Testing EdgeDriven Dollar Yen Portfolio: Setup and Result Checks

EdgeDriven Dollar Yen Portfolio combines nine fixed USDJPY strategies using H2 data. Its entries use pending stop orders, followed by protective stops, break-even management and exits based on holding time in bars.

That matters when reviewing a demo: an order can be submitted without becoming a trade, and a long list of orders is not the same as a long list of completed positions. This guide explains how to set up the test and read what actually happened.

1. Launch the correct product and symbol

Find EdgeDriven Dollar Yen Portfolio in the Market section of desktop MT5 and download the free demo. Select the installed EA in Navigator and choose Test. MQL5's demo-testing guide covers this workflow.

The free Market demo is for Strategy Tester, not an online chart on a demo or live account. The deposit used below is simulated; it is not a request to fund an account.

In the tester, select your broker's actual USDJPY instrument, including its suffix if it has one, and set the chart period to H2. Use hedging position accounting and make sure H2 history is available. The tester's advanced account settings expose the risk-management model under Margin. Do not choose Gold or another currency pair simply because the terminal accepts the selection.

2. Separate reference settings from your own account test

The published MQL5 historical reference uses:

Select Every tick based on real ticks for the reference-mode comparison. A run on your broker's USDJPY data is a broker-specific check, even when you copy the other settings. Different history and contracts can produce different trades. Retain the actual available test dates and the product version instead of describing every such run as a reproduction of the reference.

At the reference risk settings, the nominal budget for each module's entry is 0.1% of stored starting balance, before lot rounding. That is not a 0.1% maximum loss for the account or a limit on portfolio drawdown. With compounding enabled, new sizing uses current balance instead.

Choose a single test and disable optimization. In the tester Inputs, confirm TradingEnabled=true and EmergencyStop=false. Keep the news-control settings at their documented defaults for the first run and save any guard-related messages. Changing those controls creates a different test, not a repair to a disappointing result.

3. Inspect orders, deals and the Journal

After starting the run, check the Journal for initialization, history and trade-request errors. Note insufficient margin, invalid stops or volumes, and any calendar or news-guard messages rather than dismissing them because a report was produced.

In the results, distinguish pending orders from executed deals. A stop entry may expire or be cancelled without a fill. An opening deal and a closing deal are not two separate completed trades. Compare the same report field on both sides.

If trade count is much lower than expected, inspect the actual date coverage and skipped-order messages first. The EA rounds volume down and skips entries that cannot fit within the budget at the broker's minimum lot. Increasing risk just to make the count larger defeats that check.

4. Compare the full report, not a single total

Review trade count, profit factor, net profit, swap, commission and the equity curve. Keep the exact drawdown field name with its percentage and money amount. Compare equal time windows and the same compounding setting.

If the same trades have similar prices but different net results, examine costs. If entry times or directions differ, examine history, H2 bar timestamps, version and inputs. These are starting points for diagnosis, not a claim that every mismatch has the same cause.

Use the report's actual dates. A broker with a shorter history has not tested the missing years, even if the requested start date was earlier. Also avoid adding separately initialized phase tests together and treating the total as an independently verified continuous run.

5. Keep the public account record separate

Dollar Yen has a public MQL5 Signal, but that account and the historical reference use different account and risk configurations. Do not copy its settings into a reference test without making that difference explicit.

Read the Signal's current history length, account conditions and disclosed risk settings independently. A historical tester run cannot verify live calendar availability, connectivity or future execution, and an early forward record cannot establish long-term stability. Buying the EA and subscribing to a Signal are separate decisions.

6. Keep enough detail to ask a useful question

Save the report, inputs and relevant Journal excerpt. Include the EA version, MT5 build, broker/server, exact symbol, dates, model, deposit, currency and leverage when asking for technical help through MQL5 Comments or Messages. Remove account credentials and unnecessary identifiers.

Once the first run is understood, make a separate test using the account conditions you intend to evaluate. An unexplained failure is a reason to ask before buying, not a reason to hide the error or increase the risk setting.

Backtests and public account monitoring provide different information. Neither guarantees future profit, identical broker results or a maximum loss.