Research · Altcoins

Capital rotation from Bitcoin to altcoins

Rotation is a claim about where money went. What the public record shows is where prices went. The gap between those two statements is where most rotation commentary quietly stops being evidence.

Last reviewed 2026-09-21Source: Coin Metrics community network and market data documentation; CoinGecko global market data documentationMethod and data-availability only. No flow, return or market-cap figure is asserted.

Dataset, period and method

The dataset this page reasons about is the set of measures that are actually published: per-asset prices and market capitalisations from a market-data provider, and the on-chain and market metrics that Coin Metrics publishes through its community API. The period is the full published history of those series. The method is to state what a rotation claim asserts, then to ask which parts of it are observable in that data and which are not.

The distinction that organises the page is between prices and flows. A price is an observation: it is published, it is dated, and two people reading the same source see the same number. A flow is an inference: it is a claim that a particular quantity of capital moved from one asset to another, and it is not directly published for the crypto market as a whole. Market capitalisation is a price times a supply estimate, so a change in it is a change in an estimate of value rather than a record of money moving.

That distinction is not pedantry. It is the reason a rotation claim is harder to evidence than it looks. The observable part — that altcoins outperformed Bitcoin over a window — is a fact about relative prices. The unobservable part — that the outperformance was funded by capital leaving Bitcoin — is a causal story about flows, and the published data does not contain it.

What would count as evidence

The components of a rotation claim, whether each is observable in published data, and what would be required to support it.
Component of the claim	Observable?	What would support it	Where it breaks down
Altcoins outperformed Bitcoin over a window	Yes	Per-asset returns over a stated window from a named provider	A relative return says nothing about the direction of either asset in absolute terms
The market's value shifted away from Bitcoin	Yes, as a share	A change in Bitcoin's share of total market capitalisation over the same window	The share moves with the denominator, and the denominator depends on which assets the provider counts
Capital left Bitcoin and entered altcoins	No, not directly	Exchange flow data and stablecoin issuance can be consistent with it, but neither identifies a destination	A withdrawal from an exchange is not a purchase of a specific asset; the destination is unobserved
Bitcoin's move caused the altcoin move	No	Nothing in the published price record can establish this	A common factor — risk appetite, cost of capital — produces the same pattern without any transfer between the two

Last reviewed 2026-09-21Source: Coin Metrics community API documentation; CoinGecko global market data documentationThe observability column describes what the published data contains, not a finding about any period.

The first two rows are the part of a rotation claim that can be checked, and they are worth stating precisely because they are often skipped. A relative return over a named window from a named provider is a fact. A change in Bitcoin's share of a stated total is a fact. Neither of them is a flow, and neither of them identifies a mechanism.

The third row is where the claim usually overreaches. Exchange flow data can show that coins moved to or from venues, and stablecoin issuance can show that dollar-denominated tokens were created. Both are consistent with a rotation story, and neither establishes one. A coin withdrawn from an exchange may be held, lent, or moved to another venue; a newly issued stablecoin may fund a purchase of any asset or none. The exchange flows page documents how those addresses are identified and where the inference breaks down.

The fourth row is the one that cannot be closed with market data at all. Two assets can move in the same window because a common factor moved both, without any capital passing between them. A rise in risk appetite, a fall in the cost of capital, or a change in regulation can lift a whole market. Nothing in the price record distinguishes that from a transfer, and this page does not claim that it does.

What remains inference

What remains inference is the movement of capital itself. The published record contains prices, market capitalisations and on-chain quantities. It does not contain a ledger of who bought what with the proceeds of what, and for the crypto market as a whole no such ledger exists. A rotation narrative is therefore an interpretation placed on top of observable relative performance, and it should be labelled as one.

That does not make the interpretation worthless. It is a coherent story that fits the pattern, and it may well be part of what happened. The point is that the pattern is also consistent with explanations that involve no rotation at all, and the data cannot choose between them. A reader who wants to use the rotation framing should hold it as a hypothesis rather than a finding.

The practical consequence is that the observable measures should carry the argument. Breadth — the share of a defined universe that outperformed Bitcoin — and relative performance over a stated window are both checkable, and both are enough to describe what happened. The sequence page applies exactly that discipline to the proposed BTC → ETH → alt ordering, and the definition page sets out the measures in full.

Limitations

The first limitation is that no flow figure is asserted here. The page describes what exchange flow and stablecoin data can and cannot show, but it does not quote a net flow, because a net flow figure depends on the address-labelling heuristics behind it and could not be verified against a source.

The second is that market capitalisation is not money. A rise in an asset's capitalisation is a rise in price times an estimated supply, and it does not require a corresponding quantity of capital to have entered. A thin market can reprice a large capitalisation on modest volume. Treating a capitalisation change as a flow is the single most common error in rotation commentary, and the market cap page works through why.

The third is the sample. The number of distinct, well-documented rotation episodes is small, and they are not independent of one another: they occur within a single evolving market whose participants, venues and instruments have changed substantially across the period. Any generalisation from that record is a generalisation from a handful of observations, and it should be stated as such.

Sources and references

The data-availability statements on this page are drawn from the providers' own documentation, and the on-chain caveats from this site's analytics pages.

Community network and market data. Coin Metrics, API Access: the community endpoint, its coverage and its licence terms, which bound what a public analysis can be built on.
Global market data. CoinGecko, Crypto Global Market Data: the total market capitalisation and per-asset share fields used in the second row of the table.
Exchange flow inference. Bitcoin Data Guide, Exchange Flows: how exchange addresses are identified and where the inference breaks down.
Market capitalisation as an estimate. Bitcoin Data Guide, Market Cap Explained: why a capitalisation change is not a flow.