Hello, this is Tsumo.

I'll say this first.

There is no need to worry about tax return yet.

The approximately 32.2 billion yen we will introduce this time is thepast about six months of strategy test resultsby AITrendVision V1.03k.

We did not actually earn 32.2 billion yen.

However, even after subtracting that, the results during development were impressive.

In the previous GOLD verification, an initial margin of about 16,000 yen grew to about 800 million yen.

This time, we tightened the spread conditions compared to before and reviewed internal judgments in line with GOLD price movements.

As a result, about 16,000 yen increased to about 32.2 billion yen.

However, focusing only on profit does not mean the work is complete.

Currently, AITrendVision V1.03k is undergoing final adjustments toward public release.

If you’re interested in AITrendVision, please take a look here.

Two issues from the previous GOLD verification

In the GOLD verification with AITrendVision V1.02, about 16,000 yen grew to about 800 million yen.

However, upon rechecking the verification conditions, two issues were identified.

The first is the spread.

In the previous strategy test, GOLD spread was set to “2”.

Among many trading environments, even with a relatively low total trading cost, the spread is around 5.

In a typical environment, you should account for spreads up to about 20.

A spread of 2 was too favorable for GOLD verification conditions.

So this time, we first widened the spread to 5 for testing.

Applied EURUSD ruler to GOLD

Another issue is the internal EA decision method.

In the previous AITrendVision, the validation relied on a fixed pips standard based on EURUSD, regardless of price digit counts or instrument-specific price moves.

However, EURUSD and GOLD have very different price movement magnitudes.

A fixed value appropriate for EURUSD can be too strict for GOLD.

Conversely, if it’s too small for GOLD, it may be meaningless.

For example, a movement width deemed as a sufficient rise in EURUSD might be within normal price movement for GOLD.

The same applies to the spacing between additional positions.

If you apply a narrow fixed value designed for EURUSD directly to GOLD, positions can pile up in a short time.

Last time, we applied the EURUSD ruler directly to GOLD.

This time, we are revising the ruler itself to match GOLD price movements.

What changes in V1.03k

In the currently considered AITrendVision V1.03k, we are changing some judgments based on fixed pips to a relative standard that depends on market volatility.

When market moves are small and when they are large.

To avoid judging with the same fixed value.

Mainly reviewing the following:

• Change part of the fixed pips judgment to volatility-based criteria

• Adjust entry conditions to align with GOLD price movements

• Adjust spacing for additional positions according to market conditions

• Restrict BUY positions when prices rise too much

• Restrict SELL positions when prices fall too much

• Restrict initial entries that go against the direction of the higher timeframe

• Strengthen signal verification for additional positions

• Revisit dynamic spread filter

Simply loosening entry conditions is not the aim.

To normalize conditions that were too strict for GOLD.

At the same time, we suppress dangerous price ranges and frequent overlapping positions.

We are adjusting both entry and additional positions for GOLD.

Current strategy test conditions

The verification conditions are as follows.

• Version: AITrendVision V1.03k

• Instruments: GOLD

• Timeframe: M5

• Verification period: January 2, 2026 to July 10, 2026

• Verification period: about 6 months

• Test model: All ticks

• Modeling quality: 90.00%

• Discrepancy chart error: 0

• Initial margin: about 16,000 yen

• Initial lot: 0.01 lot

• Maximum steps: 13 levels

• Profit step-wise lot increase: ON

• Add 0.01 lot to initial lot for every roughly 16,000 yen profit

• Spread: 5

Note: Japanese yen conversion is calculated with 1 USD = 160 JPY.

From the previous spread of 2, this time it is widened to 5.

With the same logic, higher trading costs normally reduce profits.

Approximately 6 months of verification results

The strategy test results of V1.03k are as follows.