LombardoCycle

Claim checks

Bitcoin bottom signals, tested: death crosses, RSI, MVRV and more

Cycle “signals” usually arrive as confident screenshots. Here each is turned into an exact rule and run against every day of Bitcoin’s price history. The question is narrow: does the signal cluster near cycle bottoms, or only look that way in hindsight?

Bottom line

None of the seven X claims shows a tested, out-of-sample timing edge. The one real pattern is that the weekly 50/100 death cross clusters around bottoms, but it lags, rests on 3 resolved events, and fired before the 2022 low.

As of 2026-10-15, three signals are on: the four-week 50W streak, the monthly RSI bounce, and MVRV above its averages. The weekly 50/100 death cross and the monthly RSI bounce are each consistent with 2026-06-30 being the low. History shows that being consistent is not confirmation.

Daily 50/200 death cross

“A death cross just happened, and that happens after bottoms.”

Verdict: BEARISH-EDGE (in-sample; a late bear-market marker, not a bottom signal). Sample: 12 events since 2014, spanning 3 cycles (bottoms in 2015, 2018 and 2022).

  • Never fired within ±120 days of a cycle bottom: 0 of 12, where chance predicts about 2.
  • Median return 365 days later: −9% (11 resolved), against +84% for all days.
  • Not current: the daily 50/200 crossed up (a golden cross) on 2026-09-08. The last daily death cross was 2025-11-16.

Source: Death cross vs cycle bottoms, event study 2026-10-16; HLB project review, 2026-10-16

Daily MACD bearish cross

What was reported on Oct 14 as a “death cross”.

Verdict: NO EDGE (in-sample, not a forecast). Sample: 182 events, 2011–2026 (4 reference bottoms).

  • Fires about 12 times a year.
  • 30-day hit rate 55%, against 57% for all days.

Source: Death cross vs cycle bottoms, event study 2026-10-16; HLB project review, 2026-10-16

Weekly 50/100 death cross

“The weekly death cross marks the bottom” (X claim 1).

Verdict: INSUFFICIENT (real clustering, but it lags; in-sample). Sample: 4 events: one in each of 3 completed cycles (event years 2015, 2019 and 2022; the bottoms were in 2015, 2018 and 2022) plus 2026; 3 resolved.

  • Dates: 2015-04-26 (102 days after the bottom), 2019-02-24 (71 days after), 2022-09-11 (71 days before the bottom, followed by a further −25.5%), and 2026-07-12 (12 days after the 2026-06-30 low).
  • All 4 fell within ±120 days of a bottom, against a 19.7% base rate. The clustering is real.
  • But it lags by construction, fired before the final leg down in 2022, and was the best of about 10 moving-average variants tested, so luck is part of its record.
  • Returns 52 weeks later: +112%, +167%, +18% (median +112%, against +61% for all days).

Source: Death cross vs cycle bottoms, event study 2026-10-16; X “bottom signal” claims, claim check 2026-10-16

Weekly 50/200 death cross

The slower weekly pair.

Verdict: INSUFFICIENT (n = 1). Sample: 1 event with data (the 2022 cycle).

  • One event: 2023-02-19, 90 days after the 2022 bottom.

Source: Death cross vs cycle bottoms, event study 2026-10-16

Weekly 20/50 and EMA 21/50 death crosses

Faster weekly crosses as bottom markers.

Verdict: INSUFFICIENT (not bottom signals). Sample: 3 cycles in the weekly data (bottoms in 2015, 2018 and 2022).

  • They fired 80 to 239 days before the bottom in each cycle.

Source: Death cross vs cycle bottoms, event study 2026-10-16; HLB project review, 2026-10-16

Four weekly closes above the 50-week SMA

“Four weekly closes above the 50W means the bottom is in” (X claim 2).

Verdict: NO EDGE (in-sample, not a forecast). Sample: 7 events (6 resolved), 2015–2026: 3 completed cycles plus 2026.

  • Only 1 of 7 events came within 120 days of a bottom; the usual lag was 139–554 days.
  • Median 52-week return +133% against +77%, but the hit rate was 67% against 70%.
  • The 2021-08-15 event came 463 days before the 2022 low and was followed by −48%.
  • It fired again on 2026-10-11.

Source: X “bottom signal” claims, claim check 2026-10-16

Weekly EMA21 crosses above EMA50

An upward weekly cross as confirmation (X claim 3a).

Verdict: BULLISH-EDGE (at the n = 5 floor; a late bull confirmation, not a bottom marker). Sample: 5 events, 2012–2023, after 4 cycle bottoms.

  • Every event came 169–519 days after a bottom.
  • The edge rests mostly on two events: 2012 (+800%) and 2020 (+381%).
  • Not fired as of 2026-10-15: EMA21 75.6k against EMA50 78.3k (−3.5%).

Source: X “bottom signal” claims, claim check 2026-10-16

Weekly SMA21/50 and EMA21/SMA50 up-crosses

Variants of the same idea (X claims 3b and 3c).

Verdict: INSUFFICIENT (n = 4 each). Sample: 4 events each.

  • Too few events to judge.

Source: X “bottom signal” claims, claim check 2026-10-16

Monthly RSI(14) bounce from the low zone

“Monthly RSI bouncing from the cycle-low zone means the bottom is in” (X claim 4).

Verdict: NO EDGE (on returns; says “bottoming region”, not “bottom is in”). Sample: 9 events, 2012–2026: 3 completed cycles plus 2026.

  • Monthly RSI has never gone below 40 since 2012, so the classic ≤30 threshold gives zero events.
  • At ≤45 (a threshold chosen after seeing the lows, so tuned in-sample) there are 9 events, all near a bottom, but 4 fired before the low.
  • 52-week return +71.7% against +69.7%.
  • It has fired: 2026-07-31.

Source: X “bottom signal” claims, claim check 2026-10-16

MVRV momentum turns positive

“MVRV momentum called every cycle bottom” (X claim 5).

Verdict: NO EDGE (tested on 3 proxies; the exact formula is not public). Sample: 58–107 events per proxy, 2012–2026.

  • Proxies: MVRV crossing above its 365-day SMA, its 200-day SMA and its 155-day EMA.
  • Only 6–14% of events came near a bottom, against a 17–18% base rate: less concentrated than chance.

Source: X “bottom signal” claims, claim check 2026-10-16

Monthly MACD bearish cross, then a bottom about 11 months later

X claim 6.

Verdict: NO EDGE (in-sample, not a forecast). Sample: 6 events (5 resolved), 2014–2025.

  • The lag to the next bottom ranged from 3.5 to 35 months; only 1 of 6 fits 11 ± 2 months.
  • Median return 6 months later: −28%.

Source: X “bottom signal” claims, claim check 2026-10-16

“+8,519% / +1,882% / +564%” to the next top

A post quoting gains from a signal to each cycle’s high (X claim 7).

Verdict: EXPLAINED (matches a real signal; the size of the number proves nothing). Sample: 3 cycles (the 2017, 2021 and 2025 highs).

  • The figures match the weekly 50/100 death cross measured to each cycle’s intraday high: +8,641%, +1,709% and +471% on our closes. No daily cross fits.
  • But almost any bear-market date gives a huge “% to the next top”: the daily 50/200 cross on 2015-09-15 gives +8,221%. That is a property of Bitcoin’s cycles, not evidence for the signal.

Source: X “bottom signal” claims, claim check 2026-10-16

How the checks work

  • Exact rule, every day. A script scans 5,538 daily closes (2011-08-18 to 2026-10-15) and records every date the rule fires. Same input, same output.
  • Compared with the baseline. Returns after each event are compared with returns from every day in the sample. If they match, the signal told you nothing.
  • Near-bottom base rate. The reference bottoms are 2011-10-20, 2015-01-14, 2018-12-15 and 2022-11-21, plus 2026-06-30, which is the low so far and not confirmed. “Near” means within ±120 days, which covers about 18–20% of all days. So about 1 event in 5 lands near a bottom by chance.
  • Verdicts. INSUFFICIENT: fewer than 5 events with a resolved outcome. NO EDGE: median return within 25 percentage points of the baseline, or hit rate within 10 points. Otherwise BULLISH-EDGE or BEARISH-EDGE. Every verdict is in-sample, not a forecast.
  • Two caveats outweigh any result. In-sample: the bottoms are already known, so a rule tuned after the fact looks better than it is. Multiple comparisons: test ten variants and one will look good by luck.
  • Events are not independent. The forward windows (365 days, or 52 weeks) overlap, so events close together share most of their outcome. A count of events overstates the evidence.