How to tell if crypto is trending or ranging: 3 measures
Three simple regime measures on Binance Bitcoin by year: variance ratio, days above the 200-day average, and daily range. What each shows and misses.
To tell if crypto is trending or ranging, use three measures that ask different questions: which side of the 200-day average price is on, whether moves continue or reverse, and how big the daily swings are. On Binance's BTCUSDT perp they often disagreed: in 2022 price never closed above its 200-day average, yet the 24-hour variance ratio (a test of whether moves continue) was 1.00, which points to neither. The measure that changed most was the daily range, from a median 5.86% in 2021 to 2.87% in 2023.
These are measurements of past data, not a forecast.
For what came next above and below the 200-day average, see Bitcoin 200-day moving average.
What do trending and ranging mean?
A trending market keeps moving in one direction for a while. A ranging market goes up and down inside a band. Volatility is how big the swings are. These are three separate questions:
- Which side is price on? (above or below a long average)
- Do moves continue or reverse? (trend or back-and-forth)
- How big are the moves? (volatility)
So we measure each one on its own.
Which measures can you compute?
We used Binance's BTCUSDT perp (a futures contract that never expires), from September 8, 2019 to October 19, 2026.
Variance ratio. Variance is how spread out a set of numbers is. Take 24-hour price changes and compare their variance with 24 times the variance of 1-hour changes. If hours were independent coin flips, the ratio would be about 1. Above 1 means moves tended to continue. Below 1 means they tended to partly reverse. We computed it for 4-hour and 24-hour spans.
Share of days above the 200-day average. The 200-day simple moving average is the average of the last 200 daily closing prices. For Binance BTCUSDT it exists from March 25, 2020. We counted the days when the close was above it. Our 200-day moving average guide covers this line in more depth.
Daily range. The range is the high minus the low of a day, as a percent of the open. We use the middle (median) day.
What did Bitcoin's regime look like each year?
| Year | Median daily range | 4-hour variance ratio | 24-hour variance ratio | Days closing above the 200-day average |
|---|---|---|---|---|
| 2019 (from September 8) | 3.82% | 1.05 | 1.22 | not available |
| 2020 (average exists from March 25) | 4.03% | 0.88 | 0.89 | 87.6% (247 of 282) |
| 2021 | 5.86% | 0.99 | 0.95 | 70.7% (258 of 365) |
| 2022 | 4.06% | 0.99 | 1.00 | 0.0% (0 of 365) |
| 2023 | 2.87% | 0.97 | 1.08 | 80.5% (294 of 365) |
| 2024 | 3.72% | 0.97 | 0.95 | 80.1% (293 of 366) |
| 2025 | 2.98% | 0.98 | 0.92 | 73.2% (267 of 365) |
| 2026 (to October 19) | 2.92% | 0.97 | 1.01 | 21.2% (62 of 292) |
| All years | 3.74% | 0.96 | 0.97 | 59.21% (2,400 days with the average) |
The daily range moved the most: from 5.86% in 2021 to 2.87% in 2023.
The days above the 200-day average tell a clear story of direction: all the days of 2022 below it, and most days above it in 2020, 2023, 2024 and 2025.
The variance ratio mostly sat just under 1. Bitcoin's hourly moves tended to partly reverse a little, not to run on. The two high values come from 2019, which holds only part of a year. Differences of a few hundredths are within noise, because the windows overlap.
Do the measures agree?
Not always.
- 2022 had zero days above the 200-day average, the clearest downtrend in the table. Its 24-hour variance ratio was 1.00, and its 4-hour ratio was 0.99. Neither points to trending.
- 2020 had 87.6% of days above the average, yet its variance ratios were the lowest in the table (0.88 and 0.89).
The measures ask different questions. Price can stay on one side of a long average with many small reversals along the way. The variance ratio only asks whether the next hours move like the last hours.
So "trending" can mean "on one side of the average for months", or "moves keep going in the same direction hour after hour". Say which one you mean.
Do big moves follow big moves?
Yes, in this data, and more clearly than direction does. The correlation between one hour's absolute move and the next hour's was 0.30 for Binance BTCUSDT. The correlation of the signed moves was about −0.02. A correlation near 0 means one number tells you little about the other.
So a big hour tended to be followed by another big hour, but not by a rise or a fall in particular.
The same holds for days. We called a day "top 10%" if its absolute move was in the top tenth of the past 365 days. The day after one of the 202 such days was also top 10% in 32 cases, or 15.84%, against a base share of 9.05%. Those big days cluster in a few crashes, so the real number of separate episodes is smaller than 202. The pattern was not steady by year either: in 2023 it was 26.92% of 26 cases, but in 2025 it was 7.41% of 27 cases against a 7.4% base, no gap at all. Our post on volatility clustering looks at this in more detail.
Does the regime say what comes next?
Not on its own. We compared what Bitcoin did over the next 90 days after days that closed above or below the 200-day average.
| Market | After closing above | After closing below |
|---|---|---|
| Binance BTCUSDT (from March 2020) | Up 62.10% of 1,359 days; median +7.89% | Up 53.10% of 951 days; median +5.49% |
| Coinbase BTC-USD (from February 2016) | Up 66.79% of 2,349 days; median +16.99% | Up 48.64% of 1,472 days; median −0.98% |
The windows overlap. Counted as separate 90-day looks, there are only about 15.1 above and 10.6 below for Binance, and about 26.1 and 16.4 for Coinbase. Bitcoin also went through only a few big cycles. This is a description of the past, not a tested rule. For a tested rule built on this average, see golden cross and death cross results.
What does this not show?
It does not show that any of these measures predicts a coming move, and it says nothing about the regime today. It does not cover other coins, other bar sizes, or fees. The labels "trending" and "ranging" are ours, not a standard.
How we measured this
We used Binance BTCUSDT hourly and daily bars from September 8, 2019 through October 19, 2026. The variance ratio is the variance of rolling 4-hour and 24-hour log returns divided by 4 or 24 times the variance of 1-hour log returns. A log return is a way of writing a percent move that adds up neatly over time. The daily range is the high minus the low as a percent of the open, and each year's value is the median day. The 200-day average and the above-or-below counts use daily closes. Missing bars were left missing. We also repeated the 200-day average check on Coinbase BTC-USD.
Limits: windows overlap, so the counts are not independent trials; only surviving coins are in the study; and each year is a single history. The 2019 and 2026 rows are partial years.
Try this in Comparand
Ask your connected AI agent: "Show the volatility regime for BTC on 2026-10-15, with its realized volatility."
- hl_stats returns realized volatility, a volatility regime label and a correlation matrix for up to 20 coins on one day you choose. It costs $0.01 per call on pay as you go.
- This is Hyperliquid data, so it is a separate view from the Binance numbers above.
To get the raw bars for your own regime check, ask: "Get completed daily bars for Binance BTCUSDT from January 2020 and tell me what share of days closed above the 200-day average." venue_bars returns completed bars only and is free. Your agent does the maths; the tool does not return a regime report.
To see how past stretches that looked like today's played out, ask: "Find the ten closest past charts to the last 48 hourly bars of BTC on Binance." shape_analogs compares a coin only with its own past on the same exchange, and shows each case's path and what followed. It costs $0.01 per search from your credits.
You can try 3 coins free, read the tool docs, or check pricing. These are past cases, not a forecast. Our post on why past returns are not forecasts gives more context.
The short version
- No single number says "trending" or "ranging". Use measures that ask different questions.
- On Binance BTCUSDT, the median daily range ran from 5.86% in 2021 to 2.87% in 2023.
- 2022 had 0 days above the 200-day average, yet its 24-hour variance ratio was 1.00.
- Big hours and big days tended to follow big ones: 15.84% against a 9.05% base, from 202 cases.
- Regime comparisons rest on overlapping windows and few cycles, so read them as past cases, not a forecast.
Questions people ask
How can you tell if crypto is trending or ranging?
Use more than one measure. Three simple ones are the variance ratio (do moves continue or reverse?), the share of days with a close above the 200-day average (which side is price on?), and the typical daily range (how big are the swings?). Each answers a different question.
What is a variance ratio?
It compares how much 4-hour or 24-hour price changes vary with how much they would vary if every hour were independent. Above 1 means moves tended to continue. Below 1 means they tended to partly reverse. For Binance BTCUSDT since 2019 it was 0.96 at 4 hours and 0.97 at 24 hours.
Was Bitcoin trending in 2022?
Price was below its 200-day average on every day of 2022 on Binance BTCUSDT, so by that measure it was in a downtrend. But the 24-hour variance ratio was 1.00, which does not point to trending. The measures can disagree.
Does being above the 200-day average predict what comes next?
Not as a rule. On Coinbase BTC-USD, the median 90-day move after days above the 200-day average was +16.99%, against −0.98% after days below. But these windows overlap, and they hold only about 26 and 16 separate looks. They are past cases, not a forecast.
Do big days follow big days in crypto?
More often than the base rate, in our Binance BTCUSDT daily data. After a day in the top 10% of the past year's moves, the next day was also in the top 10% in 15.84% of 202 cases, against a 9.05% base rate. These cases cluster in a few crashes, and in 2025 the gap disappeared.