Trend-Following Backtester · Guide · Concepts 한국어
Concepts
Why not to put it all in one asset — what diversification actually buys
It is tempting to put everything into whichever asset backtested best. The reason not to has nothing to do with returns and everything to do with what is left when that asset has a bad decade.
“Best” is a statement about the past
Here is the top strategy for each asset on this site. Notice that the winner is different every time.
| Asset | Best CAGR | Its drawdown | Winning strategy |
|---|---|---|---|
| Bitcoin | 84.6% | -36.6% | Bollinger band breakout |
| Ethereum | 97.2% | -48.5% | Supertrend |
| XRP (Ripple) | 114.0% | -42.2% | Bollinger band breakout |
| Samsung Electronics | 13.2% | -50.9% | Donchian channel breakout |
| Apple (AAPL) | 20.4% | -72.6% | Donchian channel breakout |
If the winner differs by asset, there is no basis for assuming the same pairing wins next. Betting everything on the historical winner is the largest forecast you can make — inside a discipline you adopted specifically to stop forecasting.
What diversification reduces
It is not a return booster. It usually lowers your best-case return. What it lowers more is:
- The damage when one asset breaks. The deepest drawdown above is 72.6%. Concentrated, you take all of it.
- How long a bad stretch lasts. Trend following bleeds in range-bound markets. Different assets range at different times.
- The chance you quit. Which is the one that matters. A great strategy abandoned halfway returns nothing.
How to weight
Simple rules kept beat sophisticated weights abandoned.
- Equal weight — 1/N each. Dull and durable.
- Periodic rebalancing — winners grow into concentration on their own. Reset monthly.
- Band rebalancing — only correct when drift exceeds a threshold, which cuts trading costs.
Computing “optimal” weights from history is overfitting wearing a different hat — one more thing fitted to a past that already happened.
Cash is a position
A trend-following strategy sits in cash whenever there is no signal — that is the “time in market” column on every page here. Run several assets and some are long while others are flat, automatically.
That is a second layer of diversification: not just across assets, but across how exposed you are at all.