The BTC/ETH ratio

What the number is

The BTC/ETH ratio is the price of bitcoin divided by the price of ether. Read it as a quantity: how many ETH one BTC buys today.

It carries no information the two prices do not already contain. What it removes is the dollar. If both assets fall 20% in a week, every chart denominated in dollars says the same thing — the market went down — and the ratio says nothing at all, because nothing happened between the two. That is the point. The ratio is not a market indicator. It answers one narrow question: which of the two is currently winning against the other.

For anyone holding both, or choosing between them, that is the only question on the table.

Why the level on its own says nothing

Suppose the ratio reads 34. Is that high?

The honest answer is that the number alone cannot tell you, and the history of this particular pair makes the point better than any argument. In 2018 one bitcoin bought as few as nine ether. Through 2019 and 2020 it bought forty on a typical day and touched sixty. By 2022 and 2023 the pair had settled into a narrow band between twelve and twenty, and barely left it for two years. Then it climbed again.

There is no "normal" level here to compare against. A reading of 34 was extraordinary in 2022 and unremarkable in 2019. Anyone quoting a target for this ratio is quoting a number from a regime that has already ended at least four times.

What can be compared is position within history, and that requires two different measurements.

The first reading: where the ratio sits in its own past

The percentile answers the question "when has it been here before?" A 77th percentile means the ratio has spent 77% of its recorded days below today's value and 23% above it. Nothing is predicted; a fact is counted.

This reading is slow and blunt on purpose. It does not care about the last three months, it cares about the whole sample, and it is the only one of the two that survives a regime change — because it does not assume there is a level to return to. The number of days behind the percentile matters as much as the percentile itself: the same "4th percentile" means something different over three years than over six months.

The second reading: distance from the recent norm

The z-score answers a different question: how far the pair is from where it has been lately, measured in standard deviations of its own recent behaviour.

Two details matter here, and both are easy to get wrong.

First, the z-score on this site is not computed on the ratio. It is computed on the regression spread — the part of bitcoin's price that ether does not explain, after fitting one against the other. For this pair the fitted β is well below 1, which means the two do not move one-for-one, and a measure built on the raw ratio would quietly report the difference between them as divergence. The z-score page covers why the residual is the right input.

Second, the norm is a window, not a truth. "Far from normal" always means "far from the norm of the last N days" — 180 by default here. Shorten the window and everything looks extreme; lengthen it and nothing does.

The third reading: how long it used to take

The two readings above describe the present. The question that actually keeps a holder awake is the next one: if it is far out, how long did it historically take to come back?

That is what the divergence count in the figures at the top of this page measures. Every time the spread left the ±2σ band, the clock started; when it came back inside ±0.5σ, the clock stopped. The median of those durations is the closest thing to an honest answer: not what will happen, but how long the waiting used to last.

For BTC/ETH the typical return has historically been measured in weeks to a couple of months, and the longest single episode ran the better part of a year. A holder who treats the median as a schedule will be wrong often; a holder who has never looked at the distribution has no idea what they are signing up for.

What the ratio does not tell you

It is not a probability. An extreme percentile is a statement about the past, not a forecast. Spreads that are far out get further out regularly, and if the relationship between the two assets has changed, "far out" is simply the new level.

The two assets are not interchangeable. Bitcoin and ether answer to different things — supply schedule, staking yield, fee burn, the regulatory treatment of each, the flows into the products that hold them. A ratio has no opinion on any of this. When one of those changes materially, the series before the change describes a pair that no longer exists.

Reversion here is slow. Whatever the half-life on the pair page reads today, check it before assuming anything will snap back. For large, well-correlated assets it is routinely measured in hundreds of days, which is a different proposition from a trade that resolves in a fortnight.

One pair is not a portfolio. The ratio says which of two assets is ahead. It says nothing about whether holding either of them is sensible.

Where to look next

The live BTC/ETH page carries the chart, the current z-score, the percentile and the full list of past divergences with dates. The screener ranks every pair on the site by the same measurements, so a reading of 34 can be put next to what every other pair is doing today. And if the holding is BTC rather than the pair, the BTC page asks the rotation question directly: of everything quoted here, what is unusually cheap against bitcoin right now.