Moving Average Strategies That Actually Work (And Why Most Don't)
The moving average crossover is the first strategy every trader learns. It's also the first strategy most traders lose money with. The problem isn't moving averages themselves — they're genuinely useful. The problem is how they're typically used: as standalone entry signals, without context, without filters, and without understanding why they fail.
SMA vs EMA: What's the Actual Difference
A Simple Moving Average (SMA) weighs every candle equally. A 20-period SMA treats the candle from 20 periods ago the same as the most recent one. An Exponential Moving Average (EMA) gives more weight to recent prices, making it more responsive to current conditions.
The practical difference: EMA turns faster. When price reverses, the EMA will reflect it sooner. The SMA is slower but smoother. In practice, neither is universally better. EMAs work better in trending markets where you need to react quickly. SMAs work better when you want to filter out noise and stay in trades longer.
For CFD trading on 1H and 4H timeframes, EMAs are more commonly used because the edge comes from catching trends early. On daily charts, the difference matters less — both SMAs and EMAs on daily timeframes produce similar signals.
Why Most MA Crossover Strategies Fail
The classic "buy when the 50 crosses above the 200" strategy has a fundamental flaw: moving averages are lagging indicators. By the time the crossover happens, the move is often already half over. You enter late and exit late.
In trending markets, this works fine — you catch the middle of a long move. But markets trend only about 30% of the time. The other 70%, price chops sideways, and crossovers generate false signal after false signal. Each one costs you spread + commission + slippage. Ten false signals in a choppy period can wipe out the profit from one good trend.
The data is clear: pure MA crossover strategies, tested across 50+ instruments, typically show profit factors of 0.85-0.95 — meaning they lose money after costs. The win rate looks decent (45-55%), but the average win is smaller than the average loss because late entries and late exits compress the profit window.
Three MA Strategies Worth Testing
Use two MAs: a fast one (e.g., 20 EMA) for entry signals and a slow one (e.g., 50 EMA) as a trend filter. Only take long signals when price is above the 50 EMA. Only take short signals when price is below. This eliminates the majority of false crossovers that happen against the trend. Backtests show this improves profit factor by 0.15-0.25 compared to pure crossovers.
Instead of trading crossovers, trade pullbacks to a moving average. In an uptrend (price above 50 EMA), wait for price to pull back and touch the 20 EMA, then enter long. This gets you in at a better price than a crossover entry and puts your stop closer (just below the 20 EMA), improving your risk/reward. This works best on daily and 4H timeframes where pullbacks are more orderly.
Plot a 20 SMA with envelopes at +2 and -2 standard deviations (essentially Bollinger Bands). When price touches the upper band, enter short. When it touches the lower band, enter long. This is a mean reversion strategy that works in ranging markets. The key: only trade it when the ADX is below 25 (low trend strength). In strong trends, price can ride the band for extended periods and destroy a mean reversion trader.
How to Backtest MA Strategies Properly
The temptation with moving averages is to optimize the periods until you find the combination that worked best historically. This is curve-fitting, and it's the #1 reason backtested MA strategies fail live.
Proper approach:
- Test a range, not a single value — if 20 EMA works, test 18, 20, 22, 24. If the strategy is robust, all should be profitable. If only 20 works, you've curve-fit.
- Test across multiple instruments — a strategy that works on 40+ instruments with similar parameters is robust. One that only works on Gold is likely overfit.
- Use walk-forward analysis — optimize on the first 70% of data, test on the remaining 30%. If the out-of-sample performance is close to in-sample, the strategy is real.
- Include costs — MA strategies generate many trades. Test with realistic spread and commission. A strategy that's profitable at 0 spread might lose at 1.5 pip spread.
TradeTestr's Strategy Lab runs all of this automatically: 50+ instruments, multiple timeframes, Monte Carlo simulation, and walk-forward validation. If your MA strategy survives those tests, it has a genuine edge.
Optimizing Without Curve-Fitting
The rule of thumb: fewer parameters = less curve-fitting. A strategy with 2 parameters (fast MA period, slow MA period) is far harder to overfit than one with 6 (fast MA, slow MA, RSI period, RSI threshold, ADX period, ADX threshold).
If you must optimize, change one parameter at a time and look for plateaus, not peaks. If profit factor peaks at MA period 23 and drops sharply at 22 and 24, that's a spike — it won't hold. If profit factor is similar across periods 18-26, that's a plateau — it's robust.
Accept lower returns for higher robustness. A strategy that returns 15% per year across all parameter sets is better than one that returns 40% at one specific set and loses money at every other.
Related Articles
Disclaimer: CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. 70-80% of retail investor accounts lose money when trading CFDs. Backtesting does not guarantee future results. Always consider whether you can afford the potential loss of your capital.