Test Investment Hypotheses with Data and AI — Introducing FactDecode

---

When building an investment strategy, the list of factors worth testing can grow quickly.

Price trends.
Volume.
Volatility.
Relative strength.

Then there are macroeconomic variables that shape the broader market environment: CPI, employment data, policy rates, long-term interest rates, and other economic indicators.

Each one can lead to a reasonable hypothesis.

But once you have several candidates, a harder question appears:

Which factors are actually contributing to the result?

And more importantly:

What happens when you combine them?

A factor that looks powerful on its own may simply be capturing information already contained in another variable.

Meanwhile, a factor that appears weak in isolation may become meaningful when combined with other market conditions.

This is the problem FactDecode was built to explore.

FactDecode is an analysis platform for testing factor contribution, factor combinations, and investment hypotheses using market data and AI.

FactDecode is not designed to tell you what will go up next.

Instead, you choose the data and factors you want to research, and FactDecode helps you analyze how those factors relate to future market behavior using machine-learning models such as XGBoost.

You can work with market-based factors while also incorporating macroeconomic data such as CPI, employment conditions, and interest rates.

Because these datasets are often published at different frequencies, FactDecode aligns them to the target analysis timeline so they can be examined within the same research workflow.

The core process is:

Data → Factors → Projects → Run → Results

Prepare the data, define the factors you want to test, save the research conditions as a Project, run the analysis, and review the evidence in Results.

Rather than stopping at a model score, FactDecode lets you examine questions such as:

Which factors contributed most to the result?

How did different factor values affect the model?

Does the relationship still appear in validation data?

Factor Contribution, correlation analysis, SHAP, quantile analysis, and validation views help make the model more inspectable instead of treating its output as a black box.

Markets do not remain static.

Interest-rate conditions change. Inflation changes. Volatility changes. Relationships between factors can change as well.

A combination that appeared meaningful during one period may not behave the same way later.

FactDecode therefore treats analysis as something that can be repeated rather than as a one-time result.

Datasets, research conditions, Runs, and Results are managed so that you can preserve the context of an analysis and return to it later.

As new data becomes available, you can run the research again and examine whether:

• the most important factors have changed,
• factor contribution has shifted,
• validation results have weakened or strengthened,
• or the model is describing a different market environment.

The goal is not simply to discover a relationship once.

It is to build a repeatable research process for testing whether that relationship continues to deserve attention.

FactDecode is available in three plans: Standard, Pro, and Elite.

Standard — $79/month

For building hypotheses and identifying which factors appear to matter.

Standard includes the core FactDecode workflow, from data and factor preparation through analysis runs and core result review.

Pro — $249/month

For going deeper into why a factor matters and how it affects the model.

Pro adds deeper analysis through tools such as SHAP and Factor Analysis, helping you inspect factor behavior and the structure behind a result.

Elite — $799/month

For researchers who want to track how analysis results change over time.

Elite adds deeper validation and run-to-run comparison capabilities, allowing you to monitor whether previously observed relationships continue to hold as new data arrives.

You can review the full feature set, screenshots, and plan differences on the FactDecode product page.

The core analysis platform technology behind FactDecode is currently patent pending.

We will continue investing in both technical development and intellectual-property protection as FactDecode evolves as a long-term research platform.

To mark the launch of FactDecode, we are offering a 50% discount on the first month for the first 20 Founding Users.

（Code: FOUNDEROZAWA50 ）

This is not intended to become a permanent 50% discount.

FactDecode will continue to evolve after launch, and feedback from early users will play an important role in that process.

We want to hear where the workflow feels unclear, what kinds of research users want to perform, and which datasets or analysis capabilities would make FactDecode more useful.

Markets are full of claims about what works.

A particular indicator matters.

A certain market condition is important.

A specific factor predicts what comes next.

The difficult part is not finding another idea.

The difficult part is testing it.

How much does each factor actually contribute?

What changes when multiple factors are combined?

Does the relationship survive outside the data used to build the model?

FactDecode does not provide trading answers.

It provides a research environment for testing your own investment hypotheses with data, machine learning, and validation evidence.

For screenshots, detailed features, and plan information, visit the FactDecode product page.

[Explore FactDecode]

FactDecode is a research and analytics workspace. It does not recommend specific securities, tell users to buy, sell, or hold, manage assets, or provide personalized financial advice. Historical and validation results do not guarantee future performance. Users are responsible for their own investment decisions.