Does an oversold RSI mean it is time to buy?
An RSI-14 reading below 30 means bitcoin is oversold and due to rally.
Where we are now. Today RSI-14 is 52, against a threshold of 30. Shaded bands are the 21 occasions the rule fired, the same ones counted in the table below: 2011-04, 2011-08, 2012-10, 2013-06, 2014-02, 2014-07, 2015-08, 2016-01, 2016-08, 2017-07, 2018-02, 2018-06, 2018-11, 2019-09, 2021-05, 2021-12, 2022-05, 2023-03, 2023-08, 2024-06, 2025-02.
This rule has fired 21 times. A doubling followed within a year on 10 of the 21. On comparable days when it did not fire, a doubling followed 59.0% of the time.
| Times it fired | 21 | separate occasions, not days |
|---|---|---|
| Still open | 2 | fired since 2025-11-18; the 365-day window has not closed, so these are excluded from the figures above rather than counted as a success or a failure |
| What followed | 47.6% | a doubling within 365 days of that close |
| Could plausibly be | 31% to 65% | the range the true rate could take, on this many occasions, before you invoke anything but chance; a binomial interval that treats the episodes as independent trials |
| Happened anyway | 59.0% | comparable days when the rule did not fire |
| Difference | -11.3 pts | near zero means the signal added nothing |
Where the claim comes from
Wilder's RSI, 1978, applied to bitcoin by nearly every charting service and repeated constantly on social media. The 30/70 thresholds are conventional rather than derived.
How we tested it
Trigger: RSI-14 below 30. Scored as a bottom rule: did the price double within 365 days?
Only days the rule could have fired are counted. Consecutive triggers within 90 days are one occasion. The baseline excludes each trigger and the 90 days after it, because the aftermath of a signal is not a fair comparison. No verdict is given below 20 occasions.
What the numbers say
Twenty-one occasions. 47.6% followed by a doubling within a year, against 59.0% on comparable days when the rule did not fire: a 11.4-point gap, negative, on 21 observations. The largest sample on the scorecard, and the signal was followed by a doubling less often than its absence was. Under the inferential verdict the interval still reaches the baseline, so it reads indistinguishable rather than worse.
What this does not mean
It does not follow that you should buy overbought readings instead. Inverting a bad rule rarely produces a good one, and this measures a single binary outcome, not returns.
And it does not mean the threshold itself was chosen fairly. Every rule in this registry was specified by someone who had already seen bitcoin’s history, so the cut-off is fitted to the record it is being judged against. The interval above measures sampling variation between independent trials and nothing else: not the choice of threshold, not that cycle turns are few and related, and not that thirty-six rules were tested in total. Only the forward record, which fixes a rule before its answer is known, avoids this.
What would change the answer
A different threshold, horizon or outcome definition could give a different answer. Those are stated here and the code is public; we would publish a better specification if someone proposed one.
If you cannot time it, the question changes
Most of the claims in this registry cannot be shown to beat a coin flip, and several are worse than one. That is not an argument against holding bitcoin. It is an argument that the decision worth spending effort on is how much to hold, not when to buy.
That is what How much bitcoin? is for: what fixed weights actually did in a conventional portfolio, and how much the answer depended on the year you started. For the valuation state today, and what has historically followed readings like it, see Where Things Stand.
Recomputed from the daily public snapshot by fetch/scorecard.py and published as scorecard.json. Method on the methods page; every claim in the scorecard.