[Two Companies Side-by-Side Real-World Verification] AnomalyEA Operational Report Week 2 (8/25–9/1)

Hello. I’m Leo FX, developing the USDJPY-only EA “AnomalyEA.”

In this series, I run the same EA on two brokers, OANDA Securities and Gaitame Finest, at the same time and publish the results whether they are wins or losses.

The second installment covers four events from August 25 to September 1.To be brief, the result is negative.

■ Summary

・4 events: 1 win, 3 losses

・OANDA Securities −20.8 pips (−20,800 yen) / Gaitame Finest −18.4 pips (−18,400 yen)

・Cumulative (from 7/24) isOANDA +46.9 pips / Gaitame Finest +52.3 pips and staying positive

・Only on the OANDA side, 37 orders were postponed on August 25(Gaitame Finest had zero)

・The late-month London FIX leg that previously incurred a loss isavenged with +1.3 pips

■ Operating conditions (the same each time)

・EA: AnomalyEA (time-based, USDJPY-only,no averaging down・SL 25 pips / time-based settlement)

・Accounts: OANDA Securities Demo / Gaitame Finest Demo running in parallel

・Lot size:1.0 lot fixed (uniform across both)・Other defaults

■ All trades (two-broker comparison, unit: pips)

By day, it looks like this.

■ What happened

On August 25 and 28, the two consecutive short legs moved counter to the expected direction.

This EA tries to capture the real-market flow of “rising toward the value date and then dropping after passing it.” However, on these two days the trend continued toward yen depreciation even after passing the value date, so the shorts did not expand. On 8/28, both longs and shorts were negative.

On August 31, the opposite happened.

08:30  160.17  ← Long position opened
09:53  159.99  ← −18.6 pips (unwound)
09:55  159.99  ← Short position opened
11:45  159.79  ← +19.5 pips (recovered)

It was a day of strong end-of-month yen-buying, with the price steadily declining from the morning. The long was a total loss, butthe shorts on the same day recovered to end the day in positive. The two legs facing in opposite directions work well on days like this.

■ The comeback battle of the leg that previously hit a loss

In the last report, the third leg limited to month-end, “L3 (London Fix),” made its debut and hit the SL, suffering −26 pips. It was a painful loss as the price was cut by the minute-long London Fix wick, and then immediately rebounded.

This time, the second activation was +1.3 pips.

Not a flashy win. It held until time-based settlement and finished quietly with a profit. However,it was confirmed that it functioned as designed without hitting the SL. Since this leg has only operated twice so far, it’s too early to judge. Next is at the end of September.

■ Real spread measurement corner (this series’ flagship)

Maximum spreads during the value-date window (8:25–9:55). This time it becomes the9th event.

On the OANDA side, in 8 of 9 cases the spread widened to around 2.0 pips, while on Gaitame Finest the spread never exceeded 1.5 pips (except for the market turmoil at the end of the month on 7/31).The same pattern repeats, nine times. It’s hard to consider this a coincidence any longer.

■ And this time, the real harm became clear

AnomalyEA includes a safety feature: it does not place orders if the spread exceeds 1.5 pips.

On August 25, this safety trigger activated 37 times on the OANDA side.

Postponed short: spread too wide (watch for value-date widening)   ← this is 37 times

At exactly 9:55, the spread widened to 2.0 pips, so the EA continued to reject orders,and only executed 5 seconds later. At the same moment, Gaitame Finest had a maximum of 0.80 pips,no postponements and a single fill.

In the first installment, I reported 10 postponements on July 24, but this time it’s 37.Depending on the day, the EA might not be able to place orders at the most favorable moment of the value date.

For this particular case, the impact on profit and loss was about 0.4 pips (the difference between OANDA −20.8 and Gaitame −18.4 is largely due to execution timing on days other than 8/25).“37 rejections do not mean enormous loss.”That is stated accurately.

■ Plans going forward

・September’s Go-To days (GOTO days) for execution:9/4 (Fri)・9/10 (Thu)・9/15 (Tue)・9/18 (Fri)・9/25 (Fri)・9/30 (Wed)

・September 30 is the last business day of the month, so the L3 leg will activate for the third time

■ Learnings from this round (Honest Corner)

1) Negative periods come normally.

In backtests (9.4 years),the worst months are four out of ten months in a year. The worst month can be −87,899 yen per 1.0 lot, and there were months of consecutive losses. The current −20,800 yen is just a normal part of that distribution.

2) The difference between the two brokers is smaller than expected.

From the spread table, Gaitame Finest looks dominant, butthe actual profit-and-loss difference is 5.4 pips over 9 events ( +52.3 vs +46.9 ), about 0.6 pips per event. It cannot be ignored, butthis doesn’t mean “you can’t win with OANDA.”

If one wants to inflate numbers, one could say “37 rejections!”, but that would give readers a false impression.The number of rejections and the actual pips lost are two different things.

3) This series exists for this purpose.

In the first installment I wrote that results would be published whether win or lose. Having a negative report right after feels a bit uncool, butthere is no value in a series that skips losing periods.Next time’s report will be published as is.

■ Disclaimers

This article is a verification record of an automated trading program and not investment advice. The published results are past performance and do not guarantee future profits. FX trading may incur losses exceeding the principal.

・USDJPY-only EA “AnomalyEA” (GogoJungle) → https://www.gogojungle.co.jp/systemtrade/fx/82678

・Operational account records (Real Trade) → https://real-trade.tech/accounts/83716

・Free tools (Go-To calendar / account suitability checker, etc.) → «GoGoJungle vendor page URL»

・Article of the first installment → https://www.gogojungle.co.jp/finance/navi/articles/123705