Trend-Following Backtester · Guide · Strategies 한국어

Strategies

EMA crossover

Uses two exponential averages (more weight on recent prices) and trades their crossover.

Parameters 2 Assets tested 5 Costs 0.5% Optimized Sharpe · 120

The rules

Those numbers are the values optimized for Bitcoin. They differ per asset — the table below lists each.

There are 2 tunable values: fast, slow. Fewer knobs make it harder to fit noise, so a count this low keeps overfitting risk relatively contained.

Results by asset

The same strategy across 5 very different assets, each over its full history, with parameters optimized per asset. It beat buy and hold on 3 of 5.

Asset · parameters CAGRHold CAGR Max drawdownHold DD SharpeTradesTime in market
Bitcoin
fast=3, slow=25
70.7% 41.2%-57.6%-86.8% 1.4615651%
Ethereum
fast=5, slow=26
85.9% 27.1%-51.4%-95.6% 1.3912150%
XRP (Ripple)
fast=8, slow=11
67.3% 25.7%-62.5%-95.9% 1.0016841%
Samsung Electronics
fast=33, slow=36
11.8% 14.2%-45.6%-64.8% 0.5510457%
Apple (AAPL)
fast=5, slow=149
17.9% 18.8%-85.8%-82.2% 0.6918364%
1x 10x 100x 2018 2020 2022 2024 2026 EMA crossoverBuy & hold
Bitcoin 2017-09-25–2026-07-30 · EMA crossover (solid) vs buy and hold (dashed). Log scale.
These are in-sample figures. The same data chose the parameters and scored them, so they sit above what live trading would return. Use them to rank strategies against each other, not as a return target. See optimization and overfitting.

When this one works

EMA crossover belongs to the trend-following family. All of them earn in sustained directional moves and bleed costs in range-bound markets. The trade count and time in market columns above show how each variant leans.

Run this strategyCompare the others

Read next