This article explains measurement conventions, including the ones this site uses. Conventions are choices, not truths; what matters is that they are stated and applied consistently. Nothing here is investment advice. Decisions are yours.
Crypto trades every day of the year, and that single fact quietly corrupts almost every statistical comparison made between crypto results and everything else. Annualized volatility, Sharpe ratios, average daily returns, drawdown durations: all of them carry a hidden calendar assumption, and mixing the 252-day equity calendar with the 365-day crypto calendar produces numbers that look comparable and are not.
Where the calendar hides in the math
Annualization scales a per-period statistic to a yearly one, and the scaling factor is the number of periods per year. Equity convention uses 252 trading days; crypto has 365. The workhorse conversions: annualized volatility multiplies daily volatility by the square root of the period count, roughly 15.9 for equities and 19.1 for crypto, and an annualized Sharpe ratio inherits the same square-root factor from its denominator. The practical consequence: identical daily behavior produces different annualized numbers under the two calendars, about a twenty percent gap from the square roots alone. A crypto strategy annualized at 365 and casually compared against an equity fund annualized at 252 enjoys a flattering conversion nobody mentioned. Serious shops state the calendar; careless or convenient ones let the ambiguity work for them.
Three comparison traps
Trap one, the Sharpe import: quoting "hedge funds average a Sharpe near 1" as a benchmark for a crypto number computed on a different calendar, different rebalancing, and wildly non-normal returns. The comparison needs the same calendar and honestly stated assumptions, or it is theater. Trap two, the compounding mismatch: 365 compounding days per year is more compounding surface than 252; return figures spanning identical calendar windows are comparable, but per-trading-day statistics are not, and sliding between the two mid-argument is a classic marketing move. Trap three, the drawdown-duration illusion: a "forty-day drawdown" in crypto includes weekends where equity drawdowns pause the clock, so duration comparisons across asset classes silently mix wall-clock time with market time. None of these traps require dishonesty to mislead; they only require nobody checking the calendar, which is why the monitor methodology states ours explicitly.
No closing bell, no closing print
The second consequence of 24/7 is subtler than arithmetic: there is no natural daily boundary, so every daily statistic depends on an arbitrary line. We draw ours at 00:00 UTC and say so, but understand what the choice buys and costs. A different boundary produces different daily bars, different moving averages, different breakout levels, and occasionally different trend states on the same underlying prices; research that fails to pin its boundary can accidentally, or conveniently, harvest that variation. It also means the close-confirmation logic that equity systems rely on, the official auction print that settles the day's argument, has no crypto equivalent: our daily decisions key off the UTC close because a published convention beats an implicit one, not because midnight has market meaning. When you read any crypto backtest, the question "whose midnight?" is not pedantry. It moves results.
Weekends, the always-on tax and the always-on gift
Equity risk sleeps from Friday close to Monday open; crypto risk does not. That cuts both ways and both deserve stating. The tax: positions carry full exposure through weekend books that are measurably thinner, where liquidation cascades meet less resting liquidity and produce the disproportionate share of violent wicks, one of the mechanisms that shredded equity-calibrated stops in our founding transplant experiment. The gift: no overnight gaps. An equity stop can be jumped by a gap and fill far below its level; a 24/7 market's continuous tape means stops execute near their prices, wicks permitting, which genuinely improves the fidelity of stop-based risk management. A crypto system's risk model must price the weekend tax; it gets the gap-free gift in return, and honest research acknowledges both rather than whichever flatters.
Our conventions, on the record
For every number this site publishes: days are UTC calendar days closing at 00:00 UTC; the year is 365 days for all annualization; volatility and Sharpe-type figures state their daily basis and scale by the square root of 365; return windows are calendar windows; and any comparison we ever draw against equity statistics will restate both sides onto an explicit common basis or flag the difference in the same sentence. These choices are ordinary, and that is their virtue: conventions should be boring, published, and permanent, so that every extraordinary claim stands on arithmetic a reader can check rather than on a calendar nobody mentioned. Publishing the basis is what keeps this measurement rather than advice. The next time a crypto result impresses you, find the calendar before you find the feeling. Decisions are yours.