The Era of Letting AI Trade Has Arrived

Hello everyone. I amMasayan.

I have been selling MT4 EAs and indicators on GogoJungle since2021.
However,a major turning point has come with 2025.

One of the big reasons for this isthe accelerated shift from MT4 to MT5.
This trend is especially pronounced overseas.

Furthermore, just recentlyOANDA Japan (OANDA Securities) announced the termination of MT4 services.
Already, the developer of MT4,MetaQuotes has stopped issuing new MT4 accounts to new FX brokers.
Also, existing FX companies are seeing revisions or halts on license provision,rapidly accelerating the move to MT5.

Migration from MT4 to MT5 and the challenges of automated EA trading

On the other hand,MT5 automated trading EAs are not yet as widespread in Japan as MT4

.There have long been major challenges with EA automation.
Namely,you need to contract and configure a Windows-based VPS server.

Many of you may already know this, but for beginners it is a considerable hurdle.
In short,

• MT4 is on a downward trend
• MT5 is still being adopted
• EA operations require VPS and environment setup

Thus, the era where traditional EA automation alone suffices has begun to end.

The reality of “AI trading” finally materializing

In this context, we have finallyChatGPTandGoogle Gemini, among others, enabling
next-generation AI trading to become a reality.

In overseas hedge funds and large institutional investors, there are already cases where AI is used in trading.
Of course, individual investors cannot reproduce their exact environment on the same terms.

So,what AI trading can individuals invest with?

To answer this, I havebuilt two different AI trading systems for MT4 and MT5.

First AI Trade

AI Signal Trader: a system where AI analyzes news to execute trades

The first one isAI Signal Trader, an AI trade system.

It utilizes RSS feeds from news sitesRSS feeds,
and sends the latest newsto ChatGPT and Google Gemini,
where the AI analyzes the content anddecides whether to go long or short.

What is an RSS feed?

An RSS feed is a publicly available URLwhere you can fetch news and updates.
Examples include:

• RSS from economic indicator sites
• Truth Social RSS related to President Trump
• RSS from overseas economic news sites
• RSS updates from financial news

Currently,up to 10 items can be registered.
This is to reduce server load, simplify prompts for the AI, andreduce token usage.

How does it flow?

Fetch the latest articles from registered RSS feeds and pass them to the AI.
Then the AI will

• analyze statements from Trump
• analyze economic indicator results
• analyze financial news

and determine whether to go long or short against USD/JPY.

In other words, unlike traditional MT4 EA that relies on historical rates and technicals,
this is a trade system where AI makes buy/sell decisions based on recent news.

Features of News Analysis AI Trading

Abroad, in fact, there are hedge funds already running such systems in
ultra-fast, ultra-low latency environments.

However, individual investors do not need to engage in nanosecond-level battles.
What individuals should aim for is trading on a slightly longer time frame.

For example,

• Day trading holding from minutes to tens of minutes
• Day trading holding for a few hours
• Swing trades that span days

With these timeframes, AI news-analysis trading is quite realistic.

In practice, the EA I developed tends to hold positions for about
an average of 1 hour to several hours.

Also, this is publicly disclosed in real trades, so please check those if you're interested.

AI Signal Trader Real Account:https://real-trade.tech/accounts/82850?via=users_realtrade

AI Signal Trader Demo Account:https://real-trade.tech/accounts/82890?via=users_realtrade

Even if the results temporarily dip negative, reasons why it cannot be dismissed immediately

What matters here is
“Don’t rush to conclude that this AI trade will not win just because it’s negative now.”.

AI models are evolving daily.

For example,

• ChatGPT
• Google Gemini

as these models improve,
it becomes more likely that AI can interpret not just superficial meanings of news butdeeper contexts and their impact on financial markets.

In short, the characteristics of AI trading include
the potential for accuracy improvement as AI itself evolves.

I think this is one of the biggest features not present in conventional EA automated trading.

Second AI Trade

AI breakout line EA

AI trading that analyzes candlestick data by sending it to AI

The other is the method of the trader I respect,Jun FX, and the “AI breakout line EA” I developed inspired by it.

This usescandlestick data, which MT4/MT5 excels at.
Specifically,obtain the most recent 100 candlesticks, send them to ChatGPT or Google Gemini.

When sending information to AI, include items such as the following:

• High, low, open, close, and time for the most recent 100 candles
• Resistance line conditions
• Support line conditions
• Conditions for lines likely to accumulate orders
• Consider key round numbers
• Rates that many participants are likely watching (stop-loss lines)

Then AI analyzes these andplaces orders at stop-loss price in advance to follow the price movein the intended direction.

A strong point of this trading is that orders are placed in advance using stop orders.

※The stop/limit switching can be changed in properties. (Stop orders are for trend-following entries, limit orders for counter-trend entries)

The weakness of AI trading lies in lag.

This AI systemdoes not make trading decisions every tick, but rather reads the chart shape in advance and places orders at predicted prices as a waiting game.

Why conventional EAs had a hard time winning

Even with conventional EA automation,
EA that uses past rates to determine resistance/support and places limit and stop orders could be implemented.

In my experience,that alone is not enough.

Because truly successful traders look not only at simple rules but also
at the overall market flow and price ranges that market participants watch.

I have also tried to convert such logic into EA many times.
But the results often showedthe asset chart trending downward.
In other words, simply placing orders mechanically does not guarantee success.

What happens if AI captures a trader’s “concept”?

What’s interesting is
what if a successful trader like Jun FX’strading strategycould be embedded as prompts to AI?

For example,

• Which lines tend to accumulate orders
• Which price ranges attract attention
• How to switch between trend-following and counter-trend
• Where breakouts tend to occur
• Techniques that exploit human psychology

If you could feed AI theseessences of human decision-making,
AI could generate buy/sell signals based on those concepts, creating a very interesting system.

AI has already shown performance that surpasses humans in certain domains.
In trading, the era will move beyond simple rule-based approaches to
reflecting superior human methods in AI,

The full picture of this system

The system I developed this time is
not a scalp that follows instantaneous price panel movements.

Of course, in discretionary trading, rate panel-based scalping is an important element.
But fully reproducing that with EA is difficult (EA struggles to track price action).

So, the first EA holds positions for a few hours (a day trade), and the second EAplaces pre-set stop orders to ride the market direction.

Settings example

• After opening a position,close in 10 minutes
• Profit and loss limits can be set. Example:take profit at 30 pips, stop loss at 30 pips
• One-position operation is the default

EA that allows averaging down may cause bugs or infinite averaging risks.
Therefore, the design uses multiple positions (maximum 3) oroperate safely with a single position.

Of course, customization is possible and configurable via EA properties.

The mechanism needed to run AI trading

Not limited to MT4/MT5 alone

What matters here is that these AI tradescannot be completed with MT4/MT5 alone.

To communicate with LLMs like ChatGPT and Google Gemini, we use
PHPa program that can run on common rental servers.

The exchange usesJSON format.
JSON formatis, put simply,a data format readable by both humans and machines.

Concrete flow

1. Prepare data required on the MT4/MT5 side
2. Send it to the server via WebRequest
3. PHP makes a request to the AI
4. Save the signal returned by the AI on the server
5. MT4/MT5 reads that signal and makes trading decisions

Thus, the system operates as
EA + PHP + AI API.

What is required

To implement this setup, in addition to conventional EA automation, you need the following:

• MT4 or MT5
• Rental server
• PHP scripts
• API keys for OpenAI or Google
• WebRequest configuration

In other words, even if you do not write code, you need a technical stack capable of configuring servers and APIs.

How much do API usages cost?

To use ChatGPT or Google Gemini,
API keys must be issued by OpenAI or Google.

And API usage is billed based ontoken consumption.

If you send a huge number of prompts without care, it could cost several tens of dollars per day.

the average is about $0.2 per day

.

In Japanese Yen,
about 30 yen per day.

under 1,000 yen

is expected.

• how frequently you send requests
• how much data you pass to the AI
• which AI model you use

and will vary accordingly.
Nevertheless, realisticallyoperating with around 100 yen per day is practically feasible.

Source code will be published in the next article

Demo version will be published for free

To offer such a system to everyone,
the logic and source code will be published in the next article.

First,the free portionwill include a demo version,
the source code up to the communication with AI and actual MT4/MT5 entry will be fully published.

• The full EA source code
• All PHP scripts
• Basic communication setup

In other words, during the demo stage you will be able to
see AI communication and MT4/MT5 actually entering
.

About the server

A rental server on which PHP runs should work as a baseline.
Free servers may work, but typicallya monthly rental approximately a few hundred yenis realistic.

If you already have a rental server, you can use that.
If you don’t have one, you might start witha rental server with a trial period.

About API settings

AI models include

• Google Gemini
• OpenAI (ChatGPT)
• Anthropic family
• Grok family

and others.

By default, it is set to OpenAI (ChatGPT,

Google Geminiwill also be selectable.

Therefore, as initial setup, currently

• Google
• OpenAI

are the two companies used.

Please issue API keys for each service and enter them in the settings.
Basically, you will run withone of these two.

Conditions to run this EA

In the demo version, for clarity of the communication test, AI is configured to always instruct long positions.
Therefore,

• Set the API key
• Install PHP on the server (note the URL)
• Launch the EA on MT4/MT5
• Configure WebRequest(paste the noted URL)

If communication succeeds up to here,
MT4/MT5 should take long positions.

If a trade occurs in the demo, you only need to replace with the paid source code.

About the paid portion

The production source code will be published in the paid portion of the article

Andthe paid portionwill
publish all production-ready source code.

Process outline

1. Test operation with the free demo
2. Confirm successful communication (EA holds long positions)
3. Replace to production code for both EA and PHP
4. Operate with actual AI trading logic (prompts can be freely altered)

This is the envisioned flow.

These are suitable for people like this

This system requires no programming.
However, you will need to

• Rent a server
• Set up PHP
• Issue API keys
• Configure MT4/MT5

Thus, rather than being for absolute beginners, it is suited for those who

• are reasonably comfortable with PC tasks
• have used MT4/MT5 before
• are interested in AI trading
• want AI to throw trading rules and have it decide buys/sells
• want to understand the system with source code

This is the kind of content.

Next, a paid article will be sold

Price is planned at 49,800 yen

In the next article, a paid article compiling this content will be sold.Paid articlewill be released.

Price is49,800 yen (tax included).

EA name：AI Signal Trader

Description: An AI trade where AI reads news sites to make buy/sell judgments.

EA name：AI breakout line EA

Description: An AI trade that reads the last 100 candlestick values (changeable) and places orders using limit/stop orders.

■Contents of the paid article

・MT4/MT5 source code

・PHP scripts

・AI connection mechanism

・Implementation steps

・Transition from demo to production

These contents are planned to be published together.

The era of letting AI trade has already begun.

If you’re interested, please do follow and stay tuned.