Relative Value · Two legs and one relationship
Research relative value through economic relationships, hedge ratios, and invalidation tests.
Observation
Two assets move similarly, then their spread widens abruptly. Ask which risk or cash flow they share and what constrains the spread. Correlated returns prove neither a stable mean price spread nor a riskless short of the expensive side.
Mechanism
Shared factors and substitution demand may maintain relative prices; asset-specific events, supply changes, and financing constraints can permanently break the relationship. Trading income may come from spread repair or unidentified market Beta.
Hypothesis
Estimate the teaching candidate's hedge ratio in training and signal on residual deviations. Freeze estimation-update rules, holding horizons, and exits in testing. Coin-denominated, dollar-notional, and regression-Beta hedges have different exposures. Choose and explain one; do not call all of them “one-to-one.”
Data
Synchronize visible prices, volume, borrowing costs, and contract multipliers for both legs. Asynchronous prices can create lead/lag that looks like a spread. Retain asset-specific events and borrow availability. An unavailable short leg is an unexecutable trade.
Signal
Standardize the spread from past data and convert it into both legs' notionals. Submit signals together for risk checks, but fills may occur at different times. Record temporary one-leg exposure limits and remedy order. Pause entries if the residual mean or variance becomes unstable; do not keep widening the window to conceal changes.
Backtest
Account separately for each leg's positions, returns, costs, and financing before aggregation. Estimate hedge ratios from history only. Compare unhedged, fixed-notional, and rolling-Beta hedges. Statistical arbitrage needs additional spread-stability checks; see Advanced C.
Falsification
Test relationship breaks, short recalls, and unfilled single legs. Conclusions that reverse with a different estimation start, profits explained by direction, or spreads that never revert all weaken the mechanism. High historical correlation does not replace holding limits and cash-risk controls.
Exercise and project acceptance
Write three hedge definitions for the same pair of teaching assets, listing remaining exposures. Freeze one and submit spreads, both legs' PnL, Beta, costs, and invalidation records. A reviewer must distinguish spread profits from directional profits; unavailable shorts must not be recorded as filled.
Continue to the portfolio risk report to examine factors shared with other strategies.