Skip to reversion research
Mean Reversion StrategyReference / Excursion / Evidence

Pairs spread identity and hedge-ratio ownership

Require real leg sources, alignment and frozen parameters before a residual is treated as a reversion model.

A pair has two identified data objects

Both legs need exact instrument identity, permitted source, adjustment policy and observation clock. A synchronized join is not established by equal row counts alone. One leg can arrive later, be missing or use a different revised calendar. Preserve join failures and lawful access rather than selectively keeping attractive pairs. The positive-price lens does not accept signed spread values and cannot stand in for a two-leg adapter. Its spread card is a complete written question with actual leg history absent, not an alleged trained pair or genuine candidate result.

The ratio belongs to its fit scope

A hedge ratio or equilibrium parameter estimated on the full history can include evaluation information. Identify the actual fitting population, method, parameters and effective version. Applying a frozen ratio to later legs is different from refitting it on those outcomes. If an adaptive policy is predeclared, its update rules also belong to identity and evaluation. A coefficient value without its source scope is not sufficient evidence. This edition estimates no ratio, selects no pair and reproduces no proprietary residual; original fitting records remain a requirement rather than an invented completion field.

Residual labels have their own destination and scale

A fitted spread can be signed and use units unlike either leg's price. Its mean, scale, event threshold and recovery endpoint need the matching process definition. A price-level z-score or raw one-leg reversal cannot automatically calibrate that residual. Structural changes and revised joins can also change label ownership. Preserve the frozen destination during an identified event or declare a different moving policy beforehand. The path workbook teaches operator differences with positive EDU closes only. No genuine probability of spread recovery or independently stable relationship is inferred from it.

Two-leg economics remain outside the scalar label

Actual execution requires both legs' orders, quantities, fills, costs and applicable financing or borrow rules. A residual crossing does not automatically deliver a positive account result. Capital weights, partial actions and native exit ownership need complete accounting. Full research must use its authoritative asset route and independently verified master; it cannot borrow a source file or label from another class while erasing identity guards. Cloud model promotion and subscriber certification remain separate authorized steps. The site publishes documentation rather than connecting accounts or creating a live pair-trading pipeline.

Authored reference and excursion case

An authored spread card proposes two anonymous legs but supplies no actual prices, fitted ratio or aligned source versions. It therefore cannot calculate a genuine residual or claim stationarity. The correct state is written specification complete, actual source and research unestablished. Replacing its missing residual with the lens's positive-price default would change the object. The example makes that incompatibility explicit without inventing a pair, fill or return.

ObjectRecordBounded finding
Leg A and BActual sources absentNo authentic join
Hedge ratioFit artifact absentNo frozen parameters
ResidualNot suppliedPositive-price shortcut invalid
Master and accountAbsentNo financial completion
Authored gradual and overshoot paths around frozen reference 100
Authored EDU paths, not genuine market prices, return observations or calibrated odds.

Primary references and scope

Source review: 14 October 2026. Official mechanism descriptions are not this desk's paid-account test, independently captured fills or vendor-return audit.

Related reversion questions