From 2023 through Q1 2026, aggregate stablecoin float growth led BTC price by roughly 4-6 weeks with a correlation coefficient persistently above 0.7. The relationship was tight enough that several desks used stablecoin supply expansion as a leading indicator for risk-on positioning.
In Q2 2026, the correlation collapsed to 0.18. The relationship broke. Worth understanding why, because what replaced it as the leading indicator is different.
What the old relationship was
The thesis: new dollars entering stablecoins represent dry powder waiting to deploy into BTC and major alts. The lag is the time it takes for fresh stablecoin holders to size positions, KYC into venues, and execute. Empirically the lag was 4-6 weeks across the 2023-2025 period.
The mechanism made sense and the data backed it up. Stablecoin supply went up. A few weeks later, BTC went up.
What broke it in Q2
Two structural changes:
Stablecoin demand decoupled from speculation. As covered separately, USDT's Q2 growth was concentrated on Tron, driven by emerging-market currency substitution. This stablecoin flow is not waiting to buy BTC. It's substituting for local currency. The supply growth captures real economic demand for offshore dollars, not pending crypto buys.
ETF flow replaced stablecoin float as the dominant marginal buyer. Q2 ETF net flows were +$6.8B against BTC's ~$25B market cap addition. ETF buyers don't hold stablecoins as a staging asset — they buy directly from brokerage cash. The marginal BTC demand has moved upstream of crypto-native rails entirely.
The new leading indicator
If stablecoin float is no longer the cleanest leading indicator, what is?
ETF creation/redemption flow. Daily IBIT, FBTC, BITB creation activity now leads spot BTC price by 1-3 days. The lag is short because ETF creations require AP firms to source spot BTC, which directly hits exchange inventory. Correlation coefficient against next-week BTC price: 0.61 in Q2.
CME basis term structure. Front-month CME basis annualized has been a clean read on institutional positioning intent. Basis above 8% has correlated with near-term spot strength; below 4% has correlated with chop or weakness. Q2 correlation: 0.54.
Coinbase premium. Coinbase BTC price relative to Binance BTC price as a proxy for US institutional bid. The premium going positive correlates with ETF buying activity. Q2 correlation against forward BTC return: 0.47.
The aggregate of these three has correlation ~0.71 against forward BTC price, recovering most of what the stablecoin-float indicator used to provide alone.
What this changes for desk positioning
If you were running a model that used stablecoin supply as a primary input, Q2 was a quiet alpha decay event. The signal degraded slowly through Q1 and broke hard in Q2 — not a single-week structural break, but a gradual drift.
Three adjustments worth making:
Replace stablecoin float with ETF flow as the primary risk-on indicator. Daily ETF flow data is published with a one-day lag. It's a cleaner read than weekly stablecoin supply changes.
Use stablecoin metrics for emerging-market crypto-adjacent risk. USDT supply concentration on Tron is a useful read on EM dollar demand. That's its own indicator now, not a BTC proxy.
Watch for re-coupling. Stablecoin float and BTC price could re-correlate if ETF flow saturates or if a major shift moves stablecoin holders back into BTC accumulation mode. Run the correlation on a rolling 60-day window.
What this doesn't mean
The decoupling doesn't mean stablecoins are irrelevant to BTC. It means the specific lead-lag relationship that held 2023-2025 doesn't hold in 2026. Stablecoins still settle the majority of crypto trades, still provide on-chain dollar liquidity, and still expand or contract based on broader risk appetite. They just no longer lead BTC price the way they did.
Bottom line
Stablecoin supply growth as a BTC leading indicator is broken. ETF flow + CME basis + Coinbase premium together replace most of the signal. The model adjustment is required.
If you're still running stablecoin float in your weekly review without context, you're tracking a relationship that no longer exists. Rebuild the indicator stack around what's actually moving spot now.