ACE Gold EA Ver.1 ― GENESIS

Now there are entries judgments and filters
There are also TP, SL, and trailing

But I didn’t set out to create such an EA from the start

What ACE initially tried to create

That was

To draw two parallel channels on the chart

That was all

The initial concept was quite simple

What to look at is the most recent 120 candles

First the upper side

From the 120, find the higher closes
List the top 3 candidates

Connect the tops of two of them to form a line

And shift a line with the same slope downward parallelly

Lower it until all 120 candles fit inside the channel

This is the channel formed from the top

Next, the bottom

Now look for the lower closes among the 120
List the bottom 3 candidates

Connect the bottoms of two of them

And move the line with the same slope upward in parallel

Raise it until all 120 candles fit inside

What it's doing is just a vertical flip of the above

In the mind, it was very simple

“Do the same as above, but below”

So at first

“If the upper side works, the lower side should be doable in the same way”

I thought

Change high values to low

Change up to down

Change the process from moving up to moving down

It’s almost a mirror image

However

For some reason, it wouldn’t work

Even though asked for the same thing, the same result wouldn’t appear

At that time I was coding with AI

For me

“Make the same processing as the upper side, just inverted, and make it for the lower side”

I meant to instruct

But when I saw the result

Something was different

The method of selecting candidates had changed

Before I knew it, another condition was added

Parts I had told not to change were changed

I wanted to reuse the same upper-side logic, but

When it became the lower side, some different interpretation came in

Even if corrected

Another part changed

“Therefore, it should be the same!”

No matter how many times I explained, it was slightly different

In my head

Same structure for top and bottom

But when it became code, it turned out differently

I was quite frustrated around this time

“If it works on the top, it should be the same on the bottom”

“Don’t do unnecessary things”

“Don’t change that part”

I remember repeating such exchanges many times

And finally

We returned to the point of rebuilding from scratch

Restart from the first 50 lines

Rather than continually fixing complicated code

I decided to start over from the beginning

No unnecessary processing

First, look at the 120

First, pick the highs and lows

First, recognize peaks and bottoms correctly

Then build up gradually

Stop trying to complete everything at once

Rebuild from the initial 50 lines while checking

Looking back now, I think this judgment was crucial

“Don’t change on your own” became the rule

From that period, the development approach with AI also changed

Use existing code as the baseline

Don’t make big changes

Clarify the places to modify

Check before and after changes

Don’t arbitrarily change the decided algorithm

This is now taken for granted

But in ACE’s early development

It was a time when we were building not only the EA logic but also the method of development with AI itself

I think so

Put all 120 inside

The channel had one more important condition

All 120 candles must fit inside the channel

Just drawing a visually plausible line isn’t enough

A channel that only fits some candles isn’t acceptable either

Create the slope from the two selected points

Move parallel lines while keeping that slope

Envelope all 120

From the top as well

From the bottom as well

Create with the same thinking

Finally, ACE began to have a foundation for viewing the market

Still, not trading yet

At this point, ACE

Still does not BUY

Still does not SELL

No profit yet

Just watching the chart

Watch the market from above

Watch the market from below

Do those two things simultaneously

But looking back now

This was the moment ACE first gained its “eye”

ACE now has various logics

But at its core, the original form is

To create two parallel channels up and down from 120 candles

That’s where it starts

And after much effort

Finally, the channel began to move

But the next problem was more difficult

Prices approaching the line

And

Touching

What would ACE do at that moment

Would it rebound

Would it break through

Would it buy

Would it sell

Or would it do nothing

Next time, Episode 15

“It touched. Now, what will ACE do?”

ACE, which was just making channels
is finally starting to consider buying and selling