RSI appears in more trading strategies than almost any other indicator. Here's what backtesting actually shows — and why the default settings are probably wrong for you.
If you've spent any time in crypto trading communities, you've seen RSI everywhere.
"RSI is oversold — time to buy." "RSI hit 80, correction incoming." "My RSI strategy returned 300% last year."
RSI is the most widely referenced indicator in retail crypto trading. It's been around since 1978. It's free on every charting platform. And in our experience backtesting it across dozens of asset and timeframe combinations — the results are far more interesting than the standard "oversold = buy" narrative suggests.
1. What RSI Actually Measures
RSI stands for Relative Strength Index. It was developed by J. Welles Wilder and introduced in his 1978 book New Concepts in Technical Trading Systems — which makes it older than most crypto traders.
The calculation measures the speed and magnitude of recent price changes, normalized to a scale of 0 to 100.
In plain terms: RSI answers "how strong has recent upward movement been relative to recent downward movement?"
- RSI above 70: price has moved up strongly relative to recent history → "overbought"
- RSI below 30: price has moved down strongly → "oversold"
- RSI around 50: roughly balanced momentum
The classic trading logic: buy when oversold, sell when overbought. Simple. Intuitive. And — on its own — often wrong.
2. The Oversold Trap
Here's the problem with "RSI below 30 = buy signal" as a standalone strategy.
In a strong downtrend, RSI can stay below 30 for weeks or months. Every time it dips below 30, the naive strategy says buy. Every time, price continues lower. You're not catching a bottom — you're catching a falling knife, repeatedly.
Bitcoin in 2018: RSI dropped below 30 in February. Then again in March. Then again in June. Then in November and December as price fell from $6,000 to $3,200. Each oversold reading was technically correct — and each one was followed by further decline.
The issue isn't RSI itself. It's using a momentum indicator without any trend context.
RSI below 30 means: recent selling pressure has been intense. It does not mean: the selling is over.
Those are very different statements.
3. RSI With a Trend Filter: A Different Story
The results change significantly when RSI signals are filtered by trend context.
The most common approach — and the one we've tested extensively — is the RSI/SMA Cross: only take RSI buy signals when price is above a moving average (confirming an uptrend). Ignore RSI buy signals when price is below the moving average (downtrend context = falling knife risk).
The logic: in an uptrend, an oversold RSI reading means a temporary pullback in an otherwise positive environment. That's a genuine buying opportunity. In a downtrend, an oversold RSI reading means the trend is continuing. That's a trap.
This combination — RSI threshold + SMA trend filter — produces meaningfully cleaner signals than RSI alone across most assets and timeframes we've tested.
The improvement isn't dramatic in every case. But the reduction in losing trades during sustained downtrends is consistent.
4. Daily vs. Weekly: The Timeframe Effect
This is where the most significant difference shows up — and it's consistent enough to be worth stating clearly.
RSI on daily candles:
More signals. On Bitcoin, RSI crosses typical thresholds dozens of times per year on the daily chart. More signals means more trades, more transaction costs, and more exposure to false entries during sideways markets.
The 2022-2023 sideways period is particularly brutal for daily RSI strategies. Extended low-volatility grinding produces repeated oversold readings that don't resolve into meaningful bounces.
RSI on weekly candles:
Fewer signals — typically 5-15 per year depending on parameters. Each signal represents a more significant momentum shift. The weekly candle filters out the daily noise that generates false entries.
The general pattern across our backtests: RSI-based strategies tend to perform significantly better on weekly candles than daily, across most crypto assets. More signals doesn't equal more profit. On a noisy asset like crypto, it often means the opposite.
5. The Parameter Problem
RSI 14 is the default setting on every platform. Period: 14 candles. Overbought: 70. Oversold: 30.
These defaults exist because Wilder chose them in 1978 for commodities markets. They've been carried forward unchanged for 45 years.
For Bitcoin on weekly candles in 2026, are those the right parameters?
Possibly. Possibly not.
Here's the honest answer: the "right" parameters depend on the asset, the timeframe, and the market regime. What we can say is that the RSI 14 default on daily crypto charts is probably the most over-optimized setting in existence. It's what everyone uses — which means any edge it once had has been arbitraged away by the sheer number of people acting on the same signal.
Testing different period lengths and threshold levels matters. Not to find the "best" parameters through curve-fitting — that's a trap — but to understand whether the signal holds up across reasonable parameter variations. A strategy that only works with RSI 12, threshold 28.5, SMA 47 is probably not a real edge. A strategy that works reasonably well with RSI 10-18, threshold 25-35, SMA 40-60 is showing more robust behavior.
6. What the Data Shows
Without giving you a single "definitive" backtest result — which would be misleading without the context of asset, timeframe, and period — here's what consistent patterns emerge from extensive RSI testing:
| Setup | Assessment |
|---|---|
| RSI alone, daily candles | Mixed to negative on crypto during bear markets. Consistent whipsaw in sideways markets. |
| RSI alone, weekly candles | Better than daily, but still significantly improved by trend context. |
| RSI + SMA trend filter, daily candles | Meaningfully better than RSI alone. Still challenged during extended sideways periods. |
| RSI + SMA trend filter, weekly candles | The strongest consistent performer in our testing across multiple assets and time periods. Fewer trades, cleaner signals, better risk-adjusted returns than the daily equivalents. |
The comparison against Buy & Hold and Average B&H varies significantly by period. In strong bull markets, passive holding often wins — it was fully invested from the start. Over full cycles including bear markets, the RSI/SMA combination on weekly candles has historically held up reasonably well.
7. What RSI Won't Tell You
RSI is a momentum oscillator. It describes what has happened to price recently. It does not predict what will happen next.
"RSI is oversold" means selling pressure has been intense recently. Nothing more. The signal becomes meaningful only in context — trend direction, timeframe, confirmation from other signals.
The traders who lose money on RSI strategies are almost always using it as a standalone signal without trend context, on too short a timeframe, acting on every crossover without considering market regime.
The traders who use RSI effectively treat it as one input in a larger framework — not as a buy/sell button.
8. The Bottom Line
RSI is not magic. The default settings are not optimal for crypto. Daily signals without trend context are a reliable way to lose money in bear markets.
But RSI with a trend filter, on weekly candles, with reasonable parameters, has a legitimate historical record on major crypto assets. Not every period. Not on every asset. But consistently enough across multiple market cycles to take seriously.
That's more than most indicators can claim when you actually run the numbers.
Test RSI strategies with different parameters, timeframes, and trend filters on any crypto or stock asset: → tradingstrategies.work
Study the past, improve your future.
In the next post: RSI versus Golden Cross versus OBV-MACD in direct comparison. Which strategy actually wins?
FAQ:
Question: Does the simple RSI strategy "buy below 30, sell above 70" work?
Answer: Used alone, usually not. In strong downtrends, RSI can stay below 30 for weeks or months — every buy attempt catches the next falling knife. Bitcoin 2018 is the textbook example: RSI was below 30 multiple times from February through December, yet price still fell from $6,000 to $3,200. The naive logic only works in uptrends — in downtrends it systematically produces losses.
Question: What is the RSI/SMA Cross and why does it work better?
Answer: The RSI/SMA Cross combines the momentum signal from RSI with a trend filter (Simple Moving Average). Buy signals are only accepted when price is above the moving average — i.e. in a confirmed uptrend. RSI buy signals are ignored in downtrends. This avoids the costliest error: buying into ongoing declines. In our backtests, the setup consistently reduces losing trades across every crypto asset tested.
Question: Should I change the default RSI 14 / 70 / 30 settings?
Answer: Probably yes, but not through blind optimization. The defaults date from 1978 and were designed for commodities markets — crypto behaves differently. More important than the "best" value is robustness: a strategy that only works with RSI 12, threshold 28.5, SMA 47 is over-optimized. A strategy that works reasonably with RSI 10-18, threshold 25-35, SMA 40-60 shows real robustness. Backtesting Arena lets you test parameter ranges to verify robustness rather than point-optimization.
Question: Does RSI work better on daily or weekly candles?
Answer: In our backtests, consistently better on weekly candles across most crypto assets. Daily RSI produces dozens of signals per year, many of them false signals in sideways phases. Weekly RSI produces 5-15 signals per year, each one representing a more substantial momentum shift. More signals doesn't equal more profit — on a volatile asset like crypto, often the opposite.