Golden cross, RSI & co.: what popular crypto signals are really worth
We recalculated 10 popular signals for Bitcoin and Ethereum since 2017 – with rules fixed before the first calculation. Most had no measurable edge, one did.
Showed an edge
Below the 200-week line for Ethereum: 90 days later, on average +30 percentage points versus normal (10 signals).
With 20 tests, a hit can also be chance.
Tendency, not confirmed
Golden cross
On average in the claimed direction and fairly consistent – but not enough to count as proof.
Tended to go the other way
Extreme fear for Bitcoin and Ethereum even tended to work the other way round: afterwards the price did worse than normal.
No clear edge
Death crossRSI oversoldRSI overboughtExtreme greedMayer Multiple low
On their own, these signals had no measurable advantage over normal performance.
Too rare for a verdict: Mayer Multiple high, Below the 200-week line (Bitcoin), and Pi Cycle Top.
We fixed the rules on 30 Sep 2026, before calculating a single result. The comparison is always with normal performance over the same period – not with zero. This is not investment advice.
Golden cross for Bitcoin: where the signals were
Weekly closes 08/17/2017 – 09/29/2026 (Binance BTCUSDT) · each dot is one counted signal
$121,333$3,380
08/22/2017logarithmic scale09/29/2026
9 signals – point or tap for details
Golden cross
50-day line crosses the 200-day line from below
Last signal: 09/08/2026
tendency, not confirmed
Claim: is considered a buy signal: the trend is turning up (price rises).
On average in the signal's direction and fairly consistent – not enough for proof (t-value of 2 or more).
+8%+4%+25%+14%+34%+33%
30 days
90 days
180 days
after the signalnormal over the same period
Signals (90 d)
8
Hit rate
88%
normal 54%
Worst case
−4%
after a signal
Edge in the signal's direction over 90 days: +10.7 pts · t-value 1.1
By market regime: Bull −3 pts (6) · Sideways +27 pts (2) · Bear no signal
One of 20 tests – with this many tests, chance hits happen too.
All 9 signal dates
Date
30 d
90 d
180 d
04/24/2019
+47%
+82%
+51%
02/18/2020
−39%
−4%
+17%
05/21/2020
+3%
+30%
+95%
09/15/2021
+28%
+0%
−18%
02/07/2023
−12%
+19%
+25%
10/30/2023
+10%
+22%
+84%
10/28/2024
+37%
+47%
+35%
05/22/2025
−9%
+2%
−17%
09/08/2026
still open
still open
still open
Death cross
50-day line crosses the 200-day line from above
Last signal: 11/16/2025
no clear edge
Claim: is considered a sell signal: the trend is turning down (price falls).
On average the price tended to move against the signal's direction afterwards.
+2%+4%+16%+14%+31%+33%
30 days
90 days
180 days
after the signalnormal over the same period
Signals (90 d)
9
Hit rate
44%
normal 46%
Worst case
+60%
despite the warning
Edge in the signal's direction over 90 days: −2.4 pts · t-value −0.2
By market regime: Bull no signal · Sideways −19 pts (4) · Bear +0 pts (5)
One of 20 tests – with this many tests, chance hits happen too.
All 9 signal dates
Date
30 d
90 d
180 d
03/31/2018
+34%
−10%
−3%
10/26/2019
−23%
−9%
−19%
03/25/2020
+12%
+44%
+56%
06/19/2021
−13%
+33%
+34%
01/14/2022
−2%
−7%
−53%
09/12/2023
+4%
+60%
+167%
08/10/2024
−6%
+26%
+58%
04/07/2025
+23%
+38%
+55%
11/16/2025
−7%
−26%
−16%
RSI oversold
RSI (14 days) falls below 30
Last signal: 06/02/2026
no clear edge
Claim: is considered a sign that a rebound is coming (price rises).
On average in the signal's direction, but so uneven that it may be chance.
+5%+5%+31%+17%+29%+32%
30 days
90 days
180 days
after the signalnormal over the same period
Signals (90 d)
21
Hit rate
62%
normal 54%
Worst case
−39%
after a signal
Edge in the signal's direction over 90 days: +14.4 pts · t-value 0.7
One of 20 tests – with this many tests, chance hits happen too.
All 8 signal dates
Date
30 d
90 d
180 d
06/14/2022
−7%
+1%
−23%
07/23/2022
−5%
−15%
−6%
06/05/2023
+19%
+1%
+53%
08/17/2023
−0%
+42%
+87%
10/02/2023
+29%
+54%
+153%
06/05/2026
+4%
+33%
still open
07/08/2026
+4%
still open
still open
08/11/2026
+20%
still open
still open
Pi Cycle Top
111-day line crosses 2 × 350-day line from below
Last signal: 04/12/2021
too few signals
Claim: is considered a signal of a cycle top (price falls).
−17%+4%−43%+15%−8%+37%
30 days
90 days
180 days
after the signalnormal over the same period
Signals (90 d)
1
Hit rate
100%
normal 44%
Worst case
−43%
despite the warning
Edge in the signal's direction over 90 days: +58.2 pts
By market regime: Bull +66 pts (1) · Sideways no signal · Bear no signal
One of 20 tests – with this many tests, chance hits happen too.
All 1 signal dates
Date
30 d
90 d
180 d
04/12/2021
−17%
−43%
−8%
How we calculate
Daily closes from Binance from 08/17/2017, Fear & Greed Index: alternative.me from 02/01/2018. Longer Bitcoin histories from other exchanges can only be used commercially with a separate licence – which is why the years before 2017 are missing.
A signal counts on the day the condition becomes true. Further signals of the same kind in the following 30 days don't count.
We compare the return after 30, 90 and 180 days with the normal return over the same span – on all days on which the indicator could be calculated.
Verdict over 90 days: fewer than 8 signals = “too few signals”, a t-value of 2 or more in the claimed direction = “edge measured”, 1 or more = “tendency, not confirmed”, otherwise “no clear edge”. For warning signals, the edge is how much weaker the price ran than normal.
Limits: signals can overlap over 90 and 180 days – the t-value then overstates the certainty. The data covers roughly two market cycles. Hypothetical, retrospective, no fees.