Golden cross and death cross: every Bitcoin case measured
Coinbase Bitcoin was higher 90 days after 7 of 8 golden crosses, but only 8 cases exist. See each cross, 10 more coins, and a rule after fees.
Bitcoin was higher 90 days after 7 of the 8 golden crosses we could measure on Coinbase (87.50%), with a middle move of +20.50%. Across all days, it was higher 90 days later 59.80% of the time, with a middle move of +9.79%. After death crosses, price was still higher 90 days later in 5 of 9 cases. That sounds like a clear story, but it rests on 8 or 9 cases, and a rule built on the crosses did not beat simply holding Bitcoin. These are past cases, not a forecast.
What are a golden cross and a death cross?
A moving average is the average of the last N closing prices. The 50-day line reacts fast. The 200-day line reacts slowly.
A golden cross is the first day the 50-day line closes above the 200-day line. A death cross is the opposite. Both lines look backward, so a cross shows up after price has already moved. For Coinbase BTC-USD the 200-day line first existed on February 4, 2016.
How many crosses has Bitcoin had?
On Coinbase BTC-USD (Bitcoin priced in US dollars), from February 4, 2016 to October 19, 2026, there were 9 golden crosses and 9 death crosses. Moves run from the cross-day close.
| Golden cross date | 30 days later | 90 days later | 180 days later |
|---|---|---|---|
| 2019-04-23 | +42.46% | +86.59% | +48.86% |
| 2020-02-18 | −39.28% | −4.51% | +16.99% |
| 2020-05-21 | +3.24% | +29.76% | +95.09% |
| 2021-09-15 | +28.14% | +0.46% | −17.60% |
| 2023-02-07 | −12.38% | +19.15% | +24.93% |
| 2023-10-30 | +9.77% | +21.86% | +83.86% |
| 2024-10-28 | +37.22% | +46.68% | +35.38% |
| 2025-05-22 | −8.56% | +2.29% | −16.84% |
| 2026-09-08 | +4.14% | not yet | not yet |
The newest golden cross, on September 8, 2026, has no 90-day result yet, so 90-day and 180-day counts use 8 cases. No death cross had followed it by October 19, 2026.
| Death cross date | 30 days later | 90 days later | 180 days later |
|---|---|---|---|
| 2018-03-30 | +37.11% | −14.55% | −5.73% |
| 2019-10-26 | −23.03% | −8.99% | −19.12% |
| 2020-03-26 | +11.68% | +37.40% | +55.83% |
| 2021-06-19 | −13.08% | +33.31% | +34.24% |
| 2022-01-14 | −2.36% | −7.28% | −53.08% |
| 2023-09-12 | +3.52% | +59.55% | +167.15% |
| 2024-08-10 | −6.37% | +25.65% | +58.47% |
| 2025-04-07 | +22.64% | +38.01% | +54.74% |
| 2025-11-16 | −6.72% | −25.90% | −16.06% |
What happened after the crosses, compared with a normal day?
The median is the middle result when you sort all the moves. "Higher" counts cases where price was above the cross-day close.
| Starting point | 30 days: higher (median) | 90 days: higher (median) | 180 days: higher (median) |
|---|---|---|---|
| After a golden cross | 6 of 9 (+4.14%) | 7 of 8 (+20.50%) | 6 of 8 (+30.16%) |
| After a death cross | 4 of 9 (−2.36%) | 5 of 9 (+25.65%) | 5 of 9 (+34.24%) |
| Any day with a 200-day line | 57.56% (+3.05%) | 59.80% (+9.79%) | 64.14% (+26.09%) |
Golden crosses looked better than a normal day at 90 days, but not every time. February 18, 2020 fell −39.28% over the next 30 days. September 15, 2021 was nearly flat at 90 days (+0.46%) and −17.60% at 180 days.
Death crosses did not behave like a warning. In 5 of 9 cases price was higher after 90 days. The September 12, 2023 cross was followed by +59.55%. The January 14, 2022 cross was followed by −53.08% at 180 days.
With 8 or 9 cases, one more result could move these shares a lot. See why past crypto moves are not a forecast.
Does it matter if price is above or below the 200-day line?
This is a state, not an event. On each day, we asked whether the close was above or below the 200-day line, and what came next.
| 90 days later | Days counted | Higher | Median |
|---|---|---|---|
| Close above the 200-day line | 2,349 | 66.79% | +16.99% |
| Close below the 200-day line | 1,472 | 48.64% | −0.98% |
The gap is bigger than for crosses. But neighboring days share the same future, so they are not separate tests. The days above the line hold about 26 separate 90-day looks, and the days below about 16. Price also stays on one side for long stretches: in 2017 it closed above the line on all 365 days, and in 2022 on 0 of 365. For more on this line alone, see Bitcoin and its 200-day moving average.
Did the same thing hold for other coins?
We repeated the test on 10 Binance USDT perps. A perp is a futures contract that never expires. Coinbase and Binance are never pooled. The last two columns use the rule explained in the next section.
| Coin (Binance perp) | Golden crosses higher after 90 days | All days higher after 90 days | Follow-the-cross rule, after fees | Buy-and-hold, same span |
|---|---|---|---|---|
| BTCUSDT | 6 of 6 | 58.40% | +581.37% | +804.01% |
| ETHUSDT | 4 of 5 | 56.91% | −12.83% | −20.14% |
| SOLUSDT | 5 of 5 | 49.64% | +30.74% | +314.29% |
| XRPUSDT | 2 of 9 | 51.05% | −10.67% | +340.78% |
| DOGEUSDT | 2 of 4 | 39.57% | −21.21% | −6.26% |
| ADAUSDT | 3 of 5 | 40.74% | −65.31% | −36.38% |
| AVAXUSDT | 2 of 5 | 39.50% | −11.29% | −79.68% |
| LINKUSDT | 2 of 6 | 46.35% | −64.74% | −62.68% |
| BNBUSDT | 5 of 6 | 57.77% | +40.04% | +57.75% |
| TRXUSDT | 2 of 6 | 68.87% | +68.28% | +217.38% |
Bitcoin and Solana looked good. XRP, LINK and TRX did not: only 2 golden crosses each were higher after 90 days. Every coin has between 4 and 9 cases, and these perps began trading in 2019 or 2020, so each saw mostly one boom and one bust. Compare a coin only with its own buy-and-hold, since the spans start on different dates.
Would following the crosses have beaten holding?
We tested a simple rule on Coinbase BTC-USD. Buy at a golden-cross close, sell at the next death-cross close, and stay in cash in between. We charged 0.05% on each buy and sell, a typical taker fee (paid when an order trades right away).
From April 23, 2019 to October 19, 2026, the rule gained +654.11% after fees. Buy-and-hold over the same span gained +1,379.15%. The rule held Bitcoin on 59.19% of days. Its worst fall from a peak was −65.85%, against −76.67% for buy-and-hold.
The result depends on the period. Through 2022, the rule gained +287.53% against +206.82% for buy-and-hold. From 2023 on, it gained +94.60% against +395.08%. It made 8 completed trades, plus one open since September 8, 2026. Five of the 8 gained. One trade, May 21, 2020 to June 19, 2021, gained +291.19% and carried much of the total.
Across all 12 markets, the rule finished ahead of buy-and-hold in only 3 and behind in 9. This is one price path per market, not proof that the rule works or fails. Fees barely mattered here because the rule traded so rarely. See how fees change a backtest for faster rules.
What does this data not show?
- It does not show why price moved. A cross comes after the move that made it.
- It gives no chance for the next cross. Eight cases are too few.
- It does not show that 50 and 200 days are best. We did not tune them.
- It leaves out spread, slippage and perp funding (a small regular payment between traders).
- It uses coins still trading today. Coins that faded away are missing.
How we measured this
We used daily closing prices for Coinbase BTC-USD and ETH-USD and for 10 Binance USDT perps, up to October 19, 2026. We averaged the last 50 and last 200 closes. A golden cross is the first close with the 50-day line above the 200-day line after it was at or below it the day before. A death cross is the reverse. Forward moves run from the cross-day close and count only when every day in the window exists. Fees apply only to the rule, and buy-and-hold uses the same start and end dates.
Try this in Comparand
Connect Comparand to your AI app and ask:
- "Get daily bars for Coinbase BTC-USD from 2017 to today, find each day the 50-day average closed above the 200-day average, and show the move 30, 90 and 180 days later." The venue_bars tool returns the open, high, low and close for each completed daily bar, and it is free. Your agent does the averaging; the tool does not return crosses. Comparand's Coinbase daily bars start in January 2017, which covers every cross in the tables above.
- "Find the 10 closest past daily charts to Coinbase BTC-USD's last 200 days." The shape_analogs tool, with timeframe 1d and window 200, compares the whole shape instead of one rule. It only looks at the same coin on the same exchange. For each past case you get its own path and its own forward move 50, 100 and 200 daily bars later, in bp (100 bp is 1%). It costs $0.01 per search and gives no average or hit rate.
You can try 3 coins free, read the docs, or see pricing. To test any rule like this one without fooling yourself, see how to backtest a crypto chart pattern.
The short version
- On Coinbase BTC-USD, 7 of 8 golden crosses were higher after 90 days (87.50%, median +20.50%), against 59.80% for all days.
- After 9 death crosses, price was higher after 90 days in 5, with a median of +25.65%.
- A rule that bought at golden crosses and sold at death crosses gained +654.11% after fees, against +1,379.15% for buy-and-hold.
- Across 12 markets, the rule finished ahead of buy-and-hold in 3.
- These are very small samples. Past cases, not a forecast.
Questions people ask
Does the golden cross work for Bitcoin?
In a very small sample, price was usually higher afterward. On Coinbase BTC-USD, 7 of the 8 golden crosses with a full 90-day window were higher 90 days later (87.50%), with a middle move of +20.50%. Across all days, 59.80% were higher after 90 days. Eight cases is too few to call a rate.
Does a death cross mean Bitcoin will fall?
Not in these past cases. After the 9 death crosses on Coinbase BTC-USD, price was higher 90 days later in 5 (55.56%), with a middle move of +25.65%. In the other 4 it was lower, by as much as −25.90%.
How many golden crosses has Bitcoin had?
On Coinbase BTC-USD, from February 4, 2016 (the first day a 200-day average exists) through October 19, 2026, there were 9 golden crosses and 9 death crosses. The newest golden cross was on September 8, 2026.
Would a rule that follows golden and death crosses have beaten buy-and-hold?
Not for Bitcoin. Buying at each golden cross and selling at each death cross on Coinbase BTC-USD from April 23, 2019 gained +654.11% after 0.05% fees per side, versus +1,379.15% for buy-and-hold over the same span. Across 12 markets, the rule finished ahead of buy-and-hold in only 3.
Why 50 and 200 days?
They are the pair most people use, so we used them and did not tune them. Other lengths would give different crosses and different results.