At this point, let me tell you about the process and steps of writing this EA. First, we’ll test Backtest on a single currency pair, USD/CHF with $1,000 capital on BDSwiss broker. The backtest system setup details are summarized inTable 1

Table 1 Backtest system setup summary

From the Backtest results, we found that the profit factor was quite higher than expected. The average annual growth rate was not bad. Additionally, the risk wasn’t too high (Table 2)

Table 2 shows Metrics values from Backtesting

*CARG = Compound Annual Growth Rate or average annual portfolio growth rate with compounding, calculated by Quant Analyzer Free V.4.9.2 software

** In real trading, Relative DD values are lower due to control methods beyond Backtesting, which will be discussed in the next section

After inputting the Backtest results into myfxbook, we found monthly profits of 3.51% with only 33% Drawdown. Total profit over 15 years reached an impressive 84,131% … So why does monthly profit seem low?

The answer is because it’s an EA that adjusts Lotsize according to capital, creating a compounding effect until eventually reaching Lotsize saturation for each broker. BDSwiss broker has a Max Lot of only 50 Lot (simply put, it reached max lot, so profits were lower in later years)

Looking at order holding duration, we see that Wisdom closes orders as quickly as 10 minutes and holds orders for a maximum of only 4 hours which makes it a decent Scalping EA for Forex trading, suitable for people who don’t have time to monitor frequently

Additional Market Testing is one of the Robustness tests specifically designed to test EA Overfitting. Originally, Wisdom was developed to trade USDCAD pair, which in testing we call In Sample Data (IS)

Therefore, we add new data sets called Out of sample Data (OOS). We’ll add 5 more OOS currency pairs: USDCHF, NZDUSD, AUDCAD, AUDUSD, and EURCHF, each for 15 years

The Additional Market Testing results were excellent because Wisdom performed well on all 6 currency pairs without blowing the account. The highest Drawdown period was during 2013-2016. Additionally, Wisdom can work well on 15-18 other currency pairs

Stress Test is testing the EA under high-risk conditions to evaluate performance and flexibility . This stress test is divided into 2 modes: Crisis History mode and Random Delay mode

Crisis History mode testing

This test runs the EA during 3 economic crisis periods that affected Forex trading:

• The Great Recession in 2008 to 2009
• The European Crisis in 2010
• COVID-19 Crisis in 2020

We excluded the Ukraine war period since it passed easily. The EA performance during various crisis periods with $300 capital shows that Wisdom performed very well, and we’ll summarize the results in a table for easy understanding

Table 3 Backtest testing during various economic crises on BDSwiss broker Classic account type

Random Delay mode testing

We conducted Random Delay mode testing to identify factors affecting Wisdom EA’s trading. We initially hypothesized that signal transmission speed affects our EA since it’s a Scalping type EA

This test uses Backtest with Tick data suite software using 99.9% data from dukascopy bank, which isthe most comprehensive data availablein MT4

Table 4 Backtest testing on BDSwiss broker Classic account type

a = value is 1-60

b = the market order delay is 20-60 / the pending order delay is 20-60

c = the market order delay is 170-200 / the pending order delay is 100-150

From the Backtest results, we can see very satisfactory numbers. The win rate when placing orders is very high. Additionally,profits reached 1285% – 3483% while Relative DD remained low at only 21.65% – 26.86% which is considered standard EA level if it doesn’t exceed 30-40%

Performance analysis from various Metric parameters

FromTable 4 we can see 3 Tests with different details to demonstrate differences in each technique:

• Test 1 is standard backtesting with Fixed spread and no Optimized Slippage. What makes it special compared to many others is we use 99.9% tick data and simulate the trading account realistically
• Test 2 is backtesting with additional features from Test 1 by adding Variable Spread and Optimized Slippage to simulate using VPS with low Latency (Ping)
• Test 3 is backtesting similar to Test 2 but simulates using VPS with very high Ping

We can see thatusing good VPS can increase our profits by 170%. Don’t forget that what I tested were just default factory values. If we optimize parameters, there’s potential for even more profit and possibly reduced DD

For anyone looking for good VPS, I recommend MT4 Cloud – excellent service and we canchoose Server Location close to our Broker. This helps achieve Ping below 60 ms which isvery beneficial for Scalping EA orders as itreduces Slippage chances

Walk Forward Optimizations (WFO) is another test in the Robustness Test category. The purpose is to test consistency of Profit and Drawdown from the EA as well as reliability of Profit and Drawdown values

Additionally, WFO is considered anindirect Overfitting test for the EA. In this test we’ll set the EA to work with Fix lot to eliminate compounding effects and we’llconduct 10-year WFO with 8-year IS and 2-year Validation OOC (80%/20%)

The WFO results are quite satisfactory, as both %Profit and %Drawdown show Confidence values over 70%, which means we can generate profits with consistent risk across different time periods.

Monte Carlo (MC) simulation is one method for testing EA robustness. The purpose of MC simulation is to randomly simulate possible scenarios when we run the EA, where the software continuously changes OHLC data, resulting in different backtest results each time.

For this MC test, we tested 6 currency pairs using the Resampling method, allowing a 5% probability of trading errors (standard), running the EA for 15 years, and performing 1,000 simulations.

Test results show that at 95%CI the portfolio doesn’t blow up, with MAX%DD = 91.36% while Return/DD is 38.52, which is excellent value and totally worth running.

Although the Equity versus traded graph doesn’t form a straw broom shape, all possibilities are within good criteria. From visual analysis, we can infer that market price changes during EA trading periods significantly impact results. We’ll discuss these factors and ways to avoid poor outcomes.

Looking at predictions for continued EA performance, even in the worst case scenario, it can still generate high profits without portfolio liquidation or blowing the account.

Incubation testing means running the EA on a demo or live account to test for errors or bugs. We tested on a demo account for over a month and found bugs as expected.

After finding bugs, we fixed and improved the system, slightly developing the original model to add stability to the Trailing Stop system.

Wisdom EA Real Market Testing (Forward Test):

After running for a short time to check system and overall performance, we found:

1. The system is stable with no bugs from the EA itself.
2. Sometimes there’s order execution delay, but this is due to using VPS with excessively high ping (>100 ms)
3. Sometimes losses occur due to VPS crashes. You should check VPS frequently. If you don’t encounter this problem, I recommend MT4Cloud service – it’s truly excellent. I’ll include the link here: mt4.cloud/Register
4. EA functions work normally as during backtest.

Trading Performance in Forward Test:

When trading forex in real conditions, several factors can disrupt EA profitability:

1. Red box news during early morning hours (3-5 AM Thailand time) causes many orders but also many wrong direction orders, creating slippage. Impactful news includes:
   • Federal Open Market Committee (FOMC),
   • Consumer Price Index (CPI),
   • NON-FARM,
   • Interest rate news,
   • Jackson Hole meeting,
   • and AGI news, etc.
2. 1 Day relative performance (USD) is a key indicator for predicting order probability as it compares currency strength in stable countries. When this value exceeds 1, there’s higher chance of profitable orders, except when paired with red box news!! You should check this value at 23:30.
3. Currency pair selection is crucial since the EA is quite sensitive to spreads. Trade pairs with low spreads, noting that each broker has different spread values.
4. Proper lot size setting is more than half the battle. From experience, lot size = 20 might be overtrading because with $1,000 capital, the EA sets lot size = 0.32, which I see as profit with increased risk. For very safe pairs, you can set lot size = 30.
5. Zone selection affects losses. From experience, for riskier pairs I use only 1-2 zones: zone 2 and zone 3, since zone 1 has high order probability but also slippage risk. This depends on individual user management.
6. Stop trading during holidays and bank closures as it greatly affects EAs trading during early morning hours. If wrong direction orders open when banks are closed and trends develop, accumulated profits can disappear easily.

Let’s examine the EA’s strengths and weaknesses in table format for easier reading:

1. Importantly, past performance doesn’t “guarantee” future results. Testing software on demo accounts before live use is always good practice.
2. All investments carry risk, so thorough study before forex trading and investing is essential.