[GOLD Feature vol.3] Getting AI to stop you with a “Do not enter” warning — A new frontier in gold scalping with a learning AI that remembers losses

“More than deciding when to enter, the ability to stand aside determines your trading results.”

Even so, consistently making accurate “stand aside” decisions is likely difficult for many traders.
A signal appears. Conditions are met. But something feels off—now a tool has arrived thathas AI learn from past data and express that “something feels off” in language by design.

In [GOLD Feature vol.1], we introduced “no-discretion scalping,” which completely removes discretion ((▶︎Click here for vol.1)
In [GOLD Feature vol.2], we introduced “second-speed scalping,” where you get in and out within a few to a few dozen seconds. (▶︎Click here for vol.2)

This time, in [GOLD Feature vol.3], the approach is a bit different.Instead of an entry “trigger,”we focused relentlessly ona filter that“stops” entries—its performance—and that is the design behind“AI Scalper GOLD.” developed by trader Leo.

While many AI tools aim to improve entry precision, Leo deliberately chose the opposite idea.
“Reducing losses directly increases wins.” Based on this belief, he trained AI on past losing patterns and equipped it with a function to automatically filter out low-expectancy situations.

In this article, we thoroughly explain everything from the core design philosophy to the developer’s own grounds for reliability.

1. Who is the developer, Leo?

Leo has a career ofover 10 years as an FX trader, and he also works as an indicator and EA developer. His total social media following exceeds10,000, and he has achievements includingmultiple placements in trading contests and EA competitions.

He first tried FX in 2010 but faced initial setbacks. A retry in 2016 also didn’t pan out and he withdrew. Although he made significant profits during the 2017 crypto bubble, he looks back on it as “luck.” Afterwards, he apprenticed under a professional trader he knew, then focused seriously on FX during the COVID era. Encountering EAs on GogoJungle became a turning point, leading him to pursue indicator and EA development.

Leo’s words about this journey are memorable.

I couldn’t win with discretion, but I could with EAs—
that excitement led me down the path of development.

As he dug deeper into logic, the “winning scenarios and losing scenarios” became clear. Leo says that process also led to improving his discretionary skills. After trial and error, he arrived at the approach of “conquering the market with mechanisms,” and now proudly releases “AI Scalper GOLD.”

2. Why scalping GOLD now — background of development

“Too few signals for day trading tools”—this kind of user feedback was the starting point. Switching to scalping was a natural choice to increase the number of entries. However, at the same time, it creates a new challenge thatmore entries also mean more losses.

Leo confronted this contradiction head-on, going throughover 30 complete rebuilds, and the answer he reached was the “losing-pattern learning filter.”

“GOLD’s volatility creates 100–200 pips of opportunity in a short time. I wanted a design that could run sufficiently even in 1–2 hours after getting home. My top goal was to create a tool that part-time traders can use comfortably.”

The reasons for choosing GOLD are also clear.It has high volatility and is well-suited to scalping, and moreover, even with just 1–2 hours in the evening, there tend to be opportunities to target 100–200 pips a day. It’s a currency pair that naturally fits the lifestyle rhythm of part-time traders.

↑ +100 pips with a 1-minute scalping move

Before release, he asked members of his own community to serve as testers tothoroughly verify reproducibility. He pursues a design that’s not “a method only I can win with,” but “a method that brings anyone closer to the same results.”

3. What is the “losing-pattern learning filter”?

Most AI tools focus on improving entry methods. But the design philosophy of “AI Scalper GOLD” faces the other way.Keep entries themselves simple. Leave the filtering to AI—that’s the core.

▌ Why make entries simple?

“The more a scalper earns in the hundreds of millions of yen, the more they prefer simple methods. The real gap between beginners and experts isn’t entry technique—it’s judgment on when to stand aside.”

Patterns where a high breaks but turns out to be a fakeout, points that look like ranges but are actually trend reversals—AI learns the characteristics of these “low-expectancy situations” from past data. By not showing signals, iteliminates poor-quality entries at the root.

▌ Designed to work in both ranges and trends

The filter is designed to work in both range and trend markets. By not issuing signals in “ambiguous” situations, it suppresses the very swings in judgment. The learning accuracy of losing patterns leads directly not to win rate but toimprovements in expectancy.

“Reducing losses directly increases wins”—this inverted idea is precisely what makes “AI Scalper GOLD” stand out not as an entry tool but as afilter tool.

4. A logic that works even as an EA — grounds for reliability

In general, because human judgment intervenes in sign tools, it’s hard to see the tool’s true capability. In the case of “AI Scalper GOLD,” its logic isalso under test as an EA, and has recorded excellent profit performance in backtests.

When converted to an EA and run for one year, it grew sevenfold.

“We deliberately provide as a sign tool a logic that can still win when automated. If we made it an EA, human judgment couldn’t intervene in response to changing market conditions. First, I want people to use it as a sign tool and confirm with their own eyes—that’s why we chose the current form.”

As for Leo’s most recent results, he publicly sharedturning margin from¥300,000 → ¥1,300,000 in about a month and a half (including other trades). What he cites as the biggest gain isn’t asset growth but that hestopped hesitating over entry decisions. The sense of security provided by the filter elevates the quality of trading itself.
*EA version is also planned for sale.

Right after sales began, we received reviews like this:

“Just following the signals is simple, yet I started winning.”

★★★★★

I’m amazed at how much I can win with this tool—it’s shocking.

I tried it skeptically with a small amount, but in just three days since I started, I’ve made over ¥100,000 in profit.

It’s simple to just follow the arrows, so even as a beginner I could trade easily. I’ll keep using it!

5. Recommended for these types of traders

Leo says it’s particularly suited to the following traders:

01Part-time traders who work during the day

With high-volatility GOLD scalping, there are plenty of opportunities even within 1–2 hours after coming home. You don’t need to stick to the charts for long hours; it’s designed to let you trade efficiently within limited time.

02Those who want to eliminate hesitation in entry decisions

Because the AI filter screens out low-expectancy situations, it reduces hesitation over “enter or stand aside.” Even during busy evening hours, entry decisions are completed by simply following the signals.

03Those who want stable results with scalping

Instead of increasing the number of entries, it narrows them to high-quality entries. This offers one approach to the scalping-specific issue where “more trades also mean more losses.”

“AI Scalper GOLD” turns GOLD’s volatility into a weapon and was born from the inverted idea of “reducing losses.” The crystallization of a filter design reached after over 30 rebuilds is aimed at everyone who, despite the constraints of being a part-time trader, wants to trade without hesitation.