This paper introduces a fluctuation-theorem-like diagnostic for financial markets by comparing holding-time distributions of long and short trading positions generated from the same price history using symmetric take-profit and stop-loss rules. The log-ratio of the two distributions, Φ(dp), exhibits a robust crossover: it remains nearly constant at short durations and becomes approximately linear at longer durations, implying an exponential directional asymmetry. This structure is observed across equity indices, individual stocks, and cryptocurrencies. The tail slope defines an effective market temperature, Tm = 1/|k|, which serves as a coarse-grained measure of fluctuation intensity and reveals temporal patterns not visible in price series alone, including a global minimum around the enactment of the One Big Beautiful Bill Act in July 2025.

The authors identify the minimal stochastic mechanism using a Bachelier diffusion benchmark. While independent first-passage statistics produce exponential tails in holding-time distributions, they yield a saturating Φ(dp) because long and short positions share the same leading decay rate. In contrast, when positions are opened sequentially along a single price trajectory, pathwise correlations between overlapping trades generate direction-dependent subleading relaxation spectra in a coarse-grained Markov description. This converts correlations into a linear tail in Φ(dp), reproducing the empirical asymmetry. The mechanism is thus a finite-time, pre-asymptotic effect of coarse graining, analogous to hidden entropy production in stochastic thermodynamics. The work establishes a symmetry-based diagnostic of irreversibility in financial markets and demonstrates its utility for detecting regime shifts and system-wide correlations.