The long-term holder (LTH) vs short-term holder (STH) framework is the foundational cohort-analysis approach in contemporary on-chain analytics. It partitions Bitcoin's UTXO set by age: coins held 155+ days are LTH supply; coins held less are STH supply. Glassnode calibrated the 155-day threshold empirically as the point at which spend probability drops sharply. The framework distinguishes the conviction cohort (LTH) — patient, accumulation-oriented — from the reactive cohort (STH) — responsive to current price action, often the marginal buyer at peaks and the marginal seller during corrections. LTH supply growth marks accumulation; LTH distribution marks late-cycle dynamics; LTH realized price acts as structural support; LTH SOPR > 2 has recurred at cycle tops. Every cohort-specific variant of the Cluster 1 valuation metrics and the broader HODL waves decomposition build on this foundation.
Why this note matters
The LTH/STH cohort framework is load-bearing for the on-chain section in three respects:
- It is the foundation for every cohort-specific metric variant. The Cluster 1 valuation notes (Realized price, MVRV ratio, NUPL, SOPR) all reference LTH and STH variants as operationally more informative than the aggregate. Without the 155-day threshold convention, “LTH MVRV” and “LTH realized price” have no operational meaning.
- It is the most reliable single behavioral signal in on-chain analysis. Across the 2013, 2017, 2021, and 2024-2025 cycles, LTH supply dynamics have produced more reliable cycle-positioning signals than any other on-chain metric: accumulation through bear markets, distribution into bull-market peaks, and LTH realized price as structural support.
- It is the gateway to the broader cohort framework. The HODL waves note generalizes to a multi-age-band decomposition. The binary framework here is the operational simplification and the prerequisite for engaging the richer supply-by-age machinery.
The 155-day threshold is empirical and somewhat arbitrary, but the framework’s utility has been substantial enough to become standard across Glassnode, Coin Metrics, Checkonchain, and the broader on-chain literature.
What the framework defines
The two cohorts.
- Long-term holder (LTH) supply: UTXOs that have not moved on-chain in 155 or more days. Conceptually: the conviction cohort — holders who have demonstrated, through 155 days of non-movement, that they are not reactive to short-term price action.
- Short-term holder (STH) supply: UTXOs that have moved on-chain within the last 155 days. Conceptually: the reactive cohort — holders whose coins are still in active circulation and who are more likely to be responsive to current market conditions.
The 155-day calibration. Glassnode’s empirical research determined that the spend probability of a UTXO drops sharply once it has been held for approximately 155 days. The threshold marks an empirical inflection point: coins held longer have historically been much less likely to be spent on any given day. The threshold is not theoretically derived; it is the level at which holder spending behavior empirically diverges between the two cohort regimes.
What the cohort assignment is not. The LTH/STH partition is a property of UTXOs, not of holders. A single individual can hold both LTH UTXOs (older holdings) and STH UTXOs (recent purchases). When a holder receives new Bitcoin (whether by purchase, mining, or transfer-in), those UTXOs start as STH supply and migrate to LTH after 155 days of non-movement. When a holder spends an LTH UTXO, the resulting change output becomes STH supply because it was created today.
The migration dynamic. STH supply continuously migrates to LTH supply as time passes (155 days after each UTXO’s creation). Conversely, LTH supply migrates to STH supply when LTH UTXOs are spent. The net direction of migration is the operationally informative signal:
- Net STH → LTH migration: more UTXOs aging into LTH status than LTH UTXOs being spent. Signals patient holding; characteristic of accumulation phases and early-cycle bull markets.
- Net LTH → STH migration: more LTH UTXOs being spent than STH UTXOs aging into LTH status. Signals long-term-holder distribution; characteristic of late-cycle bull markets.
How the cohort dynamics work mechanically
LTH supply growth in bear markets. During bear markets and accumulation phases:
- Spot price is depressed; LTH cohort generally refuses to spend at depressed prices
- STH purchases at low prices age into LTH status after 155 days
- Net effect: LTH supply grows substantially during bear markets
The pattern is one of the most reliable on-chain signatures. The 2018-2019 bear market saw LTH supply grow from approximately 11.5M BTC to approximately 13.5M BTC; the 2022 bear market saw LTH supply grow from approximately 13.8M BTC to approximately 14.7M BTC. The accumulation is structurally visible in the LTH supply curve.
LTH supply distribution in late-cycle bull markets. During late-cycle dynamics:
- Spot price reaches elevated levels; LTH cohort begins taking realized profit
- LTH UTXOs are spent (creating new STH supply); STH purchases at high prices haven’t yet aged into LTH
- Net effect: LTH supply declines as the cycle approaches its peak
LTH supply has historically peaked roughly 6-12 months before cycle peaks and declined through the late-bull-market distribution phase. The 2017 peak saw LTH supply decline from ~13M BTC to ~12.3M BTC in the year leading up to and through the peak; the 2021 peak saw similar dynamics with LTH supply declining from ~13.8M BTC to ~13.2M BTC through the April-November 2021 distribution window.
The supply-shock dynamic. When LTH supply growth accelerates while available STH supply contracts, the resulting supply tightness has historically preceded major bull-market initiations. The “supply shock” framing — coined variously by Willy Woo, James Check, and others — captures this: LTH accumulation removes coins from the actively-traded float, creating a setup where modest incremental demand can produce outsized price impact.
Asymmetric reactive-cohort behavior. STH supply behaves more symmetrically around price action than LTH supply:
- STH net inflows (purchases) increase during bull markets as new participants enter
- STH net outflows (sales) increase during corrections as recent buyers take losses or modest profits
- STH supply oscillates with shorter cycle dynamics than LTH supply
The STH cohort is closer to the marginal buyer/seller in cycle dynamics; LTH cohort is closer to the structural-holder reserve.
What the framework tells you
The primary cycle-positioning signals.
| LTH supply behavior | Cycle phase | Operational reading |
|---|---|---|
| Sustained growth (multi-month) | Accumulation; bear-to-early-bull transition | Patient long-term cohort building positions |
| Peak followed by sustained decline | Late-cycle distribution | LTH cohort taking realized profit; cycle-top risk elevated |
| Cliff-edge decline (rapid drop) | Climax distribution | Most aggressive distribution signal; near or at cycle peak |
| Resumed growth after decline | Bear-market initiation through accumulation | New cycle setup |
The pattern has been reliable across multiple cycles. Specific LTH supply level inflections have varied across cycles, but the directional pattern (growth → peak → decline → growth) has been consistent.
LTH realized price as structural support. The cohort-specific realized price for LTH UTXOs — computed using the same methodology as aggregate Realized price but restricted to LTH supply — acts as a stronger structural support level than aggregate realized price. The reasoning: LTH UTXOs are by definition coins held through 155+ days of volatility without being spent; the holders are demonstrably not reactive to ordinary price movements. The LTH realized price represents the cost basis at which the conviction cohort is willing to hold; significant breaks below it historically mark deep capitulation.
STH realized price as resistance. The cohort-specific realized price for STH UTXOs — average cost basis of coins held less than 155 days — acts as resistance during bear-market rallies. Reasoning: STH coins are recent acquisitions; when spot price recovers to STH realized price, recent buyers are at break-even and may sell to exit. The level is a recurring rejection point in bear-market rallies until enough STH supply has either capitulated (spent at loss) or aged into LTH status.
LTH SOPR > 2 as late-cycle signal. The cohort-specific SOPR for LTH UTXOs — the realized profit ratio on LTH-cohort spent outputs — provides a refined cycle-top signal. LTH-SOPR > 2 means long-term holders are realizing at least 2× profits on the coins they’re spending. The signal has occurred at the 2017 cycle peak, the April 2021 intra-cycle peak, and the November 2021 cycle peak. The threshold has attenuated across cycles consistent with Diminishing returns thesis.
LTH NUPL and LTH MVRV. The cohort-specific variants of the NUPL and MVRV ratio valuation metrics inherit the cohort framework. LTH NUPL is operationally more informative than aggregate NUPL for cycle-top calling; LTH MVRV extreme readings have been the most reliable cycle-positioning signal in the contemporary on-chain framework.
Behavioral interpretation. The framework’s underlying behavioral claim: holders who have held through 155+ days of price volatility have demonstrably different psychology from holders whose coins were recently acquired. The LTH cohort accumulates patiently, distributes only at strong profit, and acts as the structural-holder reserve. The STH cohort is closer to the marginal-trader population whose behavior tracks current price action. The framework operationalizes this behavioral distinction through the simple age-threshold partition.
Empirical track record
Cycle-peak LTH supply dynamics.
| Cycle | LTH supply peak timing | LTH supply decline to spot peak | Approximate LTH supply at cycle peak |
|---|---|---|---|
| 2017 peak (Dec 2017) | ~mid-2017 | ~700K BTC | ~12.3M BTC |
| 2021 peak (Apr 2021) | ~late-2020 | ~500K BTC | ~13.5M BTC |
| 2021 peak (Nov 2021) | ~mid-2021 | ~600K BTC | ~13.2M BTC |
| 2024-2025 peak (Aug 2025) | ~late 2024–early 2025 (LTH distribution into the top) | attenuated | attenuated |
The pattern: LTH supply peaks 6-12 months before spot peaks and declines as distribution accelerates. The magnitudes have attenuated cycle-over-cycle consistent with Diminishing returns thesis.
Cycle-bottom LTH supply dynamics. Conversely, LTH supply has continued growing through major bear markets and reached its growth-rate peak near cycle bottoms:
| Cycle | LTH supply at cycle bottom | LTH supply growth in preceding bear |
|---|---|---|
| 2018 bottom | ~13.5M BTC | ~2M BTC growth from 2017 peak |
| 2022 bottom | ~14.7M BTC | ~900K BTC growth from 2021 peak |
The growth-through-bear-market pattern is one of the most reliable on-chain signatures. It reflects the conviction cohort’s patient accumulation through depressed prices.
LTH realized price as support. Historical episodes where LTH realized price has acted as support:
- 2019 H1: spot price tested and reversed above LTH realized price levels through the bear-market accumulation
- 2023 H1: spot price tested LTH realized price (then around $19,000-22,000) and reversed substantially upward
- Various intra-cycle pullbacks in 2024-2025 have respected LTH realized price as support
The pattern is not deterministic — sufficiently severe macro shocks can break the support — but the historical regularity has been substantial.
Cross-validation with Cluster 1 metrics. LTH cohort signals typically align with extreme readings in aggregate MVRV, NUPL, and SOPR. When LTH supply is declining, LTH SOPR > 2, LTH NUPL > 0.75, and aggregate MVRV Z-score > 7 simultaneously, the cycle-top signal is much stronger than any single metric provides. The systematic frameworks (Check, Ryan) deploy this cross-validation routinely.
The 155-day threshold’s empirical stability. The 155-day calibration has held remarkably well across cycles. Glassnode’s research on spend-probability inflection points has revisited the calibration periodically; the threshold has migrated only slightly (some refinements suggest 150-180 days depending on cohort definition). The robustness of the threshold across diverse market regimes is part of why the framework has become standard.
Limitations
The 155-day threshold is empirical, not theoretical. No mechanism in the Bitcoin protocol enforces or even motivates 155 days as a cohort boundary. The threshold was calibrated by Glassnode’s empirical research; alternative thresholds (90 days, 1 year, 6 months) would produce different cohort assignments and different operational signals. The standard convention has stuck because it has worked, not because it is theoretically derived.
Holder vs UTXO identity confound. The framework partitions UTXOs, not holders. A holder with both old (LTH) and new (STH) UTXOs is split across cohorts. Behavioral interpretation requires the assumption that aggregate UTXO-level dynamics approximately reflect aggregate holder-level dynamics — a reasonable but not perfect approximation. For high-precision research, entity-clustering frameworks (multiple UTXOs assigned to the same entity) are more rigorous.
Custodial and ETF distortion (2024+). Bitcoin held in custodial-exchange wallets and ETFs shows up at the wallet level. The UTXOs that constitute these custodial holdings have their own age dynamics that don’t reflect the beneficial owners’ holding behavior. ETF creation/redemption flows produce LTH-to-STH or STH-to-LTH migrations that reflect operational mechanics rather than holder economic decisions. The framework is increasingly muddied as more Bitcoin sits in custodial structures.
The cohort dynamics can be gamed. Sophisticated holders can perform cohort-positioning maneuvers — moving coins just before they age into LTH status, or splitting holdings to maintain specific cohort allocations. These maneuvers are rare in aggregate but exist; the framework’s signals can be marginally distorted by intentional cohort engineering.
STH cohort blends very-different sub-populations. The STH cohort (coins held less than 155 days) blends day-traders, intra-week holders, monthly buyers, and 3-month-and-counting buyers. The blend is heterogeneous; aggregate STH metrics average behaviors that are operationally quite distinct. Some frameworks decompose STH into shorter-age bands (HODL waves) for finer signals.
Cycle attenuation affects calibration. The magnitudes of LTH supply swings (peak-to-trough) have declined across cycles. The 2017 cycle saw ~700K BTC LTH-to-STH migration through distribution; the 2021 peak saw ~500-600K BTC. Future cycles will likely see further attenuation. The directional pattern (growth → peak → decline) remains; the specific magnitudes need ongoing recalibration.
Net-position signal can lag. LTH supply changes are aggregated over UTXO-level events. The smoothness of the LTH supply curve means short-term shifts (a few days of unusual movement) don’t appear strongly. The metric is operationally a multi-week-to-multi-month signal, not a real-time positioning tool.
The framework assumes psychological stability across cohorts. “Conviction cohort” and “reactive cohort” are operational labels for the empirical behaviors of UTXO populations. The labels assume the underlying holder psychology is approximately stable — that LTH UTXOs are held by holders with conviction-cohort psychology, not just by holders who have been too inattentive to move their coins. The assumption is largely defensible but has limits (lost coins, abandoned wallets, key-loss situations all show up as LTH supply that doesn’t reflect active conviction).
Counter-arguments and tensions
”155 days is just curve-fitted to historical data”
The argument: The 155-day threshold was calibrated by Glassnode on historical data. Any threshold-based framework risks overfitting: the framework looks robust on the calibration sample because the threshold was chosen to make it look robust. Future cycles may produce regime changes that invalidate the calibration.
Response: Partially right. The threshold is empirical, and out-of-sample performance is the test. The framework has held across the 2018-2019, 2020-2022, and 2024-2025 cycles since the calibration was first published — these are genuinely out-of-sample tests. The threshold has been remarkably stable across diverse market conditions. The honest reading is that the framework has more analytical content than pure overfitting but the threshold-calibration question deserves continued attention as new data accumulates.
The custodial-cohort blindness problem
The argument: As more Bitcoin sits in ETFs, corporate treasuries, and custodial exchanges, the UTXO-level cohort dynamics reflect custodial operational behavior rather than beneficial-owner economic decisions. The framework’s signal is increasingly contaminated. Pre-2024 calibrations may not apply to post-2024 regimes.
Response: Substantial concern. The framework is genuinely affected by the post-2024 custodial shift. ETF UTXOs aged into LTH status reflect ETF-creation activity, not retail accumulation. The mitigation: cohort-restricted variants that exclude known custodial wallets (where wallet attribution is available) can recover most of the signal. The framework needs adaptation, not abandonment, but ongoing calibration is required.
”LTH supply growth is just time passing”
The argument: Any non-zero amount of held Bitcoin will eventually age into LTH status as long as it isn’t spent. LTH supply growth during accumulation phases is mechanically guaranteed; the metric is measuring time passing rather than meaningful behavior.
Response: Partially right but understated. Yes, time-passing mechanically produces LTH supply growth in the absence of spending. But the rate of LTH supply growth, and the magnitude of net LTH-to-STH migration during distribution phases, both carry behavioral information beyond pure aging. The metric’s operational value is in deviations from the mechanical baseline: rapid LTH growth signals stronger-than-baseline accumulation; rapid LTH decline signals stronger-than-baseline distribution.
The lost-coin contamination
The argument: A substantial fraction of Bitcoin’s supply consists of genuinely lost coins (forgotten keys, dead holders without inheritance plans, abandoned wallets). These coins show up as LTH supply but reflect no active holder psychology — they’re not “conviction holders,” they’re just unreachable. The framework systematically overweights the conviction-cohort signal because it can’t distinguish active conviction from inability to act.
Response: Real concern but stable. The lost-coin overhang affects the level of LTH supply but not its changes — lost coins don’t move on-chain, so their contribution to LTH supply is constant. The framework’s operational use is based on changes in LTH supply (growth and decline), not on absolute level. The lost-coin overhang is implicitly calibrated into the threshold and the historical patterns.
”LTH realized price as support is just historical regularity”
The argument: The pattern of LTH realized price acting as structural support is empirical regularity across the 2018-2019, 2023, and other test cases — but the sample is small and the support-resistance dynamics aren’t mechanically guaranteed. Sufficiently severe macro shocks could break LTH realized price as a support level.
Response: Right. The support is empirical, not structural. The mechanism is plausible (LTH cohort has cost basis at LTH realized price; willingness to capitulate below this level requires strong macro pressure), but the future regularity is not guaranteed. Users should treat LTH realized price as a probabilistically meaningful level, not a hard floor. The framework’s predictions should be paired with macro context (see Bitcoin and global liquidity, Bitcoin and the ISM PMI cycle).
Binary cohort vs continuous age framework
The argument: The binary LTH/STH partition aggregates heterogeneous sub-populations within each cohort. A 160-day-old UTXO is grouped with a 5-year-old UTXO under “LTH”; a 1-day-old UTXO is grouped with a 150-day-old UTXO under “STH.” The cohort framework loses the richer information available from age-banded decomposition (HODL waves).
Response: Substantively right at the level of analytical refinement. The binary framework is an operational simplification; HODL waves is the richer continuous-age decomposition. The honest reading is that binary LTH/STH is the operational gateway and HODL waves is the deployment-ready advanced framework. Both have their roles: binary for quick cycle-positioning communication; multi-band for refined analysis.
”STH cohort signals are too noisy”
The argument: The STH cohort blends day-traders, weekly buyers, monthly accumulators, and almost-LTH (140-day) holders. STH-level metrics aggregate fundamentally different behaviors. The cohort definition is too coarse for operational use.
Response: Right. STH metrics are operationally noisier than LTH metrics for this reason. The systematic frameworks (Check, Ryan) lean more heavily on LTH metrics than STH metrics for cycle-positioning. STH-level metrics are most useful as confirmation signals (STH cohort distress confirms broader cycle stress) rather than as primary signals. The HODL waves framework provides STH sub-decomposition for users who need finer signals.
Cycle attenuation breaks the threshold magnitudes
The argument: LTH supply swings have attenuated across cycles. The 2017 distribution involved ~700K BTC LTH-to-STH; the 2021 cycle saw ~500-600K BTC; future cycles will see less. The framework’s specific thresholds (what counts as “significant LTH decline”) are migrating downward and require recalibration each cycle.
Response: Real. The directional pattern is more stable than the magnitudes. Users should focus on direction-of-change and acceleration-of-change rather than absolute thresholds. The framework remains useful but the operational calibration is dynamic.
Open questions for further development
- Should the 155-day threshold be recalibrated for the post-ETF regime? ETF flows have introduced new cohort dynamics. Specific threshold adjustments — perhaps custodial-aware cohort definitions — would strengthen the framework.
- What is the appropriate way to handle lost-coin contamination? The lost-coin overhang is constant in absolute terms but distorts absolute-level analysis. Frameworks that explicitly model the lost-coin fraction may produce more reliable cohort metrics.
- How does the framework integrate with entity-clustering? Multiple UTXOs belonging to the same entity should arguably be aggregated for cohort analysis. Entity-clustering frameworks (Glassnode entities, Chainalysis attributions) provide partial paths; the integration with cohort analysis is an active research direction.
- Should STH be sub-decomposed? The STH cohort blends very-different sub-populations. A binary STH-A (less than 90 days) and STH-B (90-155 days) sub-partition may produce more refined signals; HODL waves provides the continuous-age decomposition that fully addresses this.
- What is the cycle-attenuation calibration for future cycles? The magnitudes of LTH supply swings have declined across cycles. Predicting future swing magnitudes — and recalibrating operational thresholds accordingly — is an open analytical challenge.
- How does the framework engage hyperinflation or major-fiat-regime-change scenarios? Holder behavior may shift fundamentally if USD-denominated valuation becomes uninformative; the cohort framework may need re-grounding in alternative denominations.
- Can the cohort framework be derived from first principles rather than empirically calibrated? A behavioral model of holder spending probability as a function of holding age would ground the 155-day threshold in theory rather than empirics; current work is largely empirical.
Canonical sources for this note
Primary framework sources
- Glassnode research, various pieces introducing and refining the LTH/STH cohort framework and the 155-day threshold calibration — the canonical source
- Coin Metrics, various analyses of UTXO age-based cohort dynamics — adjacent treatment
- Checkonchain platform — James Check’s analytical framework leaning heavily on the LTH cohort signals
- David Puell, various adjacent analyses
Practitioner literature
- James Check, extensive Glassnode Week On-Chain newsletters during the 2020-2023 tenure — applied LTH/STH analysis across multiple cycles
- James Check, ongoing Checkonchain platform analysis 2024+ — refined cohort framework for current conditions
- Ryan (On-Chain Mind), various video analyses applying LTH/STH framework accessibly
- Willy Woo, various pieces on supply-shock dynamics related to LTH accumulation
Adjacent on-chain literature
- HODL waves framework — the continuous-age generalization of the binary LTH/STH partition; see HODL waves
- Coin Days Destroyed — complementary velocity-and-age metric; see Coin Days Destroyed
- Various analyses of entity-clustering and its integration with cohort analysis
Critical perspectives
- Various engagements with the threshold-calibration question
- Critiques of the binary partition as too coarse
- Within-Bitcoin debates about custodial-cohort contamination of cohort signals
Related notes
- On-chain analytics and market psychology — sub-MOC parent
- Realized price — definitional foundation; produces LTH realized price and STH realized price variants
- MVRV ratio — produces LTH MVRV and STH MVRV variants
- NUPL — produces LTH NUPL and STH NUPL variants
- SOPR — produces LTH-SOPR and STH-SOPR variants
- HODL waves — continuous-age generalization of the binary cohort framework
- Coin Days Destroyed — complementary cohort-behavior metric; supply-weighted velocity
- Whale behavior — entity-size cohort framework; complementary partition
- Exchange flows — custodial-flow framework
- Miner flows — miner-cohort framework
- Sentiment indicators — off-chain sentiment proxies
- Psychological phases of the market cycle — synthesis where LTH/STH dynamics mark phase transitions
- Using on-chain data for macro positioning — operational bridge to macro frameworks
- The Power Law model — longer-horizon trajectory framework
- Four-year halving cycles — cycle structure cohort dynamics characterize
- Diminishing returns thesis — cycle-over-cycle attenuation framework cohort swings empirically demonstrate
- Bitcoin and global liquidity — macro framework
- Monetization S-curve — adoption framework cohort accumulation empirically operationalizes
- Bitcoin fixed supply and issuance schedule — supply foundation
- The halving - Mechanism — schedule affecting cohort dynamics
- Portfolio approaches to Bitcoin — practical allocation framework cohort signals inform
- James Check — primary contemporary anchor; cohort framework developer
- Ryan - On-Chain Mind — adjacent contemporary anchor
- Dylan LeClair — adjacent on-chain voice
- Giovanni Santostasi — Power Law modeler; adjacent
- Plan B — S2F framework (engaged critically)