Updated: Apr 25
Putting an Old system to the Test!
“Without data, you're just another person with an opinion.”
― W. Edwards Deming
3 SMA Trend Following Swing System
One of the first strategies that most traders started trading with, usually comprises of some form of moving averages. Moving averages are great statistical tools used to gauge trend and direction, but can one be profitable using a systems that only uses only Moving Averages?
Lets test an old system that I used in the past. The rules are as follows:
Test period: Jan 2009 to March 2022
GBPJPY / M30timeframe
Starting Capital: 10,000
Risk per trade: 5% of balance
Market position is flat (No open position)
Medium MA (70 SMA) is above Slow MA (200 SMA)
Fast MA (20SMA) crossed above 70 SMA
Open Long Order with 5% risk
Move Stop Loss to Breakeven at 100pips
Take Profit: 400pips
Stop Loss: 150Pips
Exit when 20SMA crossed back below 70SMA
Short rules are the opposite of Rules for Long
These are the test outcomes.
No. of trades: 984 trades
Avg win: 1,416.80 / Avg Loses: $600.12
Max Consec Wins: 5 / Max Consec Losses :10
Avg Consec wins: 1.5 / Avg Consec Loss: 2.55
Stagnation in Days: 755 / 15.62%
Winning Percentage: 35.31%
This is a Long and Short Model. There are slightly more short trades (almost 8% more shorts).
Trades by Year, excluding 2022, is between 65 to 79 per year.
There are a total of 4 losing years since 2009. 9/13 profitable years (excluding 2022, but its profitable so far).
Trades that lasted more than 4 days or more, are overwhelmingly profitable.
Trades triggered between 08:00 to 10:00 suffered the most losses. These are the London Market opening hours. These could be due to the high volatility during first 1-2 hrs of London Open. Moving average cross over entry might not be a good trigger.
Trades by Duration: 4 days is the mode
Average Holding periods are 66.5hrs (2.8 days)
Overall, this moving average system seemed to work. But do bear in mind that it has only abt 35% winning rate. Meaning for every 100 trades, you will only have 35 profitable trades.
Distribution of losers are also non linear. One could over 10 losing trades in a roll before encountering a winner.
Another one of the key reason it worked is because this was tested on GBPJPY, Yen pairs tend to trend, especially so for GBPJPY. Applying a moving average based trend following style system on other currencies might not have the same outcome.
I test the same system with the exact same rules (no optimization done) on CHFJPY, EURJPY and USDJPY and these are the outcome.
CHFJPY. Worked well from mid 2009 to early 2011 / 2017 to Current
EURJPY. Pretty similar curve to GBPJPY.
USDJPY. The system do not seemed to be going anywhere on USDJPY.