Long-term vs short-term holders: how on-chain analysts split the market

Why sort holders by how long coins sit still?#

Every Bitcoin output is stamped with the block that created it, so each coin has an age: the time since it last moved, not since it was first mined ^[3]. Glassnode’s analysts reasoned that coin age is a window into behaviour. Coins that change hands frequently probably belong to traders and newcomers; coins that have not moved for months probably belong to people holding through price swings ^[2].

In March 2020 they used that idea to define two groups — short-term holders (STH) and long-term holders (LTH) — and to build versions of popular metrics for each group ^{[2] [1]}. If you have read about HODL waves, this is the same coin-age data cut into two pieces instead of many bands.

Where does the 155-day threshold come from?#

The number is empirical. Glassnode looked at the history of Bitcoin outputs and calculated, for each age, the probability that an output would be spent within the next few days or weeks. Young outputs are very likely to move again soon; the probability falls quickly with age and then flattens out somewhere between about 100 and 200 days ^[2].

To pick an exact cut-off, they compared the slopes of those probability curves for several windows (7 to 120 days). In each case the point where the curve settled was at 155 days, so 155 days became the minimum age for long-term-holder coins ^[2]. Note what this is: a fitted threshold from one data provider’s analysis of past behaviour. It is not written into Bitcoin, and a different method could produce a different number.

How the threshold was derived

How is a holder classified as long-term or short-term?#

The first versions classified individual outputs: an output younger than 155 days counted as STH supply, anything older as LTH supply ^[2]. Glassnode’s current supply metrics go further in three ways ^[1]:

• Entity adjustment. Addresses believed to belong to the same owner are grouped into one entity using heuristics and clustering — an estimate, not a certainty.
• Volume-weighted age. The entity’s whole balance is classified by its average purchase date, weighted by amount, rather than coin by coin.
• Smoothing. Instead of a hard switch at day 155, a smooth transition centred on 155 days with a width of 10 days stops sudden jumps when a large balance crosses the line.
• Exchanges removed. Coins held in exchange wallets are excluded, so LTH supply plus STH supply plus exchange balances equals the circulating supply.

Both entities bought exactly the same kind of coin today. One purchase is absorbed into an old balance, the other dominates a small one — the same logic Glassnode describes for its 10% and 90% cases ^[1].

How do long-term and short-term holders differ?#

Descriptions summarise Glassnode’s metric guides; they are generalisations about coins, not facts about any individual owner.

One mechanical point explains a lot of chart patterns. When a long-term holder spends coins, their accumulated age is reset: the coins become young again and count as short-term supply ^[1]. So STH supply tends to rise when older coins are being sold or moved, and LTH supply tends to rise in quiet periods when coins simply age past 155 days. Glassnode’s guide associates those swings with bull and bear phases; that is an interpretation of a few cycles, not a guarantee.

Which metrics use the STH and LTH split?#

Two families are most common. STH-MVRV and LTH-MVRV divide market value by realized value for each cohort’s coins only ^{[4] [5]} — see realized price and cost basis for the building blocks. STH-SOPR and LTH-SOPR take the SOPR idea — the price when an output was spent divided by the price when it was created — and apply it to each cohort ^{[6] [7]}.

Glassnode notes that past macro tops came with LTH-SOPR above 10 and past bottoms with values at or below 0.6 ^[7]. Those are descriptions of a handful of cycles that some analysts use as reference levels — not thresholds that must repeat.

What are the limits of the holder split?#

The split is only as good as the age data under it, and age resets whenever coins move. If you send your own coins to a new wallet and the provider fails to link the two wallets, your “long-term” coins become “short-term” overnight. Glassnode identifies entities using heuristics and clustering algorithms ^[1]. That is an estimate, so by our reading it can sometimes group unrelated owners together or miss a link.

Lost coins are another distortion. Coins older than five years rarely move and are often assumed to be lost or out of circulation ^[8]; they still count as long-term supply and grow older forever. Finally, the cohorts describe coins, not people: Glassnode’s guides call LTHs “smart money” and STHs “weak hands”, but those are characterisations, not things the ledger can show.

What mistakes do beginners make here?#

• Thinking 155 days is part of Bitcoin

  It is a threshold chosen by a data provider from historical spending patterns. The protocol knows nothing about it.

• Reading cohorts as people

  A single person can hold both young and old coins; an exchange may hold coins for thousands of users. The split classifies balances, not personalities.

• Forgetting that moving coins resets age

  Self-transfers and wallet upgrades can turn long-term coins into short-term coins if the provider does not link the addresses.

• Using LTH-SOPR levels as triggers

  Values above 10 or below 0.6 coincided with a few past turning points. A small number of cycles cannot establish a rule.

Frequently asked questions#

The bottom line#

The long-term versus short-term split is a useful lens: it separates coins that move often from coins that sit still, and it powers cohort versions of MVRV and SOPR. It rests on a statistical threshold, address clustering and the assumption that age reflects behaviour — so treat it as an estimate.

To see the full age distribution rather than two buckets, read the HODL waves profile, then check the limits of on-chain data before relying on any cohort chart.

Sources#