The Golden Guardian, why there are few deals for the official forward — We disclose trades from our verification account
Golden Guardian — I will answer the questions we have received
Hello, this is Yoko.
Recently, regarding the Golden Guardian, we have received several inquiries like, “Although official forward trading is infrequent, is it actually functioning well?” We seem to have caused some concern, so I will answer here without waiting for the next installment of the Guardian series.
First, the conclusion
・Guardian is originally an EA with few trades (1,596 trades in 12 years, about 10 per month)
・The official forward had a period when trading stopped due to an immediate post-release issue (fixed in v1.04 on September 7)
・Another factor may be environmental differences in the official forward, which could also explain the low trading activity (undergoing verification)
・In our demo account, from September 3 to October 2, about one month, it traded at a pace close to backtesting, 14 times
Guardian is originally a “wait EA”
Guardian is a no-averaging, one-position operation EA that only moves when the price exceeds the previous day’s high or low. In 12 years of backtesting, it has traded 1,596 times,about 10 times per month, about 2–3 times per weekpace.
This is an average statement, so there are weeks when it hardly moves depending on market conditions. The low trading frequency is not a malfunction; it is a characteristic of this EA.
What happened in the official forward
① Post-release issue
Immediately after release, in some account environments, capital management calculations were not performed correctly, preventing new orders from being placed. In the official forward as well, trading did not occur for a while due to this effect.
It was fixed on September 7 with v1.04, after which trading resumed. We apologize for the concern.
② Environmental differences (under review)
There is another possible factor we are considering.
Guardian has a safety mechanism that omits new entries when spreads are wide. The daily timeframe switch that Guardian targets tends to widen spreads around those times across any account.
The environment of the official forward seems to show wider gold spreads than the accounts we use for operation and validation, so we think entries may be omitted for that reason. We are continuing to verify, and will inform you here as soon as we know more.
How our demo account moves
To show real movement, here is the trading of Guardian as we verified it inthe demo accountfrom September 3 to October 2, all 14 tradesas-isWe are running it alongside other EAs in a demo account with ample funds, so we changed lot sizes in the middle; thus, the table’s profits and losses are allconverted to 0.1 lot amountsThe broker name of the account is undisclosed.
| Date | Direction | Profit/Loss (converted to 0.1 lots) |
|---|---|---|
| September 3 | Sell | +¥250 |
| September 4 | Sell | -¥94 |
| September 8 | Sell | +¥247 |
| September 10 | Sell | +¥155 |
| September 11 | Sell | -¥12,868 |
| September 14 | Sell | +¥4,235 |
| September 16 | Sell | +¥203 |
| September 17 | Buy | -¥2,769 |
| September 21 | Sell | +¥5,998 |
| September 22 | Sell | +¥707 |
| September 23 | Sell | +¥2,575 |
| September 25 | Buy | -¥2,494 |
| October 2 | Buy | +¥24,718 |
| October 2 | Sell | -¥2,697 |
| Total (14 trades, 9 wins, 5 losses) | +¥18,166 | |
This table shows all Guardian trades from September 3 to October 2. In actual lots, September 3 was 0.01, September 4–11 were 0.5, and from September 14 onward, 0.1. The amounts in the table are the actual profits/losses adjusted to 0.1 lots.
About 14 trades in roughly one month.At a pace of about three trades per week, it moved at a frequency close to backtesting (about 10 trades per month).
Wins: 9, Losses: 5.Win rate is about 64%, lower than the 12-year backtest (77.1%).
Wins can extend through trailing.The buy on October 2 was +¥24,718 (0.1 lot).
Just sharing only the good points is not our style, so here are aspects of concern as well.
The biggest loss is the sell on September 11.It reached stop loss, which was -¥12,868 in 0.1 lot equivalent (actually -¥64,338 at 0.5 lots). We do not average down, but when stop loss is reached, that amount of loss will occur.
The total for this period is +¥18,166 in 0.1 lot terms(If you sum with actual lots, it would be -¥32,300). The two events on September 11 loss and October 2 gain heavily influenced the total. Excluding these two, it’s about +¥6,300.
Fourteen trades are still few, and I think it’s premature to draw conclusions from this alone.
Also, since this is a demo account, there are differences from a real account in order execution and spreads. The official forward uses different accounts and lots, so you cannot compare the amounts directly. Please view this as only an example of movement.
Backtests were performed with what spread level?
We also receive questions like, “With what spread were the product page backtests conducted?”
The twelve-year backtest wasconducted using MT5 Strategy Tester with real tick data (history quality 99%)and spreads were not fixed;they were account variable spreads based on actual price movements at the time. The numbers on the product page were tested with a “low-spread account” where spreads are narrow.
On accounts with wider spreads, performance decreases. Therefore, under the same conditions, we also tested with a“Standard account”(an account with wider spreads than the low-spread account).
| Low-spread account (as shown on product page) |
Standard account | |
|---|---|---|
| Profit factor | 3.36 | 2.51 |
| Maximum drawdown (as a percentage of balance) |
3–4% | 5% |
| Win rate | 77.1% | 69.6% |
| 12-year net profit (with 1,000,000 yen margin) |
About 7,980,000 yen | About 6,690,000 yen |
July 2014 – July 2026, 0.50 lot fixed, based on past verification results, not a guarantee of future performance.
When spreads widen, the numbers drop. However, even with Standard account, the profit factor remains around 2.5 and maximum drawdown around 5%, which are practical levels.
In other words, in environments with wider spreads than the Standard account, performance may drop further, and entries may be omitted more often. Please check the gold spread for your account in advance.
For those considering
・Low trading frequency is due to the EA’s nature, not a malfunction.
・Accounts with narrow spreads tend to show better performance and more entry opportunities (see the table above). Please check the gold spread for your account in advance.
・Do not judge based solely on the official forward trades. It is most reliable to test movements on your own account starting with smaller lots.
Future plans
We will share both the good and bad points about the official forward as we learn them here.
Also, from October 5, we started running Golden Guardian solo on a real account (Standard account). When trading accumulates, we will also announce that here.
We also plan to deliver a Guardian episode in the latter part of the serialized “Yoko and Kenji's GOLD EA Mastery Log.” There, we will discuss Guardian’s mechanism in more detail.
Thank you for reading to the end.
Yoko
Golden Guardian
View the Golden Guardian sales page ▶※This article presents an example of trades from our demo account and does not guarantee future profits. Please make investment decisions at your own risk.