Does a 200-day trend filter on Bitcoin hold up?
Short answer: In a backtest from July 2020 to October 2026 with fees, $100 held only while BTC closed above its 200-day exponential average grew to $935, against $911 for buy and hold, and the worst drop was 56% instead of 77%. It reduced the damage of bear markets; it did not beat holding by much.
Where the rule stands today
| Last daily close | $84,093 on 2026-10-21 |
|---|---|
| 200-day EMA | $76,120 |
| Close vs EMA | +10.5% |
| State | Above since 2026-08-20 (62 days) |
| Switches in the last 365 days | 8 |
Most recent switches
| Date | Close moved | Close |
|---|---|---|
| 2026-08-20 | above the EMA | $73,025 |
| 2026-05-11 | below the EMA | $81,746 |
| 2026-05-10 | above the EMA | $82,210 |
| 2025-11-03 | below the EMA | $106,583 |
| 2025-10-31 | above the EMA | $109,608 |
| 2025-10-30 | below the EMA | $108,323 |
| 2025-10-23 | above the EMA | $110,078 |
| 2025-10-22 | below the EMA | $107,567 |
This page shows where a tested rule stands. It does not recommend buying or selling anything, and there are no alerts. A rule that held in the past can fail from here.
The rule
- Each day at the UTC close, compare BTC's close with its 200-day exponential moving average (EMA).
- Close above the EMA: hold BTC. Close below: hold cash (USDT). Switch at that close.
- Costs: 0.10% per switch. No leverage. The first 200 days are only used to warm up the average.
Backtest results
$100 invested at the start of each window, values on 2026-10-21.
| Start | Buy and hold | 200-day filter |
|---|---|---|
| 2020-07-01 | $911 | $935 |
| 5 years | $139 | $183 |
| 3 years | $280 | $256 |
| 1 year | $78 | $109 |
| Worst drop since 2020-07 | 77% | 56% |
The filter lagged in the 3-year window, which was mostly a rising market, and helped in windows that included a bear market. The same rule on ETH also cut the worst drop, from 79% to 62%.
Method and limits
- Data: Binance BTCUSDT daily candles from 2019-01-01. The live section uses the same source and the same EMA definition as the backtest script.
- Fills at the daily close; real fills a few minutes later would differ slightly.
- One asset and one parameter: 200 days is the most common version of the idea, and it was not tuned. Six years include only two full bear markets, so the sample of switches is small.
- Taxes are not included. In many countries every switch to cash is a taxable sale.
Verdict
HOLDS, narrowly: similar return to holding with a much smaller worst drop. Not financial advice.