Financial econometrics program

ARGOS

Two-Asset Cointegration & Statistical Arbitrage Engineering

Two-Asset Cointegration & Statistical Arbitrage Engineering

$1,800

6 phases · 18 modules · capstone2 assets · not basketsRecommended next: HARMONIA
Course roadmap · 6 phases
Phase 1
RELATION
Two-asset foundations
Phase 2
SPREAD
Construction + mean reversion
Phase 3
POLICY
Signal, OU, costs
Phase 4
TIME
Stability, tracking, breaks
Phase 5
SEARCH
Selection + validation
Phase 6
LAB
Pipeline + project
Six technical phases. Eighteen academic modules (I–XVIII) live inside them. Module XVIII is the capstone. Coursework clock is not sealed.

Mission and vision

Mission

Teach two-asset cointegration as a falsifiable hypothesis: discover the pair, engineer the spread, separate trading policy from selection, and validate under costs, stability, multiplicity, and causal OOS.

Vision

Traders who can defend a bivariate relation statistically, economically, and temporally — not a high-correlation z-score recipe, and not a basket.

Why ARGOS

ARGOS is Two-Asset Cointegration & Statistical Arbitrage Engineering.

The object is `(A, B)`: a bivariate relation turned into a falsifiable hypothesis, then into a statistical-arbitrage strategy that can survive costs, stability, multiplicity, and causal OOS. It is not a high-correlation z-score recipe. It is not HARMONIA (3+ assets and baskets). It is not ARB EST (triangular cryptos). It is not the retired TAC bundle.

Who it is for

Algorithmic and systematic traders, quantitative-strategy developers, and financial engineers. Prior cointegration is not required. Statistics, linear regression, basic time series, programming, and basic backtesting are.

What it is not

Not baskets. Not Johansen rank. Not Granger–Lee. Lecture bodies are not authored yet. Recommended next: HARMONIA.

What you'll be able to do

  • Separate correlation screening from cointegration as relationship discovery
  • Engineer a hedge ratio and a spread — Spread = A − βB — as a hypothesis, not a default
  • Treat ADF, half-life, crossing, amplitude, and OU as diagnostics, not a P&L rank
  • Keep trading policy distinct from pair selection, and costs inside the hypothesis
  • Validate at scale under multiplicity, stability, and causal untouched OOS

Course content

Phase 1 — RELATION 3 modules
Module 1 — Statistical Relationships Between Two Assets Outline
Correlation ≠ Cointegration Read
Module 2 — Stationarity & Integration Outline
Integration order is a precondition, not a trade Read
Module 3 — Engle–Granger Cointegration Outline
Stationarity ≠ Profitability Read
Phase 2 — SPREAD 3 modules
Module 1 — Spread Construction Outline
Spread = A − βB is an engineered object Read
Module 2 — Mean-Reversion Diagnostics Outline
A vector of diagnostics, not a P&L rank Read
Module 3 — Economic Opportunity Outline
Fast mean reversion ≠ high net return Read
Phase 3 — POLICY 3 modules
Module 1 — Signal Engineering Outline
Optimal trading policy ≠ optimal pair selection Read
Module 2 — OU & First-Passage Trading Logic Outline
First-passage logic is a policy, not a discovery test Read
Module 3 — Costs & Executability Outline
Friction is part of the hypothesis Read
Phase 4 — TIME 3 modules
Module 1 — Pair Stability Outline
A cointegrated pair can decay Read
Module 2 — Dynamic Tracking Outline
Kalman tracking ≠ relationship discovery Read
Module 3 — Structural Breaks Outline
A broken equilibrium is retired, not re-optimised forever Read
Phase 5 — SEARCH 3 modules
Module 1 — Large-Scale Pair Search Outline
C(N, 2) is a search, not a list of independent trades Read
Module 2 — Multiple Testing Outline
Candidate count ≠ independent trial count Read
Module 3 — Causal OOS Validation Outline
OOS ≠ untouched OOS Read
Phase 6 — LAB 3 modules
Module 1 — Pair Discovery Pipeline Outline
One flow, not a collage of notebooks Read
Module 2 — Failure Taxonomy Outline
A refused pair must name its failure class Read
Module 3 — Capstone: Pair Discovery Engine Outline
A reproducible discovery process, not two tickers Read

Course content is Phase → Module → Chapter using ARGOS's six phases (RELATION → SPREAD → POLICY → TIME → SEARCH → LAB). Module numbers restart inside each phase. Module XVIII is the capstone inside Phase 6. Lecture bodies are not authored yet. Recommended next: HARMONIA.

Instructor

Elmer Niño

Elmer Niño

Founder, VTAlgo Group · Systems Engineer · MSc in Computer Science

Algorithmic trader since 2009, running trading as an enterprise. Methodology built on walk-forward analysis, genetic optimization, PCA against over-optimization, the Triple Barrier Method, and 9–12 month live validation before capital.