This page explains what data goes into Stock Rollbar’s content and how it is produced. You can only read a report properly if you can check where its claims come from.
Who Writes This
Prunus mume — operator of Stock Rollbar
A fairly ordinary person in Korea, in my mid-thirties. I do not
work in finance, I am not a licensed investment adviser, and I hold no certifications or
institutional track record.
As an individual investor, in 2026 I made roughly what an average Korean salary
pays from the market, and I aim to repeat that in 2027 and 2028. But that came from my
own discretionary judgement. What this blog is testing is the question that follows:
take the human judgement out, and can rules alone do the same thing?
Investing principle — find fundamentally sound businesses
What I trust is not the shape of a chart but a company that actually makes
money. Is revenue growing? Do operating margins hold up? Is cash genuinely coming in?
Is the debt load survivable? A name that fails these drops out of the pool no matter how
good its technical setup looks. The system exists to check those conditions across
hundreds of names every day, without tiring and without emotion.
Fundamentals are the goal; the system is only how I find them.
That is why the stock reports here do not say “buy” or “sell.” They start from
“how does this company make money?” I am learning as I write them, and the intent
is that we look at the same things together.
Where the system stands today
The system currently runs on a paper-trading account, and across 111
filled-and-closed positions its expected value is negative (as of July 2026). That
number is published on this site too. My personal results and this system’s results are
two different things, and not blending them is the minimum condition for this record to
mean anything.
So what I can offer is not an expert forecast but a record: what criteria
picked a stock, when the system entered and exited, and what turned out to be wrong —
with the numbers attached. If this experiment ends in failure, I will write that it
failed.
Data Sources
- Prices, financials, consensus — Yahoo Finance: closing prices, quarterly results, valuation multiples, and analyst price targets.
- Primary filings — SEC EDGAR for US names, DART (Financial Supervisory Service) for Korean names.
- News headlines — Google News RSS. We show the headline and the publisher, and the headline links to the publisher’s original article. We do not reproduce article bodies.
- Trading records — actual orders and fills from our own system via the Korea Investment & Securities (KIS) API.
Third-party data can be delayed or wrong. Quarterly results in particular can take more than a day to appear after a company reports; reports affected by this carry a notice and the section is refilled once the data updates.
How Reports Are Produced
Our daily reports and stock deep dives are produced by a pipeline we built that gathers and computes the data, then generates the written sections from it. There is no reason to hide this, so we state it plainly.
- Numbers, tables, and charts are computed directly from the sources above. They are not hand-edited.
- Written sections (At a Glance, How It Makes Money, Scenarios, news commentary) are drafted by a large language model that receives those numbers as input. The model is instructed never to invent figures absent from the supplied data.
- Sections differ by company. A loss-making company gets balance-sheet strength and cash runway instead of earnings multiples; a dividend payer gets dividend metrics. The questions that matter genuinely differ.
- Human oversight — the operator reviews what is published and corrects or removes anything with a data error or inappropriate framing.
How Stocks Are Selected
Which names we cover is decided by the criteria below — never by advertising or partnerships. Fundamentals come before technicals: a stock that fails this gate is dropped no matter how good its chart looks.
- Four consecutive profitable quarters — a hard requirement, verified against quarterly filings from DART (Korea) and SEC EDGAR (US).
- Missing data means excluded, not admitted (fail-closed). Letting a name through when the check fails would mix in unverified companies, so we drop it instead.
- Valuation floor — names below our value-score threshold are excluded.
- Technical signals — trend and volatility conditions rank only the names that already cleared the above.
The principle is to ask first whether revenue is growing, whether operating margins hold, and whether cash actually arrives. That is why the reports start from “how does this company make money?” rather than “buy” or “sell.”
About the Trading Records
The trading records published here come from a paper-trading account used to validate the system. It runs under live prices and realistic execution rules, but no real capital is deployed. Both the calls that worked and the ones that ended in losses are published.
Corrections
When a data or factual error is found after publication, we fix it — and where a correction changes a number, we note that in the report. If you spot an error, please tell us via the contact page.
Advertising and Conflicts of Interest
This site is funded by advertising. We accept no payment or consideration for covering any particular stock, and advertisers have no involvement in which stocks we cover or what the reports say. Coverage is chosen automatically by our scanner’s scores and by market events such as earnings dates. For advertising and cookies, see the privacy policy.
Disclaimer
Everything on this site is for information only and is not investment advice or a solicitation. The operator is not a licensed investment adviser and does not provide advice tailored to individual circumstances. All investment decisions and their outcomes are your own responsibility.
Last updated: 2026-07-31