Multiplicative Contractions, Additive Recoveries: Functional-Form Restrictions on Risk Exposure Dynamics

Battle Count: 50

The paper provides important insights into the functional form of risk exposure dynamics during stress vs. calm regimes, which is directly relevant to risk management, position sizing, drawdown-aware trading strategies, and understanding liquidity spirals. The magnitude-dependent drawdown-recovery asymmetry (R3) has implications for mean-reversion timing and crash-recovery trading. However, the paper is primarily a mechanism-consistency test rather than a direct trading signal generator, and the exposure proxy (FINRA margin debt) is noisy. The findings inform when intermediary-capital constraints are operative, which affects strategy design during stress periods.

The core empirical test (regime-interacted OLS with HAC SEs) is straightforward. The Cox proportional-hazard model and episode detection algorithm add moderate complexity. The null model simulations (Heston SV with Milstein scheme, Markov RS, block bootstrap) require careful calibration and numerical implementation. The theoretical derivation is elegant but the empirical pipeline involves multiple steps: detrending, regime classification, interaction regression, episode detection, bootstrap CIs, and Monte Carlo null comparisons. Python scripts are provided but require data assembly for international indices and CFTC.

The paper includes a replication package (Section F) with Python scripts (run_all.py, exposure_headline.py, r3_regression.py, episode_detection.py, null_models.py, exposure_cot.py) and bundled data (S&P 500 daily prices, FINRA+VIX monthly). Dependencies are listed (numpy, pandas, scipy, statsmodels, lifelines, requests, yfinance). However, international index data and CFTC data are not bundled, and the CFTC companion test is not estimated. The master script reproduces headline numbers from cached inputs.