| Term | What it measures | Formula | Why a Cointelegraph reader should care |
|---|---|---|---|
| Log return | Every price move on these dashboards is expressed as a natural logarithm rather than a plain percentage. It is a small change with a useful property: log returns can be added together across hours, days and weeks, and they treat a gain and an equivalent loss symmetrically. | return = ln( price now ÷ price before ) log returns can be summed across hours, days and weeks | Anyone who has watched a position fall by half knows it takes a double to get back to even, and plain percentages quietly hide that asymmetry. If you are comparing a figure here against one from another site, it is worth checking which convention that site uses, because the two drift furthest apart in exactly the volatile stretches you most want to measure. |
| Realized volatility | This is how much an asset has actually been moving. The dashboards measure it as the standard deviation of hourly returns over the window you choose, then scale the result to a yearly rate. | volatility = typical size of an hourly move × √8760 8760 hours in a year; the window is the past 1, 7 or 30 days | If one asset is running at twice the volatility of another, the same amount of money in each will swing twice as hard. Sizing two positions equally in dollars therefore does not size them equally in risk, and this is the number that tells you how far to adjust. Watching it over several weeks also separates a market that is genuinely settling down from one that is only pausing. |
| Annualization | Every volatility figure is quoted as a yearly rate so that different assets and different windows can be compared on a single scale. Because crypto trades continuously, the conversion uses all 365 days rather than the 252 trading days that equity markets assume. | yearly volatility = k-day volatility × √( 365 ÷ k ) typical day ≈ yearly figure ÷ 19 · typical week ≈ yearly figure ÷ 7 | A reading of 30% does not mean the asset will move 30% this week, and the conversion back down is easy enough to do in your head: divide by about 19 for a typical day and by 7 for a typical week. That turns an abstract figure into a range you can weigh against your own stop distance. |
| Upside volatility | This applies exactly the same calculation as realized volatility, but counts only the hours in which the price rose. | upside volatility = the same calculation, keeping only the up-hours | Volatility tends to get treated as a synonym for danger, when half of it is the movement that makes holders money. An asset whose swings are mostly upward is being called risky for what is really a strong rally. The more useful signal is which half is growing: when upside is climbing while downside stays flat, the market is becoming more rewarding rather than more dangerous. |
| Downside volatility | The mirror image, counting only the hours in which the price fell. The two halves fit back together exactly, so an asset's total volatility is fully accounted for by its upside and downside components. | downside volatility = the same calculation, keeping only the down-hours the halves reunite: total² = upside² + downside² | This is the half that corresponds to losing money, and it is the one to hold against your stop distance. When downside begins running above upside, the falls are arriving faster than the rallies even though the headline volatility figure may not have moved at all, and stops calibrated on that blended number will sit too close to the price. |
| Volatility spread | The spread subtracts one half from the other. A positive number tells you the rallies have been the sharper moves, while a negative one means the falls have been doing more of the work. | spread = upside volatility − downside volatility | It compresses a great deal into a single figure: whether the market is currently paying holders or punishing them. What matters most is the moment the sign changes, because the spread tends to turn before the price trend does. A market that quietly stops rallying hard and starts falling hard is changing character while the chart still looks flat. |
| Quintile regime bands | Rather than comparing assets against one another, every reading is ranked against that same asset's own trailing year and sorted into one of five bands, running from Very Low to Very High. | band = where today sits in that asset's own past year, split into fifths bottom fifth = Very Low · top fifth = Very High | This is what lets you judge a number without already knowing the asset well. Ninety percent volatility might be an ordinary week for a meme coin while forty percent is an emergency for BTC, and the bands make that translation for you. It is the only practical way to scan a watchlist of assets that otherwise have nothing in common. |
| Measurement period | Every figure is calculated over a rolling window of either 1 day, 7 days or 30 days of hourly data. | window = the last 1, 7 or 30 days of hourly data | The window is part of the number, so it pays to match it to how long you actually hold. Someone holding for a week should be reading the 7-day figures, and someone trading intraday the 1-day. Setting a 1-day reading for one asset against a 30-day reading for another is the easiest way to talk yourself into a conclusion the data does not support. |
| Term | What it measures | Formula | Why a Cointelegraph reader should care |
|---|---|---|---|
| Current Volatility Snapshot | This shows the live total, upside and downside volatility for whichever asset you have selected, with each figure placed against that asset's own range over the past year. | today's total, upside and downside volatility, each vs its one-year bands | Before reacting to a move, it is worth knowing whether the move is actually unusual. The bands run that comparison for you, so a day that sounds dramatic can be checked against the asset's own normal in a couple of seconds. |
| Volatility Over Time | The same three measures are plotted as a history rather than as a single reading, which makes the shifts between calm and turbulent stretches visible. | the same three series, drawn through time | A number on its own cannot tell you which way conditions are heading. The same reading means a market settling down if it follows a turbulent month, and one waking up if it follows a quiet one. It is the direction of travel, rather than the level, that tells you whether to give a position more room or less. |
| Rolling Return | This tracks the cumulative gain or loss over the preceding window, plotted through time. | return = price now ÷ price k days ago − 1 | Read alongside volatility, it answers whether the turbulence was worth enduring. Two assets can have moved with identical violence and delivered opposite results, and this is the line that tells them apart. |
| Term | What it measures | Formula | Why a Cointelegraph reader should care |
|---|---|---|---|
| Forecast volatility | This is the volatility the model expects over the next 1, 7 or 30 days. It is produced fresh every hour for each covered asset and quoted annualized, exactly like the realized figures sitting beside it. | expected volatility over the next 1, 7 or 30 days | Everything else on these dashboards describes a week you can no longer trade. The comparison that matters is against the realized figure next to it: when the forecast sits higher, the models are expecting a livelier week than the one just finished, which is a reason to look at your position size now rather than after the move. |
| GARCH model | The forecast comes from a GARCH model, a design that has been standard on institutional desks since the 1990s. It captures two things markets reliably do: turbulence arrives in clusters, and a fall raises expected volatility more than a rally of the same size. | next volatility = a base level + today's move (falls count extra) + a memory of recent volatility | This explains behaviour that would otherwise look like the model overreacting, when the forecast jumps the morning after a selloff and then drifts back down through a quiet fortnight. Because falls count for more, a forecast climbing while the price sits still usually means recent downside is being carried forward. |
| Term structure | The forecast is calculated at every horizon from 1 to 90 days and drawn as a curve, with the same curve from 7, 30 and 90 days ago laid behind it. | one forecast per horizon, 1 to 90 days out, joined into a curve | The shape reads as a timeline. A curve sloping upward says the market is quiet now but expects more movement later, while one sloping downward says the current stress is expected to pass. Comparing today's curve against last month's shows whether expectations are building or draining, which is usually what decides between hedging now and waiting. |
| Forecast dial | The gauge places each forecast within a coloured risk band drawn from three years of history, which is a longer memory than the single year the heatmap works from. | band = where the forecast sits within three years of that asset's history | The two views can disagree, and the disagreement is informative rather than a fault. A reading that looks extreme against the past year but ordinary against three years describes a very different situation from one that looks extreme against both. |
| Volatility Regime Heatmap | The heatmap shows realized and forecast volatility for every covered asset at 1, 7 and 30 days, with each of the six cells per asset coloured by its band. | each cell = the band for one asset at one horizon, realized or forecast | Reading down a column is the quickest way to tell whether a move belongs to one asset or to the whole market. The pattern worth watching for is a board where the realized side is calm and the forecast side is hot, which is the signature of a compressed market in which ranges are more likely to break than to hold. |
| Term | What it measures | Formula | Why a Cointelegraph reader should care |
|---|---|---|---|
| Sector classification | The top 100 assets are grouped into nine sectors according to what they actually do, running from Blockchain Infrastructure and DeFI through to Gaming and Meme. | top 100 assets → 9 sectors | Crypto stopped trading as a single block years ago. Setting an asset against its own sector is what separates a move that belongs to that asset from a rotation the whole sector is taking part in, and it often reveals that a view someone holds about crypto is really a view about one corner of it. |
| Market-cap weighting | Each sector and size figure is a weighted average of its members, with larger assets counting proportionally more. | group figure = average of members, weighted by market cap | It is worth knowing what a group number actually represents. In a sector dominated by one or two large assets, the figure is largely describing those assets, so a sector reading can be a single coin wearing a disguise. A glance at the members first tells you whether to treat it as a broad signal. |
| Sharpe Ratio | Sharpe divides the return earned above the risk-free rate by the total volatility taken to earn it. It has been the standard risk-adjusted measure since the 1960s. | Sharpe = extra return earned ÷ total volatility taken | It puts a violent asset and a placid one on the same footing. As a rough guide, a reading near zero means the risk went unrewarded, above 1 is respectable and above 2 is strong. Two assets that returned the same amount are not the same trade if one of them managed it at half the Sharpe. |
| Sortino Ratio | Sortino works the same way but divides only by downside volatility, so an asset is not penalised for moving sharply upward. | Sortino = extra return earned ÷ downside volatility only | In a market where the best days are also the wildest, Sharpe quietly punishes assets for rallying. Reading the two together is what makes them useful: when Sortino sits well above Sharpe, the volatility that made an asset look dangerous was mostly gains, and its headline risk figure overstated the real threat. |
| Term | What it measures | Formula | Why a Cointelegraph reader should care |
|---|---|---|---|
| Size cohorts | The top 100 are divided by market-cap rank into five groups of twenty, running from the majors down to the tail. | five groups of twenty, by market-cap rank | This settles the recurring question of whether moving further down the size curve is being rewarded. When the majors deliver more return with less volatility, the extra risk in small caps is not being paid for and there is little reason to reach for it. When the tail begins to out-earn the majors on a risk-adjusted basis, that is one of the earlier signs of a speculative phase. |
| Term | What it measures | Formula | Why a Cointelegraph reader should care |
|---|---|---|---|
| Value at Risk (VaR) | Value at Risk marks the loss a single day should stay inside, at whichever confidence level you select, measured over a 24-hour horizon. | VaR = the line the worst 1 day in 20 falls below at 95% confidence; rarer settings move the line further out | It draws the line between an ordinary bad day and a genuinely exceptional one. At 95% confidence you should expect losses to stay within it on nineteen days out of twenty, so a day that breaks through is telling you something rather than simply being unlucky. |
| Filtered Historical Simulation | The estimate comes from Filtered Historical Simulation. Past returns are first stripped of their own volatility to leave the raw surprises behind, and those surprises are then re-sized to the volatility expected tomorrow. | take history's surprises → re-size them to tomorrow's conditions → read off the worst days | This is why the figure moves with conditions when simpler versions do not. A plain historical estimate treats a crash from three years ago as just as relevant as a quiet Sunday, whereas this one re-dresses every past shock in tomorrow's expected weather. The result tightens in calm markets and widens in stressed ones, which is what you want it to do. |
| Confidence level | The same asset can be quoted at 95%, 99% or 99.5%, and each setting reaches further into the tail. | 95% = 1 day in 20 · 99% = 1 in 100 · 99.5% = 1 in 200 | A 99.5% figure is not a gloomier forecast than a 95% one, it is simply a rarer event, and the two are not comparable. Before setting a VaR against another asset or another publication, check that both are quoted at the same level. |
| Expected Shortfall | Expected Shortfall averages the losses on the days that break through the VaR threshold, answering the question VaR deliberately leaves open. | Expected Shortfall = the average of the days beyond VaR | VaR tells you where the tail begins and this tells you how far into it you typically go, which makes it the better number to size against. Working from VaR alone amounts to planning for the mildest of the bad days. |
| Expected Tail Median | Expected Tail Median takes the middle of those same losses rather than their average. | Expected Tail Median = the middle of the days beyond VaR | Reading it next to Expected Shortfall shows you the shape of the tail. When the average is much worse than the middle, a handful of extreme days are doing most of the damage and the risk is lumpier than the average alone suggests, which is an argument for holding a smaller position than that average would imply. |
| Skewness | Skewness measures which way the distribution leans. A negative reading means the large surprises tend to be losses rather than gains. | skewness = the lean of the returns negative = big losses more common than equally big gains | It tells you which direction to expect the shocks from. In a negatively skewed market the big moves are falls, so stops fill worse than expected and gaps open downward more often than upward. The practical consequence is that getting out during a decline costs more than an average day would lead you to budget for. |
| Kurtosis | Kurtosis describes how much of the total risk is concentrated in rare days. A calm, bell-curve market scores 3, and anything higher means the extremes carry more of the weight. | kurtosis = the weight of the tails a calm, bell-curve market scores 3; crypto usually scores far higher | Crypto sits well above that benchmark, which is a formal way of saying it spends long stretches quiet and then moves enormously. A calm month is therefore weak evidence that an asset is safe, and position sizes are better set with the rare day in mind than the typical one. |
| Drawdown | A drawdown is how far an asset currently sits below its own previous peak, tracked continuously on hourly prices. | drawdown = price now ÷ highest price so far − 1 | Volatility is the statistician's measure of risk, while drawdown is the one holders actually live through. It is what decides whether somebody sits tight or capitulates, and looking at an asset's drawdown history before buying is the most direct test of whether a position is the right size. |
| Maximum Drawdown | The maximum drawdown is the deepest peak-to-trough fall over the period being examined. | max drawdown = the deepest such fall in the period | The question to ask yourself is whether you could have held through it. The consequence for a leveraged position follows immediately: at two times leverage, a 50% drawdown is not a painful stretch but a total loss. |
| Duration and recovery | Alongside the depth, the dashboard records how long each decline lasted and how long it then took to climb back to the old peak. | days from peak to bottom · days from bottom back to the old peak | Depth is only half of what breaks a holder. The same fall is a very different experience depending on whether it recovered in a fortnight or ground on for a year, and it is the long ones that shake people out at the bottom. This is the number that tests whether your intended holding period is realistic. |
| Ulcer Index | The Ulcer Index scores the depth and persistence of declines over a rolling 30-day window, counting deeper falls more heavily. | Ulcer = average depth of the last 30 days' drawdowns, with big falls counted extra | Unlike volatility it ignores the upside entirely and penalises time spent below the previous high. Two assets with identical volatility can look completely different here, one dipping and recovering while the other sinks and stays down. When the Ulcer reading rises while volatility holds steady, the market is becoming more painful to hold without looking any more volatile. |
| Frequency of extreme moves | This counts how often, over the last 30 days, the 24-hour move exceeded 5%, 10%, or 2.5 times that asset's own usual swing. The final threshold adapts to each asset. | the share of the last 30 days' moves beyond ±5%, ±10%, or 2.5× the asset's usual swing | It turns the tail statistics into something you can simply count. A month without a single extreme day points to a compressed market, while a cluster of them marks one in an altogether different state. Because the last threshold scales to the asset, it flags moves that are genuinely unusual for that coin instead of penalising the ones that are always volatile. |