BZ-CL

Universal Pairs
Trading System

The Universal Pairs Trading System is a production-grade algorithmic framework built to identify, test, and trade statistically cointegrated asset pairs across any asset class — energy futures, agricultural spreads, metals, and international equities. It runs a full statistical pipeline from cointegration testing through signal generation, Kelly-based position sizing, and ATR-anchored risk management.

BZ-CL — Brent vs WTI crude — is the flagship pair. Out of 15+ pairs systematically screened, it delivers the strongest risk-adjusted profile: a 14–28 day half-life, confirmed cointegration at p = 0.0025, and 39% rolling cointegration frequency across 252-day windows.

Why Brent
vs WTI?

Brent Crude (ICE, BZ=F) and WTI Crude (NYMEX, CL=F) are the two dominant global oil benchmarks, but they reflect physically distinct markets. Brent prices seaborne North Sea crude — the international reference. WTI prices US inland crude at Cushing, Oklahoma — the domestic reference. Their prices are structurally linked through crude quality differentials, trans-Atlantic shipping costs, pipeline capacity, and refinery demand.

This structural economic linkage creates a persistent statistical equilibrium. When geopolitical shocks, inventory imbalances, pipeline outages, or sanctions widen the Brent-WTI spread beyond its equilibrium range, arbitrage forces — refiners switching crude grades, tanker routing changes, export/import flows — work to close the gap. This is the fundamental driver of mean reversion.

The log spread S_t = log(BZ_t) − β·log(CL_t) is stationary with a half-life of 14–28 days. This means the spread reverts to its historical mean in roughly 2–4 weeks on average — an actionable timeframe for systematic signal generation.

Statistical
Foundation

Pairs trading operates on the principle of cointegration — a stronger condition than correlation. Two price series are cointegrated if a linear combination of them is stationary (mean-reverting), even if each series individually is non-stationary (I(1)). This guarantees a long-run equilibrium relationship and provides the theoretical justification for mean-reversion signals.

Seven-Stage
Signal Pipeline

Each trading day the system runs the full pipeline from raw price data to a binary position signal. Multiple filters are applied sequentially to reduce false positives while preserving genuine mean-reversion opportunities. The cointegration gate alone eliminates 61% of trading days, keeping the system inactive until structural conditions hold.

Backtest
Performance

Tested on daily futures data from 2014–2025 (11 years) with 5 bps per-side commission and 2 bps average slippage modeled. $500,000 initial capital. Zero look-ahead bias — all statistics computed on expanding in-sample windows only.

Visual
Analysis

Price Series — Brent (BZ=F) vs WTI (CL=F)

Log Spread · Z-Score · Entry / Exit Signals

Equity Curve — $500,000 Initial Capital

Trade Entry / Exit Overlay on Spread

Monthly Returns Heatmap

Rolling Cointegration Status — 252-Day Windows

Full Trade
Journal

All 28 completed trades from the backtest period. 25 wins, 3 losses. The three losses are concentrated in 2016–2021 — the post-WCS-widening period when Brent-WTI dynamics were most disrupted by US shale export growth and Permian basin bottlenecks. The system has been profitable on every trade since late 2021.

Win/loss is determined by spread P&L direction before friction costs. Portfolio return % includes all transaction costs and Kelly position sizing.

Screened
Universe

The system was systematically applied to 15+ pairs across energy, agriculture, metals, and international equities. BZ-CL was selected as the flagship on the basis of confirmed cointegration, actionable half-life, and best risk-adjusted performance.

★ BZ-CL selected as flagship on confirmed cointegration + actionable half-life + best risk-adjusted profile. Non-cointegrated pairs are tracked but not live-traded as primary strategies. Data from multi-pair backtest run 20260511.