by a new entrant. Building and maintaining this asset is a core responsibility of every Kalaya founder, partner, and contributor.

27.Research Universe Roadmap

Current research is focused on FX instruments. This is a deliberate starting point, not a permanent limitation.

27.1 Why FX First

27.2 Planned Universe Expansion

PhaseAsset ClassStatus
1FX — Major and Minor PairsActive
2Equities & IndicesPlanned
3Bonds & RatesPlanned
4Commodities — MetalsPlanned
5Commodities — EnergyPlanned
6Commodities — SoftsPlanned
Current research focus ≠ long-term universe.

The methodology, validation framework, and pipeline are instrument-agnostic. Universe expansion requires data access, cost model calibration, and infrastructure testing — not methodology changes.

28.Monitoring and Governance

28.1 System Monitoring

A system status JSON is written after every execution cycle, providing a complete operational snapshot without requiring an active IBKR connection. Fields: kill_switch_active, connection_status, system_state, orders_today, open_orders, rejected_orders_today, last_block_reason. (Source: quant-pipeline/monitoring/system_status.py)

28.2 Non-Negotiable Constraints

28.3 Governance Principles

Kalaya governance is lean and purpose-built. It exists to ensure allocation decisions are made on evidence; the research pipeline is producing candidates continuously; live strategies are monitored on a defined schedule; the operating model remains current; and the kill switch and execution guards are tested and functional. All governance decisions are recorded.

28.4 Execution Logging

Every order attempt is logged with timestamp, symbol, side, quantity, order type, status, and block reason. This provides the governance audit trail and contributes to the institutional knowledge base. (Source: quant-pipeline/execution/execution_logger.py)

29.Reporting Framework

29.1 Strategy Research Report

Every strategy that completes validation receives a structured HTML research report, generated programmatically from pipeline outputs. The report is designed to be scannable in under 30 seconds and is published to the research portal immediately upon generation. (Source: quant-pipeline/reporting/html_report.py)

29.2 Research Portal

All strategy reports are published to the Kalaya Capital Research Portal — a static HTML site organised by instrument and strategy type. The portal is the primary access point for all research outputs and is version-controlled and deployed via GitHub. (Source: kalaya-research-portal/)

29.3 Classification Standard

ClassificationCriteriaImplication
PASSAll gating metrics pass on ≥1 instrument. Cross-market consistency confirmed.Advances to portfolio construction consideration
WATCHLISTMarginal metrics, insufficient data, or regime-dependent edge.Monitored. Not deployed. Re-evaluated on additional data.
REJECTOne or more gating metrics fail on all instruments.Archived as institutional knowledge record. Not eligible for resubmission without new evidence.
Part VI
Status

30.Current State

Operational status as of 30 May 2026.

Research Platform
Python pipeline operational. Multi-instrument runs executing.
IBKR Data
Live data via ib_insync. MIDPOINT data confirmed for FX.
Research Reporting
Research portal live. 34 reports published across 5 instruments.
Strategy Discovery
D1 FX24 momentum breakout series completed. Signal validation experiments underway.
Operating Memorandum
v0.4 published. First baseline version.
Portfolio Candidates
In progress. No strategies at PASS status yet.
Paper Execution Infrastructure
Built and guard layers implemented. Live session validation pending.
Paper Trading — Active
Not yet active. Requires approved strategy + infrastructure validation.
Live Portfolio
Not yet active. Follows paper trading and governance sign-off.
LLP Governance
Under development. Members' Agreement, reporting and onboarding frameworks in progress.

31.Current Phase — Research & Infrastructure Validation NEW

Kalaya Capital is currently in its Research & Infrastructure Validation phase. This phase has clear objectives and defined success criteria before the next phase begins.

Current Objectives
  • Validate research pipeline — producing reliable, reproducible results
  • Validate execution infrastructure — paper orders flowing, guards functioning
  • Validate reporting framework — portal live, reports auto-generated
  • Publish and maintain operating model
  • Establish LLP structure — governance, members' agreement, onboarding
Success Criteria
  • First PASS-classified strategy (cross-market validated)
  • Paper execution operational in a live IBKR session
  • July LLP engagement session completed
  • LLP Members' Agreement drafted
Next Phase
  • Controlled Paper Portfolio
  • Portfolio selection from validated candidates
  • Infrastructure hardening
  • LLP operationalisation
  • Member onboarding
  • Reporting cadence established
  • Small-scale live deployment preparation

32.What Has Not Yet Been Proven NEW

Intellectual honesty requires stating clearly what remains unproven.

Kalaya Capital has built a rigorous process and demonstrated research capability. The following capabilities have not yet been demonstrated in live conditions and should not be claimed as proven:

CapabilityCurrent Status
Long-term profitabilityNot yet demonstrated. No validated strategies have been deployed to live capital.
Live execution performanceNot yet demonstrated. Paper execution infrastructure built; live session validation pending.
Portfolio construction effectivenessIn development. No live portfolio has been constructed. Framework is architected.
Regime allocation effectivenessResearch objective. No empirical evidence linking regimes to Kalaya strategy performance yet exists.

This transparency is not a weakness. It is the correct intellectual posture for a business at this stage of development. The value Kalaya offers at this stage is the quality of the process, the rigour of the research, and the discipline of the infrastructure — not claimed performance that does not yet exist.

33.July 2026 — First LLP Engagement Session UPDATED

Fact vs Assumption:

Fact FACT — An external session was delivered on 30 April 2026 to Andrew and Matt, presenting a weekly breakout strategy validation across five FX instruments. (Source: kalaya-research-portal/Andrew_Matt_30.04.26/)
Assumption ASSUMPTION — A July 2026 session is planned. Specific date, venue, and final attendee list are not yet confirmed.

33.1 Session Objective

The July session is the first formal LLP engagement — an opportunity to present the Kalaya vision, operating model, governance approach, research methodology, and progress to date. It is a forum for feedback, discussion, and relationship building. It is a milestone on the path toward live deployment — not the deployment itself.

Specific objectives for the session:

33.2 Success Test

The session is successful if attendees leave believing that Kalaya Capital is:

Serious
This is a considered, long-term business — not a hobby project
Structured
There is a clear process, framework, and operating model
Disciplined
Risk controls, validation gates, and governance are real and enforced
Thoughtful
Honest about what has been built and what has not yet been proven
Building Something
A business worth being part of over the long term

33.3 Demonstration Scope

AreaWhat to Show
PhilosophyOperating Memorandum v0.4. The Kalaya operating system: principles, process, and long-term vision.
ResearchResearch portal with published validation reports. Cross-instrument results. The pipeline flow.
InfrastructureIBKR paper execution stack. Guard layers. Kill switch. System monitoring.
GovernanceLLP structure. Corporate entities. Governance framework under development.
ReportingAuto-generated HTML reports. Portal publish workflow. Audit trail.
Intellectual honestyWhat has been built. What has not yet been proven. The realistic path to live deployment.

33.4 Path to October

The July session is not intended to demonstrate live capital deployment. Its purpose is to validate the operating model and build the foundations for the next phase. The path from July to a controlled live portfolio involves:

The October timeline is directional, not committed. ASSUMPTION Each step above is a dependency. Progress will be reported through regular LLP governance sessions.


Appendix A.Approval Gate Thresholds Reference

(Source: quant-pipeline/evaluation/approve.py)

MetricThresholdType
profit_factor_net≥ 1.20Hard Gate
trade_count≥ 50Hard Gate
expectancy_net_pct> 0.0%Hard Gate
max_drawdown_net_pct> −30.0%Extreme Only
Sharpe / Calmar / Payoff / Consec LossObservation onlyNot Gated
p95 MC drawdownReported, not gatedNot Gated

Appendix B.Instrument Reference

Current FX universe. (Source: quant-pipeline/config/instruments.yaml)

InstrumentIBKR SymbolSec TypeExchangeSpread (bps)Slippage (bps)
EURUSDEURCASHIDEALPRO1.50.5
GBPUSDGBPCASHIDEALPRO2.00.5
USDJPYUSDCASHIDEALPRO2.00.5
AUDUSDAUDCASHIDEALPRO2.00.5
NZDUSDNZDCASHIDEALPRO3.00.5
USDCADUSDCASHIDEALPRO2.00.5
USDCHFUSDCASHIDEALPRO2.00.5
EURGBPEURCASHIDEALPRO2.00.5
XAUUSDXAUCMDTYSMARTTBC0.5

Appendix C.Glossary

TermDefinition
AlphaReturn attributable to systematic edge — not market exposure.
Behavioural CategoryOne of four market behaviour types: Momentum/Breakout, Mean Reversion, Trend Pullback, Reversal/Exhaustion.
Capital CompetitionThe process by which validated strategies compete for portfolio allocation based on evidence, regime fit, and portfolio correlation.
Decision RecordA documented record of a governance, research, or allocation decision — including the rationale and evidence considered.
Edge DegradationThe weakening or disappearance of a strategy's positive expectancy due to regime shift, crowding, or structural change.
ExpectancyMean return per trade, net of all costs. Positive expectancy is required for approval.
Execution GuardPre-trade safety layer validating every order before it reaches the broker.
IB GatewayInteractive Brokers' server application providing TCP interface for automated order placement and data retrieval.
Institutional KnowledgeThe accumulated research history, decision records, lessons learned, and operational experience embedded in Kalaya's systems and documentation.
Kill SwitchFile-based emergency stop that blocks all order placement immediately without system restart.
Monte CarloSimulation technique resampling historical trade returns to estimate the distribution of possible outcomes.
Portfolio Risk LayerThe risk management layer between strategy validation and capital allocation, responsible for sizing, diversification, exposure, concentration, and regime adjustment.
Profit FactorGross profit divided by gross loss. ≥ 1.20 net is required for approval.
RegimeA persistent state of market behaviour — assessed on growth and inflation axes, or by structure and volatility.
Walk-ForwardOut-of-sample testing splitting data chronologically to test strategy stability across time segments.