crypto-market-structurelisted

# Crypto market structure Name the venue, instrument, quote currency and bar boundary before comparing prices or returns. Continuous crypto trading is a calendar convention; it does not guarantee continuous liquidity, uninterrupted exchange access or a single settlement convention for every derivative. All measurements below were reproduced on 2026-09-14 by `scripts/market_structure.py`, seed 20260910. They use synthetic prices and are not empirical estimates of venue behavior. ## Annualization follows the observation frequency Use `fin_skills.api.conventions.annualization_factor()` and `annualize_sharpe()` when the package is installed. The standalone script falls back only when `fin_skills` itself is absent; it does not silently mask broken package imports. For ordinary calendar-day crypto returns, the library's annualization convention is 365. It uses 8,760 for hourly observations and 1,095 for eight-hour funding observations. A weekday strategy still needs weekend portfolio returns accounted for if risk is measured in calendar time. Choosing 252 merely because an equity library defaults to it is not a conversion. ✅ Measured on the same 1,460 calendar-day rows: | Statistic | Factor 365 | Factor 252 | |---|---|---| | Sharpe, mean / sample SD times square-root factor | 0.8432 | 0.7006 | | Annualized volatility | 65.81% | 54.69% | Their ratio is `sqrt(252/365) = 0.8309`; the 252 result is 16.9% smaller for this positive Sharpe. For negative Sharpe the magnitude shrink