RSI (Relative Strength Index)
Here's how RSI is calculated. Take the last 14 candles, average the gains of the up candles and the losses of the down candles, then compare the two. When gains outweigh losses, RSI sits above 50; when losses dominate, it falls below 50. For example, if the average up move was 2% and the average down move 1% over the last 14 candles, RSI comes out around 67. Nothing but up moves pushes it toward 100; nothing but down moves, toward 0.
The key point: RSI measures how lopsided recent movement has been — not whether the price is high or low. Whether Bitcoin is at $100,000 or $50,000, RSI only looks at how one-sided the last 14 candles were. That's why RSI can read completely differently today versus yesterday at the exact same price.
The thresholds are convention, hardened over time. Calling below-30 oversold and above-70 overbought comes straight from Welles Wilder, who proposed those levels when he created the indicator in 1978 — there's no law-of-physics reason behind the numbers. Some traders track the more extreme 20 and 80 instead.
Many traders expect a bounce below 30 and a pullback above 70 — the psychology that the selling is exhausted, so price should come back at least once. But there's an opposite property: in a strong downtrend RSI can sit below 30 for days while price keeps falling, and in a strong uptrend it can pin above 70 and keep climbing. An extreme RSI can simply mean the current trend is that strong.
So RSI isn't a button that says 'buy now' or 'sell now' — reading it as a status display, 'the market has been this lopsided lately,' is the accurate use. What price actually did next is something you count from historical data, not something you feel.
What the data actually shows
Barobara counts every past instance of RSI signals firing on the Bitcoin chart and publishes what price actually did next, as is. Spoiler: in most stretches, up and down cases don't stray far from 50/50. Where a condition looks like a high win rate, it's often an illusion created by a tight target — or the expected value turns negative once fees are subtracted. See for yourself: RSI oversold (1h) actual results, RSI overbought (1h) actual results, and for the deeper extreme, RSI deeply oversold (1d). How fees affect the outcome is covered in the fee guide.Common misconceptions
The most common misconception is 'RSI below 30 = buy signal.' Below 30 is a status flag — 'a lot was sold recently' — not any assurance that price is likely to rise from here. In the actual historical data, bounces and further drops are mixed in roughly equal measure. The second misconception is 'RSI above 70 = crash incoming' — early in big bull runs, RSI can pin above 80 while price climbs much further. Overbought isn't a crash warning; it can just as well mean the buying pressure is that strong.
FAQ
Q. Should I buy when RSI drops below 30?
There's no fixed answer. RSI below 30 only tells you a lot was sold in a short period — whether price rises or falls next is a separate question. When Barobara counted every past case, most came out close to a coin flip. We'd recommend checking the actual distribution before deciding.
Q. Why is the RSI period 14? Can I use other numbers?
14 is the convention its creator proposed; plenty of people use 7 or 21 or something else. A shorter period fires signals more often but with more noise; a longer one fires rarely but reacts slowly. Neither grants any ability to predict the future.