BTC analysis · the best trigger we tested · passed 9 of 9 checks
The Best BTC Trigger: the 200-Day SMA, a 3% Band and Falling-SMA Sells
Five rounds, 21 rules, every one fixed before its test. The best trigger: buy on a daily close more than 3% above the 200-day SMA; sell on a daily close more than 3% below it, but only while the SMA is falling. 71.4% of trades won, +92.9% a trade, $1 → $24.71 since 2018 (held: $9.8, plain 3% band: $13.97).
200-day SMA rising: $72,898 now vs $71,462 20 days ago (+2.0%). No sell while it rises, even on a close below the sell level; that is what stops the sell-and-rebuy round trips.
Decided only on the daily close (00:00 UTC = 05:00 PKT). The levels move a little every day with the average. Spot BTC, no leverage. Not financial advice.
A daily close more than 3% above the 200-day simple moving average. Buy at that close.
close > ta.sma(close, 200) * 1.03
Sell trigger
A daily close more than 3% below it while the 200-day SMA is lower than 20 days ago. Sell at that close. While the SMA still rises: hold.
close < sma * 0.97 and sma < sma[20]
Trades won
71.4%
7 trades · 3% band 53.8%
BTC kept vs holding
2.52 BTC
per 1 BTC held · 3% band 1.43
$1 since 2018
$24.71
held $9.8 · 3% band $13.97
Worst drawdown
-54.3%
held -76.6% · 3% band -52.5%
Round 5: stop selling and buying back higher
Every time a trigger sells and then buys back at a higher price, you end up with fewer BTC. The plain 3% band did that 10 times in 12 sell-and-rebuys since 2018: in the 2019–2020 range before the 2020 run, at $43k → $48k in 2021, and three times around $57k–$66k in 2024. Together those round trips cut the stack to ×0.27; the band still ended ahead of holding only because it sat out the 2022 and 2025–26 crashes.
In 7 of those 10 sells the 200-day SMA was still rising: the long-term trend had not turned, price had only dipped. So round 5 changed the sell, not the buy: sell only while the 200-day SMA is lower than 20 days ago. On the map below the SMA turns red where that is true.
━ falling-SMA sells (the trigger)┈ plain 3% band╌ held 1 BTC the whole timebelow 1 = fewer BTC than holding, above 1 = more
Plain 3% band: 10 of 12 sell-and-rebuys lost BTC
Sold
Bought back
BTC kept
2019-09-26 $8,064
2019-10-26 $9,230
-12.9%
2019-11-08 $8,774
2020-01-28 $9,374
-6.7%
2020-03-08 $8,033
2020-04-29 $8,779
-8.8%
2021-05-19 $36,690
2021-08-13 $47,800
-23.5%
2021-09-20 $43,016
2021-10-01 $48,142
-10.9%
2021-12-31 $46,217
2023-01-14 $20,955
+119.9%
2023-08-18 $26,054
2023-10-20 $29,669
-12.5%
2024-07-05 $56,629
2024-07-15 $64,724
-12.8%
2024-08-04 $58,161
2024-09-27 $65,770
-11.8%
2024-10-01 $60,806
2024-10-14 $66,084
-8.3%
2025-03-09 $80,734
2025-04-22 $93,443
-13.9%
2025-11-04 $101,497
2026-08-20 $73,025
+38.6%
Falling-SMA sells (the trigger): 4 of 6 sell-and-rebuys lost BTC
Sold
Bought back
BTC kept
2019-12-12 $7,198
2020-01-28 $9,374
-23.4%
2020-03-08 $8,033
2020-04-29 $8,779
-8.8%
2021-09-20 $43,016
2021-10-01 $48,142
-10.9%
2022-02-28 $43,160
2023-01-14 $20,955
+105.4%
2024-10-01 $60,806
2024-10-14 $66,084
-8.3%
2025-11-26 $90,484
2026-08-20 $73,025
+23.5%
Rule
P1
P2
P3
P4
P5
P6
P7
P8
P9
Lost-BTC trips
Won
$1 →
Worst fall
2012–17
ETH $1 →
Beat band on coins
Plain 3% band (round 3)
·
·
·
·
·
·
·
·
·
10 of 12
53.8%
×13.97
-52.5%
×1,440
×11.64
—
3% band, sell only while the SMA is fallingthe trigger · 9/9
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
4 of 6
71.4%
×24.71
-54.3%
×2,886
×5.66
5 of 8
3% band, buy only while the SMA rises, sell only while it falls6/9
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
✗ failed
✓ passed
✓ passed
✗ failed
3 of 5
66.7%
×9.29
-62%
×5,057
×1.29
4 of 8
P1 = Average trade above zero after costs in 2018–22, 2023–26 and 2012–17
P2 = Profit factor at least 1.5 (2018–today)
P3 = Smaller worst drawdown than buying and holding
P4 = Both neighbour settings (lookback ×0.75 and ×1.25) also make money
P5 = At most 2 losing calendar years 2018–2026
P6 = Beats the 3% band on return per unit of drawdown (2018–today) and on Sharpe (2012–17)
P7 = Works on ETH too (2018–today): average trade above zero and a smaller drawdown than holding ETH
P8 = Fewer sell-and-rebuy round trips that lost BTC than the 3% band (rebuy above the sell after costs)
P9 = More money than the 3% band on at least 5 of 8 coins no rule was built on (LTC, XRP, BNB, ADA, TRX, XLM, ETC, NEO)
2012–17 = what $1 became on Coin Metrics closes (held: ×2,629); "Beat band on coins" = of the 8 untouched coins.
Untouched coin, 2018-03-25 → 2026-10-20
Held
Plain 3% band
Falling-SMA sells
Worst fall (trigger / held)
LTC/USDT
×0.39
×0.83
×0.95
-85.8% / -89.4%
XRP/USDT
×1.50
×0.49
×4.05
-79.3% / -85%
BNB/USDT
×57.27
×17.38
×45.86
-69.2% / -76.1%
ADA/USDT
×0.92
×13.88
×13.53
-72.3% / -95.2%
TRX/USDT
×6.85
×3.54
×7.03
-75.8% / -83.1%
XLM/USDT
×0.62
×2.29
×2.20
-78.5% / -90.5%
ETC/USDT
×0.54
×0.25
×0.51
-89.6% / -95.5%
NEO/USDT
×0.03
×0.93
×0.42
-89.5% / -98.6%
What it fixed: 10 round trips that lost BTC became 4. Gone: Sep 2019 $8.1k → Oct 2019 $9.2k, Nov 2019 $8.8k → Jan 2020 $9.4k, May 2021 $37k → Aug 2021 $48k, Aug 2023 $26k → Oct 2023 $30k, Jul 2024 $57k → Jul 2024 $65k, Aug 2024 $58k → Sep 2024 $66k and Mar 2025 $81k → Apr 2025 $93k. $1 since 2018 became ×24.71 instead of ×13.97 (2.52 BTC per BTC held instead of 1.43).
What it did not fix, and what it costs: Dec 2019 $7.2k → Jan 2020 $9.4k, Mar 2020 $8.0k → Apr 2020 $8.8k, Sep 2021 $43k → Oct 2021 $48k and Oct 2024 $61k → Oct 2024 $66k are still round trips that lost BTC: the SMA was already falling then, so the sell was allowed, and price came back. In crashes it sells later, because the SMA turns down weeks after the top: it sold the 2021–22 top on 2022-02-28 and the 2025 top on 2025-11-26, at lower prices than the band. Its worst drawdown is -54.3% (Aug 2019 → May 2020) against -52.5% for the band. On ETH it made ×5.66, less than the band's ×11.64 (still more than holding ETH), and it beat the band on 5 of 8 other coins, just the pass mark. This is the fifth look at 2018–2026 BTC data, so no further rule search on it.
The same trigger on ETH
No rule was designed on ETH: the same rule on ETH's own 200-day SMA, 2018-03-25 to 2026-10-20, after the same costs. On ETH the plain 3% band made more (×11.64).
200-day SMA rising: $2,157 now vs $2,117 20 days ago (+1.9%). No sell while it rises, even on a close below the sell level; that is what stops the sell-and-rebuy round trips.
Decided only on the daily close (00:00 UTC = 05:00 PKT). The levels move a little every day with the average. Spot ETH, no leverage. Not financial advice.
What your money would be now
Following the trigger
$24,712
Buying and holding
$9,802
From the first weekly close of 2018 (2018-03-25) to 2026-10-20, after 0.15% costs on every trade. The trigger ends ahead of holding when started in 2018, 2019, 2020, 2021, 2022, 2024, 2025, 2026, and behind when started in 2023: it wins by dodging crashes and loses some of every rally.
$1 in the trigger against $1 held
After 0.1% fee and 0.05% slippage on every buy and sell. In BTC 66% of the days, in cash the rest: that is how it skipped the 2018 and 2022 crashes.
━ $1 in the rule, after costs╌ $1 bought and held┈ plain 3% bandlog scale, weekly
Every buy and sell on the live chart
○ connecting…BTC $83,028IN since 20 Aug 2026 at $73,025 (+13.7%) · sell below $70,509 once the SMA falls200-day SMA $72,690 (rising +1.9% vs 20 days ago)
|time|price23 Feb 2018 – 19 Nov 2026
20 Oct 2026 · BTC $83,028 · 200-day SMA $72,690 (+14.2%) · buy line $74,871 · sell line $70,509 · SMA rising: hold IN the market · trade bought 2026-08-20 at $73,025, still holding
drag to pan · drag the price scale to zoom the price · pinch, ctrl + scroll or ＋/－ to zoom · double-click resets╌ 200-day SMA━ SMA falling: sells allowed┈ buy line +3% / sell line −3%▲ buy▼ sellhollow = made by the daily closes after the backtest · shade = holding (green won, red lost) · levels from closed days, like the card
Round 4: why the plain band sold and bought back in the same zone
After a run BTC often goes sideways for months (accumulation). The 200-day SMA flattens, price swings 10–20% either side of it, and every swing crosses the ±3% lines, so the plain band sells and then buys back near the same price, as in late 2019 and early 2020 below. It cannot tell accumulation from the start of the next run; it only sees the close against the line. Round 4 tried fixing the buy; round 5 (above) fixed the sell, and that is the trigger now.
Since 2018 it bought back within 90 days of a sell 10 times. 6 of those lost, and they are every losing trade it made. The other 4 won, and they include +317% from 2020-04-29 at $8,779 and +90% from 2023-10-20 at $29,669: the 2020 buy back was in the same zone as the March 2020 sell, and it was the best trade of all.
Sold
Bought back
Days later
Price vs the sell
Result of the buy back
2019-09-26 $8,064
2019-10-26 $9,230
30
+14.5%
-5.2%
2019-11-08 $8,774
2020-01-28 $9,374
81
+6.8%
-14.6%
2020-03-08 $8,033
2020-04-29 $8,779
52
+9.3%
+316.7%
2021-05-19 $36,690
2021-08-13 $47,800
86
+30.3%
-10.3%
2021-09-20 $43,016
2021-10-01 $48,142
11
+11.9%
-4.3%
2023-08-18 $26,054
2023-10-20 $29,669
63
+13.9%
+90.3%
2024-07-05 $56,629
2024-07-15 $64,724
10
+14.3%
-10.4%
2024-08-04 $58,161
2024-09-27 $65,770
54
+13.1%
-7.8%
2024-10-01 $60,806
2024-10-14 $66,084
13
+8.7%
+21.8%
2025-03-09 $80,734
2025-04-22 $93,443
44
+15.7%
+8.3%
The fix we tested: buy back only when price leaves the range
Round 4 turned that into a rule before testing it: keep the band's sell, but buy only on a close that is also above the highest close of the previous 90 days (amber line), the classic sign that a range is over. A second version buys only while the 200-day SMA is rising. Both were fixed in advance with two neighbour settings and had to pass eight checks: round 3's seven with P6 = beat the band, plus P8 = fewer losing trades than the band.
Plain 3% band, Jun 2019 – Sep 2020
━ BTC daily close (log)╌ 200-day SMA┈ buy line +3% / sell line −3%▲ buy▼ sellgreen shade = winning hold, red = losing one · hover a marker
Band + buy only above the 90-day closing high (round 4), same months
━ BTC daily close (log)╌ 200-day SMA┈ buy line +3% / sell line −3%━ range top (highest close of the prior 90 days)▲ buy▼ sellgreen shade = winning hold, red = losing one · hover a marker
Rule
P1
P2
P3
P4
P5
P6
P7
P8
Losing trades
Won
Avg trade
Worst fall
$1 →
2012–17 Sharpe
ETH $1 →
Plain 3% band
·
·
·
·
·
·
·
·
6 of 13
53.8%
+37.5%
-52.5%
×13.97
1.90
×11.64
3% band, buy only on a 90-day closing high7/8
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
✓ passed
✓ passed
4 of 10
60%
+40.6%
-61.8%
×9.82
1.89
×6.27
3% band, buy only while the SMA is rising6/8
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
✗ failed
✓ passed
✓ passed
5 of 11
54.5%
+31.8%
-66.6%
×5.70
2.03
×2.66
P1 = Average trade above zero after costs in 2018–22, 2023–26 and 2012–17
P2 = Profit factor at least 1.5 (2018–today)
P3 = Smaller worst drawdown than buying and holding
P4 = Both neighbour settings (lookback ×0.75 and ×1.25) also make money
P5 = At most 2 losing calendar years 2018–2026
P6 = Beats the 3% band on return per unit of drawdown (2018–today) and on Sharpe (2012–17)
P7 = Works on ETH too (2018–today): average trade above zero and a smaller drawdown than holding ETH
P8 = Fewer losing trades than the 3% band (2018–today)
Plain band buy
Band result
Breakout version
Its result
2019-04-02 $4,857
+65.5%
2019-04-02 $4,857
+65.5%
2019-10-26 $9,230
-5.2%
skipped: no 90-day closing high before the sell
—
2020-01-28 $9,374
-14.6%
2020-01-30 $9,513
-15.8%
2020-04-29 $8,779
+316.7%
2020-06-01 $10,201
+258.6%
2021-08-13 $47,800
-10.3%
2021-08-13 $47,800
-10.3%
2021-10-01 $48,142
-4.3%
2021-10-06 $55,315
-16.7%
2023-01-14 $20,955
+24.0%
2023-01-20 $22,667
+14.6%
2023-10-20 $29,669
+90.3%
2023-10-22 $29,992
+88.2%
2024-07-15 $64,724
-10.4%
skipped: no 90-day closing high before the sell
—
2024-09-27 $65,770
-7.8%
skipped: no 90-day closing high before the sell
—
2024-10-14 $66,084
+21.8%
2024-10-18 $68,428
+17.6%
2025-04-22 $93,443
+8.3%
2025-05-08 $103,262
-2.0%
2026-08-20 $73,025
+13.5% open
2026-08-21 $78,338
+5.8% open
It did what it was meant to: 6 losing trades became 4, and it skipped the Oct 2019, Jul 2024 and Sep 2024 buy backs. But the range top is often far above the buy line, so it also waited: the 2020 run was bought on 2020-06-01 at $10,201 instead of 2020-04-29 at $8,779 (+259% instead of +317%), and breakouts fail too (2020-01-30 -15.8%, 2021-08-13 -10.3%, 2021-10-06 -16.7%, 2025-05-08 -2.0%). Net since 2018: $1 → ×9.82 instead of ×13.97, worst fall -61.8% instead of -52.5%, ETH ×6.27 instead of ×11.64. It passed 7/8 and missed P6, so round 4 kept the band. The rising-SMA version did worse (×5.70).
The lesson: you only know a range was accumulation after price leaves it. Buying back inside the range costs a few small losses (-4% to -15%); waiting for the breakout costs part of every run. On BTC since 2018 the small losses were the cheaper price than waiting for the breakout; round 5 found the cheaper fix, not selling while the trend still rises. When the trigger is in cash, the live card shows the 90-day range top, so you can see whether a buy would be a buy back inside the range.
Round 3: four fixes for the whipsaws, seven checks
Each fix was fixed in advance with its textbook setting and two neighbours, and had to pass seven checks: round 2's five, beat the plain rule (P6) and work on ETH (P7). Binance 2018-03-25 to 2026-10-20 after costs unless marked.
Trigger
P1
P2
P3
P4
P5
P6
P7
Won
Avg trade
Trades
Worst fall
$1 →
Losing yrs
ETH $1 →
Plain 200-day SMA (round 2)
·
·
·
·
·
·
·
25.8%
+15.4%
31
-64.5%
×11.54
2
×15.61
200-day SMA with a 3% bandheadline until round 5 · 6/7
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
✓ passed
53.8%
+37.5%
13
-52.5%
×13.97
1
×11.64
200-day SMA, 3 closes to confirmpassed 7/7
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
41.2%
+26.4%
17
-55.3%
×10.50
1
×12.02
Weekly close vs 21-week EMA
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
✓ passed
✓ passed
52.6%
+26.3%
19
-48%
×12.52
3
×10.67
200-day SMA, only while it is rising
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
✗ failed
✓ passed
25.9%
+12.3%
27
-72.9%
×4.55
3
×3.37
Buy and hold
-76.6%
×9.80
4
×4.68
P1 = Average trade above zero after costs in 2018–22, 2023–26 and 2012–17
P2 = Profit factor at least 1.5 (2018–today)
P3 = Smaller worst drawdown than buying and holding
P4 = Both neighbour settings (lookback ×0.75 and ×1.25) also make money
P5 = At most 2 losing calendar years 2018–2026
P6 = Beats the plain 200-day rule on return per unit of drawdown (2018–today) and on Sharpe (2012–17)
P7 = Works on ETH too (2018–today): average trade above zero and a smaller drawdown than holding ETH
The band led after round 3 because it won the most: highest win rate, highest average trade, most money and one losing year. It missed P6 by a hair (2012–2017 Sharpe 1.9 against the plain rule's 1.94). The rule that passed all seven was 3 closes in a row above the line: 41.2% won, +26.4% a trade, worst drawdown -55.3%. Round 5 then kept the band's buy and changed its sell.
Four tests, two of them on data no rule was built on
Test
Trigger
Held
A year (trigger / held)
Worst fall
Sharpe
Trades
Won
Avg trade
BTC 2012–2017 (Coin Metrics closes)
×2,886
×2,629
277.4% / 271.6%
-71.2% / -84.5%
2.01 / 1.92
7
28.6%
+2479.5%
BTC 2018-03-25 – 2022
×5.51
×1.95
43% / 15.1%
-54.3% / -76.6%
0.94 / 0.56
4
50%
+102.8%
BTC 2023 – 2026-10-20
×4.48
×5.00
48.4% / 52.7%
-32% / -53%
1.16 / 1.14
3
100%
+79.8%
ETH 2018-03-25 – 2026-10-20
×5.66
×4.68
22.4% / 19.7%
-74.8% / -89.8%
0.64 / 0.64
10
40%
+133.3%
Multiples are what $1 became. In 2012–2017 it made more than holding (×2,886 vs ×2,629); in the 2023–2026 bull market it made less (×4.48 vs ×5.00). Its edge is the crashes it sat out.
Year by year
2018
2019
2020
2021
2022
2023
2024
2025
2026
Falling-SMA sells (headline)
cash
+48%
+181%
+42%
-7%
+102%
+103%
-3%
+14%
Plain 3% band
cash
+57%
+181%
+9%
cash
+76%
+56%
-7%
+14%
3 closes (passed 7/7)
cash
+63%
+163%
+2%
cash
+86%
+33%
-8%
+6%
Plain 200-day SMA
cash
+57%
+173%
-22%
cash
+94%
+81%
-18%
+20%
Buy and hold
-56%
+94%
+302%
+60%
-64%
+156%
+121%
-6%
-5%
2018 starts on 2018-03-25; 2026 runs to 2026-10-20. "cash" = no trade that year. The trigger lost money in 2 years (2022 -7%, 2025 -3%); holding lost in 4.
All 7 trades
Buy (close > line +3%)
Sell (close < line −3%, SMA falling)
Days
Result after costs
2026-08-20 $73,025
still holding (last close $83,028)
61
+13.5%
2024-10-14 $66,084
2025-11-26 $90,484
408
+36.5%
2023-01-14 $20,955
2024-10-01 $60,806
626
+189.3%
2021-10-01 $48,142
2022-02-28 $43,160
150
-10.6%
2020-04-29 $8,779
2021-09-20 $43,016
509
+388.5%
2020-01-28 $9,374
2020-03-08 $8,033
40
-14.6%
2019-04-02 $4,857
2019-12-12 $7,198
254
+47.8%
Pine Script: the trigger as a TradingView strategy
Paste into the Pine Editor on a BINANCE:BTCUSDT daily chart and open the Strategy Tester. It fills on the signal candle's close like the test, with a 0.1% commission (add slippage in the settings to match the 0.05% used here), draws the buy and sell lines and sends one alert on each buy and sell.
//@version=6
strategy("DanioX BTC trigger: 200-day SMA ±3%, sell only on a falling SMA", overlay = true, initial_capital = 1000,
default_qty_type = strategy.percent_of_equity, default_qty_value = 100,
commission_type = strategy.commission.percent, commission_value = 0.1, process_orders_on_close = true)
// daniox.com/bitcoin-analysis/best-trigger · use on BINANCE:BTCUSDT, 1D.
// Buy on a daily close more than 3% above the 200-day SMA. Sell on a daily close more than 3% below it, but only while the
// SMA is lower than 20 days ago (no sells while the long-term trend still rises). Spot, no leverage.
len = input.int(200, "SMA length")
bandPct = input.float(3.0, "Band %", step = 0.25)
slopeLen = input.int(20, "SMA falling vs N days ago")
sma200 = ta.sma(close, len)
buyLine = sma200 * (1 + bandPct / 100)
sellLine = sma200 * (1 - bandPct / 100)
falling = sma200 < sma200[slopeLen]
var bool holding = false
buySig = not holding and close > buyLine
sellSig = holding and close < sellLine and falling
if buySig
holding := true
strategy.entry("Long", strategy.long)
alert("BTC closed more than 3% above its 200-day SMA: buy trigger", alert.freq_once_per_bar_close)
if sellSig
holding := false
strategy.close("Long")
alert("BTC closed more than 3% below its falling 200-day SMA: sell trigger", alert.freq_once_per_bar_close)
plot(sma200, "200-day SMA", falling ? color.red : color.aqua, 2)
plot(buyLine, "Buy line", color.new(color.lime, 40), 1, plot.style_circles)
plot(sellLine, "Sell line", color.new(color.red, 40), 1, plot.style_circles)
plotshape(buySig, "Buy", shape.triangleup, location.belowbar, color.lime, size = size.small)
plotshape(sellSig, "Sell", shape.triangledown, location.abovebar, color.red, size = size.small)
Round 2: the five trend triggers that came first
Round 2 compared five trend-following triggers on five checks; only the plain 200-day SMA rule passed, and round 3 improved on it.
Trigger
P1
P2
P3
P4
P5
$1 since 2018
Worst fall
Trades
Avg trade
Close vs 200-day SMA
✓ passed
✓ passed
✓ passed
✓ passed
✓ passed
×11.54
-64.5%
31
+15.4%
EMA 50/200 golden cross, hold to death cross
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
×7.68
-71.4%
8
+59.1%
Turtle breakout: 55-day high in, 20-day low out
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
×9.66
-38.7%
30
+11.2%
20-day breakout, midpoint trailing stop
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
×10.24
-57.9%
71
+4.3%
90-day momentum (price above 90 days ago)
✓ passed
✓ passed
✓ passed
✓ passed
✗ failed
×8.29
-66.8%
57
+10.8%
Why it works on BTC: a band against chop, a falling SMA against rebuys
Bitcoin trends for months, but around the 200-day average it chops: price pokes through the line for a day or two and falls back. The plain rule traded every poke, and 19 of its 23 losing trades lasted a week or less. A 3% band on each side of the line ignores the chop and only acts when price has really left it. The band still sold every deeper dip in a bull market and bought back higher; waiting for the 200-day SMA itself to turn down before selling skips those dips and still sells the real bear markets. So the trigger trades 7 times since 2018 and wins 71.4% of them.
5 winning trades paid for 2 losing ones; the biggest winners made +389% (2020-04), +189% (2023-01), +48% (2019-04), +37% (2024-10), and the worst loser lost -14.6%.
What it costs you
Later exits. The 200-day SMA turns down weeks after a top, so the trigger sells bear markets later and lower than the plain band, and it holds through deep dips while the SMA still rises (worst trade -14.6%).
Drawdowns still happen. The trigger's worst fall was -54.3% (holding: -76.6%; plain band: -52.5%), from Aug 2019 to May 2020.
Bull markets. In 2023–2026 it made ×4.48 while holding made ×5.00. No trend rule tested beat holding in that bull market; the edge is the crashes it sits out.
ETH and other coins. On ETH it made ×5.66 against ×11.64 for the plain band, and it beat the band on 5 of 8 other coins, just the pass mark.
How the test was done
Round 1 (analysis page) tested 8 indicator setups with a fixed 2R target: none made real money. Round 2 tested 5 trend triggers that hold the trend; the plain 200-day rule passed. Round 3 tested 4 fixes of its whipsaws with neighbours and seven checks. Round 4 tested 2 ways to stop buying back inside an accumulation range (eight checks); neither beat the band. Round 5 tested 2 sell-side filters that stop selling while the 200-day SMA rises (nine checks, one on 8 untouched coins); the falling-SMA sell passed all nine and is the trigger.
Long-only spot BTC, decided and filled on the daily close, 0.1% fee + 0.05% slippage on every buy and sell. Settings were fixed before each run and never changed after.
Rounds 3 to 5 are the third to fifth looks at Binance 2018–2026, so the strongest evidence is the tests nothing was designed on: BTC 2012–2017 (Coin Metrics closes), ETH 2018–2026 and the 8 coins added in round 5. Scripts: signals/bot/btc_trigger_research.py and btc_trigger_round3.py to btc_trigger_round5.py.
Frequently asked questions
What is the best Bitcoin trigger you found?+
The 200-day simple moving average with a 3% band and one more condition on the sell: buy on a daily close more than 3% above the average; sell on a daily close more than 3% below it, but only while the average itself is lower than 20 days ago. From 2018-03-25 to 2026-10-20 on Binance BTC/USDT, after 0.15% costs a side, 71.4% of its 7 trades won, the average trade made +92.9% (profit factor 26.83), and $1 became $24.71 against $9.8 for holding and $13.97 for the plain 3% band, with a worst drawdown of -54.3% against -76.6% held.
Why sell only while the 200-day SMA is falling?+
Because most of the plain band's sells came while the long-term trend was still rising, and BTC was back above the buy line within weeks: 10 of its 12 sell-and-rebuys bought back higher than they sold, each one leaving fewer BTC (together ×0.27 of the stack). Waiting for a falling SMA skips the dips in an uptrend and still sells the real bear markets: 4 of 6 round trips lost BTC, and the neighbour settings (15 and 25 days) made ×22.48 and ×22.24.
Did it pass every check?+
Yes, all nine set before the run, including two on data no rule was built on: BTC 2012-2017 (Sharpe 2.01 vs 1.9 for the band; it also beat holding there, ×2,886 vs ×2,629) and 8 other coins (more money than the band on 5 of 8, just the pass mark). It is the fifth look at the 2018-2026 BTC data, so treat the untouched tests as the real evidence.
Does it beat buying and holding Bitcoin?+
From March 2018, yes (×24.71 vs ×9.80, worst drawdown -54.3% vs -76.6%), because it sat in cash through most of the 2018 and 2022 crashes. In the 2023-2026 bull market it made ×4.48 while holding made ×5.00. On ETH it made ×5.66 against ×4.68 for holding ETH, but less than the plain band there (×11.64).
Why did the old trigger sell and buy back at the same price?+
Because BTC goes sideways after a run and price crosses the 3% lines both ways. The plain band bought back within 90 days of a sell 10 times: 6 lost and 4 won. Round 4 tried waiting for a 90-day closing high before buying back (less money: ×9.82 vs ×13.97); round 5 instead stopped the sells while the trend still rises, which removed most of those round trips and made ×24.71.
Is this financial advice?+
No. It is a historical test on one asset, and the fifth look at the 2018-2026 data, so the 2012-2017, ETH and 8-coin results are the stronger evidence. Past results do not guarantee future ones. Spot only, no leverage.