Trend following on 27 CFD markets: every version lost, and here is why
🏛️ Giza Gold EA launch: $199 for the first 10 copies, then $499. Get it on MQL5 | Launch post | Watch Giza trade in public, every trade on the gold chart: thegoldheist.com
The short answer
We tested 20 versions of trend following on 27 markets you can trade in MetaTrader 5. All 20 lost money. The main reason: the swap, the fee your broker charges for holding a trade overnight, ate the profit.
The idea
Buy what is rising, sell what is falling, and hold for weeks or months as long as the trend lasts. Many professional funds trade this way across hundreds of markets.
How we tested it
We used 27 CFD markets: 10 stock indices, 12 currency pairs, gold, silver, copper and two kinds of oil, from 2012 to 2026. We tried three common types of trend rules with several settings each, both "buy and sell" and "buy only": 20 versions in total. Each market's position size was set by how much it usually moves, so a calm currency gets a bigger position than oil. Costs included: the spread, and the swap on the full position.
What happened
From 2012 to 2026:
- All 20 versions lost money
- The buy-and-sell versions: -11.9% to -26.4% a year
- The buy-only versions: -1.4% to -8.0% a year
- For comparison: simply holding the US500 made about +13% a year (without costs)
Where the money went
We split four versions into three parts: the profit before costs, the trading costs (spread) and the overnight fees (swap):
- 50-day breakout, buy and sell: +1.7% a year before costs, -1.4% trading costs, -14.4% swap
- 200-day moving average cross, buy and sell: +3.8%, -1.2%, -20.0%
- 200-day breakout, buy only: +12.0%, -0.5%, -9.6%
- 250-day momentum, buy only: +16.1%, -1.4%, -12.4%
Two problems. The buy-and-sell versions made almost nothing even before costs. The buy-only versions made good money before costs, mostly because stock indices and gold went up, but holding big positions for months meant paying the swap every night, and that took almost all of it. Our swap figures are on the careful side (on some currency pairs the swap is positive in one direction), but even at half the swap the buy-and-sell versions would still lose.
What it means for you
- Check the swap before you hold a trade for weeks. On a leveraged position, 4% a year on the full size is a big part of what a slow strategy can earn
- Many backtests leave the swap out or set it to zero. Ask the seller, or test with your broker's real swaps
- Strategies that hold trades for days instead of months pay much less of this fee. That is one reason why the index dip strategy from our third post survived
The Giza Gold EA holds its index dips for at most 10 days for this reason: https://www.mql5.com/en/market/product/203469
Details for those who want them
- Data: daily prices from Dukascopy, 2012 to 2026; oil adjusted for contract rolls
- Rules: channel breakouts (20, 50, 100 and 200 days), moving average crosses (100, 200 and 300 days) and momentum (60, 120 and 250 days)
- Positions sized by ATR (the average daily range) of each market
- Costs: spread and slippage 0.008% to 0.05% per side; swap 1.5% a year on currencies and 4% on indices, metals and oil, charged on buy and sell positions
Have a strategy, an indicator or an EA you want tested like this? Write it in the comments.
If one of our free indicators or tools helped you, a short review on its Market page helps other traders find it.
Next week: Tesla 3-6-9 numbers on gold and BTC.


