Bitcoin's 4-Hour RSI Just Hit 94.4. Here Is What Happened Every Other Time.
Bitcoin's 4-hour RSI printed 94.4 this week. Only one reading in 13 years of data has ever been higher. So I went back through every time it came close, sorted them by the kind of market they fired in, and looked at what happened next. The answer depends almost entirely on one thing.
What 94 means
RSI, the relative strength index, scores how fast and how one-sided a move has been on a scale from 0 to 100, by comparing the size of recent up-moves to recent down-moves over the last 14 bars. Above 70 is conventionally called overbought. Above 90 is rare. Above 94 has happened a handful of times in Bitcoin's life.
On a 4-hour chart, a reading in the mid 90s means price went one direction, hard, for several days, with almost no pullback. That can happen for two very different reasons. Demand, a wall of real buyers that keeps arriving. Or force, traders who bet on lower prices with borrowed money getting closed out automatically as price rises. Closing a short means buying, whether you want to or not, and that buying lifts price into the next tier of shorts.
This week was overwhelmingly the second one. More than $4 billion of short positions were liquidated, about $3 billion of it in a single day, which reporting describes as the largest short liquidation event in at least five years. Forced buyers cannot pause. That is what pins an oscillator near its ceiling.
Where 94.4 ranks
I rebuilt the full 4-hour series from Bitstamp, 32,890 bars from 2011 to today with no gaps, and computed the standard 14-period RSI. Bitstamp's first months in 2011 are excluded as illiquid launch data. The 2012 onward record is clean.
Across 13 years, the only closed reading above this week's is November 2013 at 95.0, on the final melt-up of that bubble, about four weeks before it topped. The next two, December 2016 at 94.1 and April 2019 at 94.0, both sit below 94.4. The entire 2020 to 2021 bull run never produced a reading above 92.5. The 2024 ETF melt-up topped out at 91.1.
What happened this week is more one-sided than any leg of the last bull market.
Bull years against bear years
To get a sample worth splitting I dropped the threshold to 85, which gives 60 readings since 2012. Then I sorted them by the kind of year they fired in. Bear-market years are 2014, 2015, 2018 and 2022, the calendar years that followed the 2013, 2017 and 2021 cycle tops. Everything else is a bull or recovery year.
- Bull years, 52 readings. Three months later the median return was +18%. Only 18 of the 52 were lower.
- Bear years, 8 readings. Three months later all 8 were lower, median -18%.
Same indicator, opposite result. In a bull market, overbought is just momentum and it tends to keep going. In a bear market, it has marked a local peak every single time in the record.
Here are the eight bear-year cases in full, because a claim like eight for eight should be shown, not summarised.
- January 2014, RSI 86.4, then -17% in 30 days and -53% in 90
- May 2014, RSI 86.3, then +12% and -9%
- November 2014, RSI 90.0, then -21% and -50%
- January 2015, RSI 88.5, then -23% and -29%
- June 2015, RSI 86.1, then +11% and -7%
- November 2015, RSI 92.0, then -25% and -22%
- December 2015, RSI 86.9, then -3% and -11%
- July 2018, RSI 90.1, then -14% and -14%
2026 is a bear year. This week's reading is number nine.
Trying to break it
A finding this clean deserves to be attacked before it is trusted. Define bear three different ways and the answer holds. By calendar year at the 85 threshold, 8 of 8 lower. By calendar year at the stricter 90 threshold, 2 of 2. By distance from the cycle top, every reading above 85 inside the first 12 months after a top, 4 of 4 lower.
One lens is muddier and I want to show it rather than hide it. Sort by how far price was below its all-time high instead, and readings more than 30% below the high come out close to a coin flip, 17 of 36 lower. The reason is that this bucket includes the recovery years, 2015 into 2016, 2019, 2020 and 2023, when price was still deep in a drawdown but the bear was already over. Those recovery years are where the great exits live. Which raises the obvious question of how you know whether you are still in the bear year or already in the recovery.
The timing answers that
Measure each spike by how many months after the cycle top it fired. The genuine exits, the readings that kicked off the next bull, cluster late. January 2023 fired 14 months after the 2021 top. April 2019 fired 15 months after the 2017 top. May 2016 fired about 30 months after the 2013 top.
The traps cluster early. Inside the first 12 months after a top, a reading above 85 fired four times, in January, May and November 2014 and July 2018. All four rolled over.
Today is about 10 months after the October 2025 top. In 13 years of data, nothing that early has ever been the exit.
What I tested and found nothing
A natural idea is that the Fear and Greed reading at the moment of the spike should separate traps from exits. I tested it. The index only exists from 2018, which leaves nine prior episodes above 90. Spikes that fired in fear went both ways. Spikes that fired in greed went both ways. Nine cases is too few to say more, and I would rather report a null result than manufacture a pattern.
What the sentiment series does show is the speed of this week's flip. Fear and Greed printed 11 at the July low, 46 on 19 August, and 72 today. Extreme fear to greed in seven weeks. That is not evidence of a bottom. It is a description of the crowd moving from one side of the boat to the other, which is the precondition for the next squeeze in either direction.
Structure, both sides stated
The indicator cannot settle whether this is a bear-market rally or the start of the next bull. Structure can, and right now it says two things that cut against each other.
For the bulls. Price closed back above its 200-day moving average this week for the first time in 289 days. That is a genuine change. And the July 2018 trap fired with price well below its 200-day, while the April 2019 and January 2023 exits fired above it.
For the bears. Every major high since October has been lower than the one before it. $126,272, then $97,939, then $82,833, then this week's high. A downtrend is a sequence of lower highs, and one green week does not end it.
That sequence breaks on a daily close above $82,833, the May high. If that happens, the right response is not to argue with it. It is to re-evaluate and accept that the market is changing. Until then, I treat this as a rally inside a bear, and it is the ninth of its kind.
What this does not say
It does not say price goes down from here. Eight for eight is a base rate across eight cases, and the ninth is free to be the first exception. It does not say the July low will break. Rolling over after a spike and making a new low are different claims, and the median bear-year outcome of -18% would not reach the July low from today's level. And it is one indicator on one timeframe, chosen because it produced this week's headline number. Treat it as a single witness whose testimony happens to agree with the structure, not as a verdict.
Education, not financial advice.
Common questions
Is RSI 94 bullish or bearish for Bitcoin?
It depends on the regime. Since 2012, a 4-hour RSI above 85 in a bull-market year was followed by a median +18% three months later, with 34 of 52 cases higher. In a bear-market year, all 8 cases were lower three months later, median -18%. 2026 is a bear year, so the historical base rate points lower, but a base rate is odds, not a verdict.
What is the highest Bitcoin 4-hour RSI ever?
On Bitstamp 4-hour closes since 2012, the highest closed reading is 95.0 in November 2013, on the final leg of that bubble. This week's live print of 94.4 is the second highest in 13 years. December 2016 (94.1) and April 2019 (94.0) are next. The entire 2020 to 2021 bull market never exceeded 92.5.
Why did Bitcoin's RSI spike so high this week?
Forced buying. More than $4 billion of short positions were liquidated this week, about $3 billion in a single day, the largest short liquidation event in at least five years. Closing a short means buying, and cascading liquidations produce exactly the kind of one-directional move that pins RSI near its ceiling.
Does an overbought RSI mean the Bitcoin bottom is in?
Not on its own. The genuine exits from past bears, April 2019 and January 2023, fired 14 to 15 months after their cycle tops. Inside the first 12 months after a top, a reading like this fired four times and rolled over four times. Today is about 10 months after the October 2025 top, earlier than any exit in the record.
What price would change the bearish read on Bitcoin?
A daily close above $82,833, the May 2026 high. Every major high since the October 2025 top has been lower than the last, $126,272, $97,939, $82,833, then this week's high. Closing above the May high would break that sequence for the first time, at which point the bear-market framing should be re-evaluated rather than defended.
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.