Bitcoin's Daily RSI Just Did Something It Has Never Done in a Bear Market.

By Josh Molnar · August 2026 · 8 min read

On August 21, Bitcoin's daily RSI hit 85.9. I went back through 13 years of data to find every time it ran this hot. There are exactly 14, and what they share is not what you think.

RSI, the relative strength index, scores the speed of recent gains against recent losses on a 0 to 100 scale, and above 85 is rare air. Bitcoin's daily RSI has closed there in only about 5% of all days since 2012. Finding the distinct episodes takes one honest step. Overheated readings cluster, so consecutive episodes within 90 days are chained into one period and only the hottest print of each is kept, which stops the same crash being counted twice. That leaves 14 distinct overheated periods since 2012, plus the one that began on August 21.

The part the bears get wrong

Overbought was not a sell signal. The 14 prior periods resolved at higher prices 90 days later more often than not, with a median gain of about 14%, and in the modern era, 2019 onward, it is 6 for 6 higher with a median gain of 29%. April 2019 printed 88.7 and price was 117% higher ninety days on. January 2023 printed 89.3 at the ignition of the entire last bull run. Momentum this extreme requires enormous real buying, and enormous real buying mostly happens in strong markets. Anyone who shorted purely because RSI crossed 85 has been run over for a decade.

The law. 14 for 14, the drawdown always came.

Now the other side, and it is the study's whole finding. Every single one of the 14 periods, the winners included, was hit with a peak to trough drawdown of at least 21% within 90 days of the hot print, measured on intraday wicks. Fourteen for fourteen. No exceptions in 13 years. The median fall was 36% across the whole era, and 23% in the calmer modern era, where it is still 6 for 6.

Even the biggest winners paid first. January 2023 dipped 23% peak to trough, including the March banking-crisis flush, before its +46% resolution. October 2023 dipped 22% inside the window before the ETF run. February 2024, in full ETF mania, saw price 23% below its March peak at the April low. And the gentlest case in the entire record, August 2020, still fell 21.2% before the great 2020 bull run paid it back. That 21% is the best outcome anyone who bought an 85+ print ever got.

Overbought does not pick a direction. It guarantees turbulence, and it has never once been wrong about that.

The first one ever inside a bear market

Here is what makes the current episode genuinely new. All 14 prior periods happened in bull markets, or in one case right after a confirmed bottom, October 2015, which still drew down 41% on wicks before the trend carried. Zero happened inside a bear market. August 21, 2026 is the first.

The closest cousins the record offers are the overheated summer rallies of 2014 and 2018, which pushed the daily RSI to 83.5 and 77.2, just short of the 85 bar. Both rolled over within 90 days, fell 24% to 29%, and went on to make new bear market lows. In bull markets the drawdown was a reset. In bear markets it was the trap springing. There is no third script in the data where the drawdown simply never arrives.

Where we stand, and what the law implies

We are on day 7 of the current 90-day window. The deepest shakeout so far is 8%, the August 21 flush itself, against a 21% historical floor. From this rally's high near $81,455, the gentlest repeat in the record touches about $64,000, and the modern-era median lands near $63,000. Almost the entire window is still ahead.

And the law does not care whether the trend breaks or how high price runs first. Every one of the 14, full bull markets included, took the shakeout anyway, no matter what price did in between. If this is the bear trap, history says the dip does not stop at the shallow end. If this is a new bull market, every ignition in the record still shook out 20% or more before running. Either way, patience got a better price than chasing, every single time it has been measurable.

The honest limits

Fourteen periods is fourteen periods, and a perfect record on a sample of 14 is a base rate, not a law of physics. The old-era drawdowns of 40% to 80% belong to a wilder Bitcoin than today's, which is why the modern-era median of 23% is quoted alongside. The deduplication choice is a judgment call we disclose because without it the record would double-count crashes and overstate itself. The law says nothing about timing inside the window, day 8 and day 88 both count. And if this window closes without a 21% fall, the law takes its first loss on record and we will publish that scoreboard ourselves.

The full study, all 14 periods in one table with the methodology and what would break the law, is free: The Turbulence Law.

Education, not financial advice.

Common questions

What does it mean that Bitcoin's daily RSI crossed 85?

The daily RSI above 85 is rare air, reached on only about 5% of days since 2012 and almost always in full bull markets. It means recent buying has been extreme relative to selling. Historically it did not reliably mark tops, but every one of the 14 distinct periods that reached it was hit with a drawdown of at least 21% within 90 days.

Is a high RSI a sell signal for Bitcoin?

Not by itself. The 14 overheated periods since 2012 resolved at higher prices 90 days later more often than not, and in the modern era it is 6 for 6 higher with a median gain of 29%. What the reading did predict, every single time, is turbulence: a peak to trough drawdown of at least 21% somewhere inside the following 90 days, even in the periods that ended much higher.

Has Bitcoin's daily RSI ever hit 85 in a bear market before?

No. All 14 prior overheated periods happened in bull markets or immediately after a confirmed bottom. The August 21, 2026 print of 85.9 is the first ever inside a bear market. The closest historical cousins are the overheated summer rallies of 2014 and 2018, which peaked just short of 85, fell 24% to 29% within 90 days, and went on to make new lows.

How big a Bitcoin drawdown does the Turbulence Law imply now?

The gentlest outcome in the 13-year record was a 21.2% peak to trough fall, which repeated from this rally's high near $81,455 would touch about $64,000. The modern-era median drawdown of 23% lands near $63,000. The all-era median was 36%. The law says nothing about timing inside the 90-day window, and a first-ever miss is always possible on a sample of 14.

Keep reading

We break down the market like this every day, free on Instagram and YouTube, and in depth inside the community.

Education, not financial advice. Trading involves real risk.