RSI, the Relative Strength Index, is the most used momentum gauge in crypto. It turns the last 14 candles of a chart into one number between 0 and 100 that says how one-sided the recent closes were. Coinlyzer calculates it on every timeframe for every pair, so you can screen by it instead of opening charts one at a time.
What RSI measures
Look at the recent candles and only at how each close compares with the one before. Some closes were higher, some lower. RSI asks: of all that movement, how much was upward? If the recent closes were all higher, RSI is near 100. If they were all lower, it’s near 0. If the gains and losses were the same size, it’s 50. The last 14 candles count most; older ones fade out (see below).
So a high RSI doesn’t mean a price is high, and a low RSI doesn’t mean it’s cheap. It means the recent closes were mostly in one direction. J. Welles Wilder, who introduced RSI in 1978, used it to spot moves that had run a long way in one direction.
How Coinlyzer calculates it
- For each candle, the change from the previous close: a gain if it closed higher, a loss if it closed lower.
- The first average gain and average loss are simple averages over the first 14 changes.
- Every later candle updates them with Wilder’s smoothing: new average = (previous average × 13 + this candle’s gain or loss) ÷ 14. Older candles never drop out completely; their weight just fades.
- RS = average gain ÷ average loss, and RSI = 100 − 100 ÷ (1 + RS).
Coinlyzer runs this over the last 100 candles of each timeframe (300 for 5-minute candles). After that many candles, the starting averages carry less than 0.2% of the weight, so the result matches charting tools that use Wilder’s RSI on the same exchange’s candles, apart from tiny rounding differences. A pair with fewer than 15 candles on a timeframe, such as a new listing on the weekly chart, shows no RSI there yet. A candle with no trades counts as a close with no change.
Why hourly and daily RSI disagree
The same coin, the same moment, three timeframes:
| SOL/USDT, last closed candle | 1h RSI | 4h RSI | 1d RSI |
|---|---|---|---|
| 18 Nov 2026, 05:19 UTC | 63.7 | 52.0 | 29.1 |
None of these is wrong. The daily RSI covers a few weeks, and in that time Solana fell from $86.74 to $70.28 in five daily closes before recovering to $75.19: mostly down-closes, so the daily RSI was below 30. The hourly RSI is dominated by the last day or so, when the price was bouncing back: mostly up-closes, so it was above 60. The 4-hour RSI sat in between, near 50.
That’s the useful part. A low daily RSI with a recovering hourly RSI describes a coin that has fallen hard and is now bouncing. Whether the bounce lasts is a different question; RSI only tells you what the closes did. A screen can look for exactly this combination across every pair (below).
70 and 30: what the levels mean
- Above 70 is traditionally called overbought, below 30 oversold. They’re Wilder’s original levels and still the defaults on most charts.
- Strong trends stay there. In a steady rally, RSI can sit above 70 for days; in a slide, below 30. A reading beyond 70 or 30 describes a one-sided move, not a move that must reverse.
- Some traders use 80 and 20 on volatile coins, so that only the most extreme readings stand out. In Coinlyzer the threshold is just a rule value, so you can use whichever you like.
- 50 is a trend line of sorts. RSI holding above 50 means up-closes have outweighed down-closes; some traders use it to tell an uptrend from a downtrend.
Live RSI and closed RSI
Every RSI column comes in two versions. 1h RSI treats the current price as the last close, so it moves with every trade until the hour ends. 1h RSI ◂ uses the last closed candle and doesn’t change until the next close. Use the live one to watch a move as it happens, and the closed one for screens and alerts that should only fire on a finished candle. Now vs Last explains the difference for every column.
RSI screens to try
Daily RSI under 30
Pairs whose daily RSI closed below 30: coins after a run of lower daily closes. Sort by 1d RSI to see the most extreme first.
All exchanges · 24h volume at least $1M · 1 day RSI (14) (last closed) below 30
Daily RSI low, hourly RSI recovering
The Solana pattern above: a low daily RSI with an hourly RSI above 55. Coins that fell hard and are bouncing on the hourly chart.
All exchanges · 24h volume at least $1M · 1 day RSI (14) (last closed) below 35 AND 1 hour RSI (14) (last closed) above 55
4h RSI above 70, closed candle
Pairs whose last 4-hour candle closed with RSI above 70: a strong one-sided run on the 4-hour chart, confirmed by a closed candle.
All exchanges · 24h volume at least $1M · 4 hour RSI (14) (last closed) above 70
FAQ
What is a good RSI for crypto?
There is no good or bad value. RSI describes how one-sided recent closes were: 50 is balanced, above 70 is the traditional “overbought” zone and below 30 “oversold”. Strong trends can stay beyond those levels for a long time, so RSI is context for a decision, not a signal on its own.
Which timeframe should I use RSI on?
The one that matches how long you plan to hold. Hourly RSI describes the last day or two, daily RSI the last few weeks. Looking at two together, for example daily for the trend and hourly for timing, tells you more than either alone.
Why is Coinlyzer’s RSI different from my chart?
Check three things: the same exchange and pair (BTC/USDT on Binance is not BTC/USD on Coinbase), the same timeframe, and the same candle (Coinlyzer has a live and a last-closed RSI). Coinlyzer uses Wilder’s RSI with 14 periods, the default on most charting tools.
Is RSI a Pro column?
No. RSI on every timeframe is included on the Free plan, live and on the last closed candle.
Can a screen combine RSI on two timeframes?
Yes. Each timeframe is its own column, so a screen can ask for daily RSI under 35 and hourly RSI above 55 at the same time, across every pair.
Related
Coinlyzer shows market data only. Nothing in this article is investment advice or a recommendation to buy or sell anything. Past price data doesn’t predict future prices. Figures are from Binance’s public market data and were checked when the article was updated.