Do crypto trading signals work? RSI, MACD and more tested
Quick answer
No, not in our tests. We tested 18 trading signals on real price history, each with its pass bar written before we looked: 0 passed, 13 failed and 5 were mixed or studies.
18 tests0 passed
13 failed5 mixed or studies
What it means for you
Chart signals: what this family of tests supports, in plain moves. Information only, not financial advice.
- A classic indicator told us little about the next month. Across every coin, the signal days did about the same as random days, often worse.
- A signal that 'worked' on one coin is usually luck: with this many coins and rules, some always look good.
- Treat indicators as a description of what already happened, not a prediction of what comes next.
All 18 tests, passed first
FAILEDOct 29, 2026
Over all history Bitcoin rose more after Tether printing (+7.6% vs +3.5%), but since 2024 it flipped (-3.1% vs +3.0%), so it can't be trusted. Full result →
FAILEDOct 27, 2026
No. 'Oversold' helped a little across all coins (2.3% more gain), but on the four test coins it did worse (3.3% less). Full result →
FAILEDOct 27, 2026
No, the opposite. After an 'overbought' reading, coins kept rising more than usual (4.3% more gain), so it was a bad sell signal. Full result →
FAILEDOct 27, 2026
No. MACD buy and sell crosses made almost no difference (0.7% more and 0.1% less than an average day). Full result →
FAILEDOct 27, 2026
No. A squeeze breakout did slightly worse than buying on an average day (1.4% less gain across all coins). Full result →
FAILEDOct 27, 2026
No. Buying a bounce off a rising 50-day average was no better than an average day (0.1% less gain). Full result →
FAILEDOct 27, 2026
No. Buying at the 61.8% pullback made no difference across all coins, and on the four test coins it did worse (8.1% less gain). Full result →
FAILEDOct 27, 2026
No. Small 10% dips helped a little until 2024, bigger 20-30% dips did worse than a random day, and in falling markets every dip did 6.1-8.0% worse. Full result →
FAILEDOct 27, 2026
No. Round prices like $1 or $100,000 held as floors no more often than any other price (37% vs 38%). Full result →
FAILEDOct 27, 2026
No. Buying the stochastic 'oversold' turn did slightly worse than an average day (1.0% less gain). Full result →
FAILEDOct 27, 2026
No. It looked good on average, but one Solana stretch did most of the work, and only 2 of the four test coins were helped. Full result →
FAILEDOct 27, 2026
No. Coins that had just beaten Bitcoin by 20%+ didn't keep beating it over the next 30 days. Full result →
FAILEDOct 22, 2026
No. 68% of coins that pumped 30%+ in a week did worse than the average coin afterwards. Full result →
MIXEDOct 27, 2026
Partly. A small bounce showed up since 2024 (2.8% more gain over 7 days), but not over all history, and with this many tests it could be luck. Full result →
MIXEDOct 27, 2026
Partly. New highs led to bigger gains over all history (10.7% more), but since 2024 the edge mostly disappeared (1.6% more). Full result →
MIXEDOct 27, 2026
Not the cuts. Rate cuts did worse on the four test coins (10.7% less gain); hikes did hurt over all history, but there are too few recent ones to judge. Full result →
MIXEDOct 27, 2026
Partly. After a 10%+ day, the next day was a bit weaker than usual (0.4% worse), but it's small and not reliable on the four test coins. Full result →
STUDYOct 22, 2026
Coins that crashed 30%+ in a week usually kept doing worse: 76% trailed the average coin. Full result →
Other topics
Free: the 9 rules that passed, on one page. Plus every new test by email. No coin picks, no hype. Unsubscribe anytime.
Tested on past prices with no fees unless stated; the past is no promise. Information only, not financial advice. Every test on one page →