Trend-Following Backtester · Guide · Concepts 한국어
Concepts
What is trend following? Buy strength, sell when it breaks
Trend following does not try to call the top. It buys once something is already rising and leaves when it turns. You give up being right in exchange for not being catastrophically wrong.
The mechanic: give up prediction, automate the exit
The rule fits in two lines.
- Price crosses above a reference line — buy.
- Price crosses below it — sell.
What you use as the reference line is what gives each variant its name: a moving average is SMA crossover, a volatility band is Bollinger, a recent high is Donchian. The skeleton is identical.
Note what is absent: any forecast. The strategy has no view on how high it goes or why. It only observes that it is going up.
The shape of the returns: lose often, win big
Trend following wins roughly half its trades or fewer. On Bitcoin, the best strategy here won 54% of them.
It still nets out because losers are cut small and winners are ridden. In chop it enters and exits at a small cost repeatedly; when a real trend arrives it stays for the whole thing.
Where it works and where it does not
Trend following needs trends that persist. In range-bound markets it pays costs and nothing else. The same ten strategies across five very different assets:
| Asset | Hold CAGR | Hold DD | Beat holding | Best CAGR | Best DD |
|---|---|---|---|---|---|
| Bitcoin 2017-09-25~ · 9y |
41.2% | -86.8% | 10/10 | 84.6% | -36.6% |
| Ethereum 2017-09-25~ · 9y |
27.1% | -95.6% | 10/10 | 97.2% | -48.5% |
| XRP (Ripple) 2017-09-25~ · 9y |
25.7% | -95.9% | 10/10 | 114.0% | -42.2% |
| Samsung Electronics 2000-01-04~ · 27y |
14.2% | -64.8% | 0/10 | 13.2% | -50.9% |
| Apple (AAPL) 1980-12-12~ · 46y |
18.8% | -82.2% | 1/10 | 20.4% | -72.6% |
On Samsung Electronics, holding won. In a long, grinding uptrend, the upside missed while sitting in cash exceeds the drawdown avoided. The deeper and more violent an asset's declines, the more trend following is worth.
What it actually sells
Read the drawdown columns, not the return columns. Holding Bitcoin meant surviving 86.8% from peak; the best strategy saw 36.6%.
Almost nobody holds through a 86.8% decline. They sell near the bottom and miss the recovery. What trend following sells is not the highest return — it is the probability that you are still there at the end.