Research and Publishing Policy, and Important Notices
This page explains AlpAI Lab's research aims, how decision records, test results and trading performance differ, how free registration and possible future paid access are positioned, and how we handle information and corrections.
Under what conditions is AI-assisted investment decision-making useful, and where does it fall short?
AlpAI Lab is a research publication that uses the management of our own company's funds as its subject to publish AI-driven company analysis and tests of investment methods.
We distinguish clearly among live trading, observation of analyses and scores, simulated trading, and backtests on historical data. It is operated by GRID Start, Ltd. Alongside the results, we explain what we compared and what the findings show.
Our goal is to help readers evaluate AI analyses and trading performance with the conditions and limitations in mind. This page explains how we keep these kinds of information apart, how to read what we publish, and how we handle registration, conflicts of interest, and corrections.
What We Want to Find Out in Our Research
Good performance and the effectiveness of the rule that was tested are not the same thing. How results look can change depending on the market, the period covered, and the comparison conditions.
For that reason, we examine what changed when a trading rule was altered, along with the comparison conditions. We also cover tests in which the expected effect could not be confirmed, and we explain why we did not adopt the rule and what should be checked next.
Decision Records, Test Results, and Trading Performance Are Different Kinds of Information
| What We Publish | What It Shows |
|---|---|
| Daily decision records | What we recorded about our own trading decisions, what fills we could confirm, and what remained unresolved. |
| Weekly reviews | What moved markets during the week, how those moves fit into trends over several weeks, and what to watch in the coming week. Earlier issues reviewed our own account. |
| Backtests on historical data | Calculation results from trying a rule under specific conditions, and comparisons of those results. These are not results earned from actual trades. |
| Trading performance | The performance of the account we manage with our own funds. It does not prove the effect of any individual rule or of AI alone. |
That a record of a decision exists is not proof that the decision was correct or that it will remain effective in the future. We treat having obtained a result and having completed a sufficient test as separate matters.
We Distinguish Our Decision Records from Trading Instructions to Readers
Articles may name individual companies and ticker codes, and may describe what we decided to buy or sell. These are records for our own funds and our own trading conditions. They do not recommend or instruct readers to make the same trades.
A decision made before trading is also separate from a trade actually being executed (a confirmed fill). When we publish fills, we present them separately from the decisions, as trades we were able to confirm. Each article describes decisions as of the date it covers.
We do not provide investment consultations tailored to a reader's personal financial situation, individual instructions on trade timing, or operation of securities accounts or management of assets on a reader's behalf. Daily decision records cover completed US trading days. The English edition follows the Japanese issue after review. These records are not real-time trade alerts.
Timing of Information and How Articles Are Produced
Decision records and weekly reviews are created and published by software. Checks are run before publication, but a person does not read every article before it is published. AI is also used in producing research articles.
We distinguish the information used at the time of a decision from materials checked afterward. We explain decisions and counterarguments using the sources that support them.
The date information was checked, the date a decision was made, and the date an article was published or updated may differ. Please check the period covered and the information date for each article. Errors can occur in the information or in the automated processing, and we do not guarantee accuracy, completeness, or timeliness, or fitness for any particular purpose.
How to Read Performance, and Investment Risk
The latest confirmed trading performance can be viewed without registration on the Trading Performance page. It uses TWR (time-weighted return), which adjusts for the effect of deposits and withdrawals, and is shown together with the period, currency, comparison benchmark, and interim declines. Please also note differences in comparison conditions, such as how dividends are treated.
Performance published in past articles is the value as of the reference date given in the performance section of that article. Please check the date so that you do not confuse it with the latest performance.
Past calculations, trading performance, and internal scores do not guarantee future profits. Prices of financial products and exchange rates fluctuate and may fall below the amount invested. If you invest, please verify the necessary information yourself and decide at your own judgment and responsibility.
Free Registration and Possible Future Paid Access
Currently, AlpAI Lab articles are provided free of charge. Free registration is for reading daily decision records and weekly reviews starting from the latest articles, and registration alone does not incur any charge.
If we offer articles for a fee in the future, subscription fees or per-article purchase prices will be payment for access to the articles described. They are not an investment in our company, nor deposits for managing readers' funds. A subscription also does not entitle you to individual investment consultation.
For what you can read and how to register, see Free Registration Guide; for the terms of use, see the Terms of Use.
Conflicts of Interest and Corrections
Because we trade US stocks with our own funds, we may hold or trade the stocks covered in articles. The financial interests of our company or related parties do not necessarily align with those of readers. Conflicts of interest that we identify will be disclosed in the article or on a related page.
If we find an error, we correct it. Later results do not change the record of what we decided at the time.
For inquiries about errors in articles or about our information sources, please send them to alpai@gridstart.co.jp. Our operating company and our article production policy are also described on About Us.
Japanese original of this policy.