Market EfficiencyMarket Microstructure

Tether Flows Explain 58.8% of Bitcoin's 2017 Price Surge

Summary by Robert Gorak · Published September 4, 2026 · Last reviewed September 4, 2026

John M. Griffin and Amin Shams·2019·Working Paper (SSRN)
Sample: 9,504 hourly observations (March 1, 2017-March 31, 2018)Data: Bitcoin and Tether blockchain transaction dataPeriod: March 1, 2017-March 31, 2018

Tether is a cryptocurrency "stablecoin" pegged to the U.S. dollar and used across exchanges in place of a direct dollar-banking relationship. Griffin and Shams (2019) mapped the Bitcoin and Tether blockchains in Is Bitcoin Really Un-Tethered? using 9,504 hourly observations from March 2017 to March 2018. The top 1% of hours with the largest lagged Tether flow accounted for 58.8% of Bitcoin's compounded return over that period. A single large account on the Bitfinex exchange drove most of this flow, buying Bitcoin with Tether right after price drops.

What the Study Found

On days after Tether was printed, a 100 Bitcoin increase in lagged Tether flow raised Bitcoin's 3-hour return by 3.85 basis points. That effect rose to 8.13 basis points when Tether printing coincided with a prior negative Bitcoin return. A single account, labeled "1LSg," generated 81% of Tether flows from Bitfinex to Poloniex and Bittrex. Bitcoin fell 6% relative to its surrounding days in months with the largest Tether issuance, consistent with insufficient dollar reserves at month-end. Removing just the top 1% of high-flow hours cut Bitcoin's buy-and-hold return over the sample from 481.8% to 198.5%.

"Rather than demand from cash investors, these patterns are most consistent with the supply-based hypothesis of unbacked digital money inflating cryptocurrency prices."

Griffin and Shams (2019), Is Bitcoin Really Un-Tethered?, p. 1.

Methodology

The dataset combines the full Bitcoin and Tether blockchains with intraday pricing data from CoinAPI, Blockchain.info, and Omniexplorer.info. The main tests use 9,504 hourly observations from March 1, 2017 to March 31, 2018, drawn from a broader blockchain history beginning in 2009. Wallets are grouped into exchange identities using a same-input clustering algorithm, and regressions control for lagged returns, volatility, and their interaction.

Key Statistics

Metric Finding Context
Extreme-flow hours' share of Bitcoin's return 58.8% Top 1% of hours (95 of 9,504), test period March 2017-March 2018
Share of six other cryptocurrencies' return 64.5% Same extreme-flow hours; Dash, Ethereum Classic, Ethereum, Litecoin, Monero, Zcash
3-hour return per 100 Bitcoin lagged flow, post-authorization and negative return 8.13 basis points Table II, Panel A
2SLS estimate, 1LSg flow instrumented by round-threshold cutoff 65.44 basis points Table VI, Panel C, after Tether authorization
End-of-month abnormal Bitcoin return, high-issuance months -6% Figure 9; benchmarked against surrounding days
Net Bitcoin flow between exchanges NetBTCFlow_t = (sum BTC(PLX to BFX) - sum BTC(BFX to PLX)) + (sum BTC(BTX to BFX) - sum BTC(BFX to BTX)) Equation (1), Section II.D
Combined Tether/Bitcoin flow measure Tether/BitcoinFlow_t = (NetTetherFlow_t + NetBTCFlow_t) / 2 Equation (3); averages the flows measured on the two blockchains

1LSg Account Flows vs Other Account Flows

Measure 1LSg Account Other Accounts
Share of Bitcoin's rise tied to top 1% of flow-hours 55.0% Weak or no effect for non-1LSg flows
Return reversal after a 1% price drop (top 1% flow-hours) 52 basis points Not related to the reversal
Buying below round $500 thresholds after Tether authorization t-statistic of 3.71 Not statistically or economically significant

Why This Matters

The findings suggest that a single account, not broad investor demand, was capable of moving Bitcoin's price during the 2017 boom. That kind of concentrated influence weakens the case for building regulated financial products, such as ETFs or derivatives, directly on cryptocurrency prices. It also shows how a nominally dollar-backed digital currency can function as an unaudited source of new money entering an asset market. For investors, it is a reminder that blockchain transparency does not guarantee that on-chain price discovery reflects independent buying and selling.

Frequently Asked Questions

Griffin and Shams (2019) found that 58.8% of Bitcoin's compounded return is linked to the top 1% of hours with the largest lagged Tether/Bitcoin flow. The test period ran from March 2017 to March 2018. The same extreme-flow hours account for 64.5% of the combined return on six other major cryptocurrencies.

One deposit account, labeled "1LSg," generated 81% of Tether flows from Bitfinex to Poloniex and Bittrex. Griffin and Shams (2019) identified it using a same-input clustering algorithm that groups wallet addresses into exchange identities. The same account accounted for 47% of all Tether flows leaving Bitfinex.

Griffin and Shams (2019) found that Bitcoin fell 6% relative to surrounding days in months with the largest Tether issuance, evidence against full dollar backing. Bitfinex's Bitcoin reserve wallets also showed statistically significant outflows before month-end in those months, with a t-statistic of 3.14. Other exchanges' reserve wallets showed no such pattern.

Griffin and Shams (2019) found a 6% end-of-month Bitcoin price drop in months with the largest Tether issuance. They interpret this as evidence for the "pushed" hypothesis, that Tether is printed without full dollar backing. The alternative "pulled" hypothesis holds that Tether issuance instead reflects genuine investor demand for a dollar-pegged asset.

Reference

John M. Griffin and Amin Shams (2019). Is Bitcoin Really Un-Tethered?. Working Paper (SSRN).

Read the full paper

Cite this summary

Gorak, R. (2026). Tether Flows Explain 58.8% of Bitcoin's 2017 Price Surge. Tradicted. https://www.tradicted.com/research/griffin-tether-2019/