Trend Following EMA(20/50/200): Tested on 12 Pairs

Fast average above slow, and price above the long-term average, defines an uptrend to be traded long. Mirrored for shorts. The canonical trend-following construction with triple confirmation.

The trend-following stack: entry conditions

Long when EMA(20) is above EMA(50) and price is above EMA(200); short on the mirror image; flat otherwise. The triple condition makes the rule considerably more selective: around eleven trades over the sample and roughly half the time out of the market.

Trend following: numbers across 12 pairs

Run with the portal's own engine over 518 daily bars for each of 12 pairs, costs of 1 pip round turn, canonical parameters. The method and its limitations are set out in how we run backtests.

• Median return — -1.67%, profitable on 4 of 12 pairs
• Sharpe — -0.13, drawdown — 5.5%
• Trades — 11, win rate — 19%, time at risk — 50%
• Extremes: EURJPY +13.9%, NZDUSD -9.9%

Filters cut the loss without creating an edge

A mildly negative median with four profitable pairs. The filters did their job — drawdown is noticeably lower than for simple crossovers and the number of false entries fell. But they cannot remove the central problem of trend systems: over a two-year window on a single pair, a sustained trend either happened or it did not. A spread of twenty-four percentage points between best and worst pair is the answer to that question.

Why trend systems are not tested this way

Trend-following systems are not evaluated like this. They need a portfolio of instruments and a horizon measured in years, because their profit concentrates in rare strong moves and they lose in between. Our run neither refutes nor confirms the approach — it shows that the available sample is insufficient for such a conclusion, and any claim about the method based on it would be an overstatement. For comparison: simply holding the position on the same data returned 4.79%, and this rule did not beat it — why a benchmark is not optional.

The horizon limitation

Without this caveat the conclusion would be wrong: the profit of trend systems concentrates in rare strong moves, so a two-year window on one pair simply gives them no opportunity to appear. A negative result here is evidence about the sample, not about the method. The general limitations — a short sample, correlated pairs, a signal tested without risk management — are listed in the methodology. The data is open and the run reproduces from a script in the repository.

This material is educational and is not individual investment advice. Backtested results do not guarantee similar results in the future. Trading forex carries the risk of losing capital.