BTC Metrics

Why the vertical axis is logarithmic by default

On a linear scale every equal vertical distance is an equal number of dollars: the gap from $10k to $20k looks the same as $70k to $80k. On a logarithmic scale every equal vertical distance is an equal percentage move, so a doubling looks the same whether it happens at $100 or $100,000.

That distinction matters more for Bitcoin than for almost any other asset, because the price has moved across five orders of magnitude. On a linear chart of the full history, everything before 2017 is a flat line pressed against the bottom — the entire 2013 cycle, a move of more than 100×, becomes invisible. Log scale makes cycles comparable to each other, which is the whole point of looking at fifteen years at once.

Use linear when you care about absolute dollars — sizing a position, reading a drawdown in money rather than percent, or zooming into a few weeks where the percentage range is small anyway. Switch with the Scale control above.

The moving averages, and what each is for

A moving average is the mean of the last N daily closes, recalculated each day. It strips out daily noise so the underlying trend is visible. All three here are simple averages (SMA), equally weighting every day in the window.

• 50-day — medium-term trend. Reacts within weeks, so it turns early but also whipsaws in choppy markets.
• 200-day — the standard long-term trend line across all of finance. Price above it is conventionally read as a bull regime, below it as a bear regime. The Mayer Multiple shown further down this page is simply price divided by this line.
• 200-week (1,400 days) — a Bitcoin-specific cycle floor. Price has spent only a handful of days below it in Bitcoin's history, all at the deepest bear-market lows. It rises almost monotonically, which is why it works as a floor rather than a signal.
• Realized price — not a moving average at all. It is the average price at which every coin in existence last moved on-chain: the network's aggregate cost basis. When price falls below it, the average holder is underwater.

When the 50-day crosses above the 200-day it is called a golden cross, and below it a death cross. Both are widely watched and both are lagging by construction — they confirm a move that has already happened rather than predicting one.

The honest caveat: moving averages describe the past. They are smoothing, not forecasting. They are most useful for answering "what regime are we in?" and least useful for "what happens next week?"

Further reading

• Glassnode Academy — Realized Price (what cost basis means on-chain)
• LookIntoBitcoin — 200-Week MA Heatmap (the cycle-floor behaviour, charted)
• Binance Academy — Moving Averages Explained
• Investopedia — Moving Average · Logarithmic vs. Linear Scale
• Glassnode — MVRV metric guide

What MVRV actually compares

There are two ways to put a value on the whole Bitcoin network, and MVRV is the ratio between them.

• Market value — today's price multiplied by every coin in existence. What the market says the network is worth right now.
• Realised value — take every coin individually, value it at the price it last moved on-chain, and add those up. Roughly what the network collectively paid for its coins: an aggregate cost basis.

MVRV is the first divided by the second. An MVRV of 2.0 means the average coin is worth twice what it last changed hands for — the network is holding 100% unrealised profit. An MVRV below 1.0 means the average holder is underwater.

Why anyone cares: unrealised profit is the fuel for selling. When almost everyone is deep in profit there is a lot of incentive to take some, which is the condition that has historically existed near tops. When most coins are held at a loss, the sellers have largely already sold — the condition near bottoms.

Reading the chart

The shaded zones mark levels that mattered historically. In Ratio view:

• Below 1.0 (blue) — the average holder is underwater. Every such period so far has been a bear-market low.
• 1.0–1.5 — accumulation. Modest profit, no froth.
• 1.5–2.4 — mid-cycle.
• 2.4–3.7 — elevated; late-cycle territory in past bull markets.
• Above 3.7 (red) — historically frothy. Readings here clustered near tops.

Why there is a Z-score version

The raw ratio has a problem: its extremes have shrunk each cycle as the market matured. The 2013 peak reached 4.7, but the 2021 peak only 2.8 — so "above 3.7 means a top" quietly stopped being true.

The Z-score rescales the gap between market and realised value by how volatile market cap has been up to that point, which makes different eras comparable. On this data the 2013 peak scores 7.2 and the 2017 peak 8.6, while the November 2022 low scores −0.3. Above roughly 7 has meant a cycle peak; below 0 has meant a cycle bottom.

Note what the Z-score chart shows about the last two cycles: each peak has been markedly lower than the one before it (8.6, then 3.4). Whether that is Bitcoin maturing or simply two data points is genuinely unsettled.

What it cannot tell you

• Four cycles is a tiny sample. Every threshold on this page is a description of four events, not a law. Treat them as context, not signals.
• Realised value counts any movement as a purchase. When a coin moves between wallets or onto an exchange, its cost basis resets to that day's price even though nobody bought anything. Heavy on-chain activity therefore drags realised value up on its own.
• Lost coins distort it. Coins from the earliest years that will never move again are still valued at a few cents, holding realised value down and MVRV up. Estimates of how much supply is permanently lost vary widely, and none of them are in this number.
• It says nothing about timing. MVRV sat in "frothy" territory for months before the 2017 top and for weeks before the 2021 one.

Further reading

• Glassnode — MVRV metric guide
• Glassnode — MVRV Z-Score
• Glassnode Academy — Realised Price (the cost-basis idea underneath all of this)
• LookIntoBitcoin — MVRV Z-Score chart (for comparing against another implementation)
• Coin Metrics — CapMVRVCur (the exact metric this page uses)

What the number is

A single score from 0 to 100 meant to capture how the crypto market is feeling. Low is fearful, high is greedy. The publisher labels the ranges, and those labels are what the shaded zones on the chart use — they are not thresholds we invented:

• 0–25 Extreme fear · 26–46 Fear · 47–54 Neutral · 55–75 Greed · 76–100 Extreme greed

It is built from a weighted mix of inputs the publisher describes as volatility, market momentum and trading volume, social media activity, Bitcoin's share of total crypto market cap, and search trends. The exact weights are the publisher's and are not independently verifiable.

How it is meant to be used

As a contrarian gauge. The idea, which the publisher states outright, is that extreme fear tends to appear when an asset is oversold and people are selling irrationally, and extreme greed when a correction is overdue.

Historically the extremes have been the informative part: readings at or below 25 have clustered around local and cycle lows, and readings at or above 76 around local tops. The middle of the range says very little.

Turn on the 30-day average. The raw index can swing fifteen points in a week, and single-day readings are mostly noise — the regime is what matters, not the tick.

What it cannot tell you

• It is partly circular. Volatility and momentum are price-derived inputs, so a large part of "sentiment is fearful" is really "price fell". It is not an independent signal in the way an on-chain metric is.
• It has no timing precision. The index sat in extreme greed for weeks before some tops and only days before others. It can stay pinned at an extreme far longer than a position can be held against it.
• The methodology is undisclosed and has changed. Weights have been adjusted over the index's life, so a 2018 reading of 80 is not strictly the same measurement as a 2026 one.
• Social inputs can be gamed. Any index reading public posting activity is exposed to coordinated posting and bots.
• Short history. It begins in February 2018, so it has seen one complete cycle and parts of two others — less history than anything else on this page.

Further reading

• alternative.me — the index itself (the publisher, including its own description of the inputs)
• Binance Academy — What Is the Crypto Fear & Greed Index?
• Investopedia — Fear and Greed Index (the original equities version the crypto one is modelled on)

What the funding rate actually is

A perpetual future never expires, so nothing naturally drags its price back to spot. The funding rate does that job: every eight hours, whichever side is crowded pays the other. When the perpetual trades above spot, longs pay shorts. When it trades below, shorts pay longs.

So the sign tells you which side is crowded, and the size tells you how much that crowd is willing to pay to stay there.

How big is big? Annualise it

The chart is in basis points per settlement, where 1 bp = 0.01%. That sounds like nothing until you remember it is charged three times a day:

• 1 bp × 3 × 365 ≈ 11% per year
• 3 bp ≈ 33% per year · 5 bp ≈ 55% per year · 7 bp ≈ 77% per year

At 5 bp a leveraged long is paying more than half its position value annually just to keep the trade open. That is why sustained high funding rarely lasts: it is expensive, and the positions paying it are the ones that get liquidated first when price dips.

The scale, and what each level has meant

• Below 0 — shorts are paying longs. The crowd is positioned bearish. Historically clustered around local lows, because a market where everyone is already short has few sellers left.
• 0 to 1 bp — quiet. 0.01% (1 bp) is the exchange's own baseline rate, so this is as close to neutral as the market gets.
• 1 to 3 bp — a normal, healthy uptrend. Longs pay a modest premium.
• 3 to 5 bp — crowded. 33–55% annualised is a real cost and signals eager, leveraged buying.
• Above 5 bp — very crowded. Historically the zone that precedes long squeezes, where a modest price fall liquidates leveraged positions and accelerates the move.

For scale, over the 95 days on record here funding has ranged from -0.13 bp to 1.00 bp (roughly -1% to 11% a year).

Open interest, and why it has no history here

Open interest is the total value of futures positions currently open — how much money is in the game, as distinct from which way it is leaning. Rising open interest with rising price means new leveraged money entering; rising open interest with flat price means fragility building; a sharp fall usually means positions were liquidated.

The exchange only retains about 30 days of it, so this page shows the current value and a short window rather than years. That is a limit of the free source, not an omission.

What it cannot tell you

• One venue only. This is OKX's BTC perpetual contract. A single venue is not the whole market, and positioning elsewhere — other exchanges, DEXs — is invisible here.
• Extremes persist. Funding can stay crowded for weeks. It describes positioning, not timing.
• It is a fast signal in a slow dashboard. Everything else here moves over months; funding can flip within a day.

Further reading

• Binance — how perpetual funding rates are calculated
• Binance Academy — perpetual futures explained
• Investopedia — Open Interest

Hashrate: the number itself is meaningless

Hashrate is how many guesses per second the whole network makes while mining, measured in exahashes — and one exahash is a quintillion guesses. Nobody has intuition for “1,020 EH/s”, and nobody needs it. The level tells you nothing; the direction tells you a lot.

Mining costs real money in hardware and electricity. Miners only add machines if they expect the coins to be worth more than the power. So a rising hashrate is capital being committed by the people closest to the economics — revealed preference rather than opinion.

How to read it: the 30 and 60-day averages

That is what the two averages on the chart are for. When the 30-day average falls below the 60-day, hashrate is dropping rather than merely wobbling: machines are being switched off because they no longer cover their power bill. This is usually called miner capitulation, and historically those crossings clustered near price lows — the point where the least efficient sellers have already been forced out.

The reverse crossing, 30-day back above 60-day, has tended to mark the recovery.

The honest caveat: hashrate follows price. Miners buy machines after profitable periods and switch off after unprofitable ones, so this lags by weeks to months. It confirms; it does not predict.

Coins on exchanges: what the percentage means

Today 13.46% of all mined bitcoin sits in exchange wallets. On its own that figure means little — what gives it meaning is its own history: it peaked at 17.17% in 2020-03 and has fallen fairly steadily since. So the answer to “is 13.5% a lot?” is: it is near the low end of the last several years, meaning less supply is sitting somewhere it can be sold immediately than at almost any point since 2020.

Why it matters: a coin sitting in an exchange wallet can be sold in one click. A coin in self-custody generally cannot — it has to be moved first, which takes a deliberate act. So the share of supply held on exchanges is a rough measure of how much supply is positioned to sell.

A falling line means coins moving off exchanges into private storage, which is usually read as accumulation. A rising line means coins moving toward the place people sell them. Direction matters far more than level.

• Exchange wallets are identified by heuristics, not by declaration. Labelling is imperfect and providers disagree with each other.
• Custodians and ETFs blur it. Coins held for a spot ETF are not on an exchange but are not self-custodied by an individual either.
• Large moves can be internal. An exchange reshuffling its own wallets can look like a flow that never happened.

Which of these counts toward the score, and why

Miner commitment does. The score uses the ribbon spread — the 30-day mean hashrate against the 60-day, as a percentage. Tested against known turning points it read +11.1% at the 2017 top, −12.4% at the December 2018 low and +4.5% at the 2021 top: right on direction at three of six, and strongest exactly where it matters, in the negative tail. It missed the 2020 and 2022 lows, which registered mildly positive.

So it is weaker than the valuation measures, and it is in anyway, for a specific reason: it is the most independent input on the page — a correlation of 0.10 with the MVRV Z-score and 0.17 with the Mayer Multiple. Nearly everything else here is a restatement of where price sits relative to its own history. A noisy signal that measures something genuinely different is worth more in an average than a fourth clean restatement of the same thing.

Adding it visibly improved the calibration: with four signals the November 2022 bear-market low ranked in the 38th percentile and read “accumulation”. With miner behaviour included it ranks 12th and reads “deep value”, which is what that day actually was.

Coins on exchanges does not, and it was tried. Scored on its level it called the 2013 and 2017 tops maximally cheap and the March 2020 crash low maximally hot, because the share carries a secular trend — exchanges grew from nothing to 17% of supply, then shrank as self-custody and ETFs took over — and a fixed band measures that trend rather than the cycle. Scored on its 90-day change it was right at two tops and wrong at three lows and the 2021 top. Being uncorrelated is not the same as being informative, and a component that is wrong more often than right does not earn a vote. It stays here as context, where direction over time is readable and no false precision is claimed.

Further reading

• Glassnode Academy — exchange balances
• mempool.space — hashrate and difficulty, charted
• Binance Academy — what mining actually is