Three Axes Lost Their BETA Badge by Rule, Not by Hand — Eleven Now Live on the Thai Terminal, and Exactly Why the Last Three Still Say BETA

"Impairment of investments in subsidiaries As at 31 December 2025, the Company has investments amounting to Baht 183.77 million in separate financial statements. In accordance with Thai Financial Reporting Standards, investments in subsidiaries are required to be tested when there is an indicator of impairment." That is the auditor of 2S Metal, writing a Key Audit Matter into the FY2025 audit report (page 2 of the auditor's report filed for FY2025). You can open it on the SEC filing page for 2S. Until 25 October, a passage like that came up on the Terminal under a chip marked BETA. Today it comes up without one. Nobody flipped a switch to make that happen. A number did.

This piece is the full count for the Thai market. It covers which axes are live, which ones lost the badge, which ones still carry it, and the exact figure that holds each of those back. The board below is the short version. Tap any axis to see its numbers.

Terminal axes · Thai market

Which axes still say BETA — and why

Measured 31 Oct 2026 · 09:30 Bangkok time. Rule: a BETA axis loses the badge after 14 days in a row with ≥98% of served rows verified against the source, and every other door green. Tap an axis.

All 11BETA removed · 3Never BETA · 5Still BETA · 3

• Financials (MD&A)Live
• Risk factorsLive
• Auditor opinionUnlocked
• GovernanceUnlocked
• Business & competitionUnlocked
• Insider trades (Form 59)Live
• Major shareholders (Form 246-2)Live
• Sustainability (SET ESG data)Live
• Operating qualityBETA
• SET early-warning newsBETA
• Opportunity Day (earnings calls)BETA

Auditor opinion

BETA removed

Reads

The auditor's report and its Key Audit Matters

Where you find it

Chip on Scan

Rows verified against the source

99.66% 14,202/14,251

Days in a row at ≥98%20/14

Above 98% every day since the board started measuring on 12 Oct; the 14th day was 25 Oct. Badge removed by the rule, not by hand.

Read this before the numbers: everything above and below was read from the Terminal's own pipeline snapshot on 31 October 2026, around 09:30 Bangkok time. The snapshot refreshes every five minutes and this article does not. A streak that reads 20 days today will read 21 tomorrow. A citation share can move a few hundredths as the index is rebuilt each night. Treat every figure here as dated.

What BETA means on the Terminal, and what it does not

BETA on a Terminal axis has never meant "the data might be wrong". Every row on every axis, badge or no badge, has to pass the same citation gate before you see it. The gate checks that the verbatim passage really sits on the stated page of the source document. A row that fails the check is withheld. It is not shown and it is not "shown with a warning".

What BETA has meant is narrower. The axis is new, or its coverage is not yet complete, or the way we check it is not yet as strong as on the axes we have sold the longest. The trouble is that "new" and "not yet complete" are judgments. A judgment that lives in someone's head (mine included) tends to drift towards whatever is convenient that week.

So on 12 October 2026 the badge stopped being a hand-set flag. The rule that replaced it has three parts:

1. The served citation share must be at least 98%. That is the share of rows on the axis that verify against the source document.
2. It must stay there for 14 consecutive days. Each day counts by its lowest reading, not its average. One bad refresh breaks the streak.
3. Every other door on the axis must be green as well. That covers data extraction, the index, the guard and the serve allowlist. A good citation share on an axis that is still missing filings is not a finished axis.

When all three hold, the board puts the axis on its unlocked list. The Terminal reads that list and drops the badge. If the list cannot be read (a fetch error, an empty row, anything), the Terminal keeps every badge on. A badge that stays one refresh too long costs nobody anything. A badge that drops because a request failed would be the product claiming maturity it has not measured.

The model-validation habit behind this: in a bank, a model does not graduate from "monitoring" because the team likes it. It graduates when a pre-agreed metric holds over a pre-agreed window, and the metric and the window are written down before anyone looks at the result. The 98% / 14-day rule is that habit applied to a product badge.

The count: eleven axes live, eight without a badge

Three notes so the table is not over-read:

• "Live" means you can use it today, badge or not. All eleven are served. The three BETA axes return results like any other. The badge tells you how far the axis is from the bar, not whether it works.
• The last three "never BETA" rows are company panels, not search chips. Insider trades, major shareholders and sustainability answer "what about this company". They sit in the panels under a company lookup on the Scan page rather than in the market-wide chip row.
• The headline citation figure on our other pages covers FY2021 onward. Older years carry their own coverage number and are never blended in. The percentages above are the served-index figures the badge rule uses.

The three that unlocked — what each one actually returns

The streak history shows something useful about the unlock date. All three axes have been above 98% on every single day since the board started measuring on 12 October. The 14th qualifying day was 25 October, and the badge dropped then. By 31 October the streak stood at 20 days. Their lowest daily readings over that window were 99.59% (Auditor opinion), 99.10% (Governance) and 98.85% (Business & competition).

Here is one verified row from each, all from the same company and the same fiscal year, so you can see how the three axes read different parts of one filing set.

Auditor opinion — reads the auditor's report. The passage that opens this article is a Key Audit Matter from 2S Metal's FY2025 audit report. On this axis the voice is the auditor's, not management's. That is the reason the axis exists: it is the one part of the annual filing written by an outsider who signs under professional liability. I covered how to read it in the auditor's-report article.

Governance — reads Form 56-1 Part 2. From the same company's FY2025 One Report, printed page 88 (PDF page 93):

"Currently, the company's Board of Directors consists of 10 members, comprising 3 executive directors and 5 independent directors, which accounts for half of the total directors, thereby ensuring a balance in voting and independent consideration of various matters." — 2S Metal, One Report FY2025 · SEC filing

Business & competition — reads Form 56-1 Part 1. Same report, printed page 10 (PDF page 14):

"the company can deliver goods to each branch quickly according to the needs of customers in no more than three days which is a competitive advantage for the company." — 2S Metal, One Report FY2025 · SEC filing

All three rows carry a verified: true verdict from an exact match against the source PDF. Before quoting them here I opened the PDFs and found the same text on the stated pages myself.

Read the Business axis with one caution: everything in Part 1 is the company describing itself. A competitive advantage, a market share, a delivery time — none of it is audited. The Terminal shows exactly what the company wrote and where. It does not certify that the claim is true. Treat these rows as statements to test, not facts to rely on.

The three that still say BETA, and the number holding each one back

Operating quality — one door short

Operating quality reads the MD&A through six lenses: customer concentration, FX, inventory, working capital, labour and litigation. On citations it has already done everything the rule asks. It has held 99.11% for the same 20 days as the three axes above, and it would have unlocked on 25 October on that measure alone.

It still says BETA because of the third part of the rule. Extraction for this axis has reached 17,679 of 17,683 filings. That is 100.0% when rounded, but four filings short in fact. The board treats a data door that is "partial" as not green, and the axis waits until it is.

A mistake of ours, and why it is in this article: on the morning of 31 October the Terminal was showing Operating quality without its badge. The list the Terminal reads had been built from the streak alone and skipped the other doors. So the board said "locked" while the product said "unlocked". The same morning the list was changed to include only axes the board itself marks as unlocked, and the badge came back within one refresh. I would rather tell you that than have you find two screens disagreeing.

SET early-warning news — below the bar, streak not started

This axis is a filtered set: early-warning events from announcements on the SET website since 2021, not every piece of company news. Its served citation share is 91.32% (3,082 of 3,375 rows). That is below 98%, so its streak reads zero and the 14-day count has not begun. Its lowest daily reading over the window was 91.31% and its highest 91.37%. It is stable, just stable below the bar.

The cause is coverage of the source text, not invented rows. Some announcements are scanned images with no text layer, and the system has read 3,325 of the 3,375 so far. A row whose announcement cannot be read cannot be checked word for word, so it does not count as verified. Rows that do not verify are withheld from results rather than shown with a caveat.

Opportunity Day — a different kind of evidence

Opportunity Day is the axis I think of as the moat. It holds our AI transcripts of the SET's quarterly earnings-call sessions, each row linked to the minute in the original recording. I explained the format in the Opportunity Day article.

It cannot pass the badge rule as written, and that is by design. The citation check on every other axis compares a passage with a page in a PDF. Here the evidence is a timestamp in a video, so that check does not run, and the board records no verdict at all. No verdict counts as not passing. The axis stays locked: it fails closed, which is the right direction to fail.

What we do run on it is different. Every row links to its exact minute in the clip so you can listen for yourself. Transcripts above a word-error ceiling are kept out of the index. A second-model check that re-transcribes disputed passages has been through 1,873 of 8,441 clips so far. Until spoken evidence has its own pass mark set at the same bar as the PDF axes, this axis keeps BETA. That is the honest description of where it stands.

Two parts held out on purpose

Two sub-sections are deliberately kept out of the serve promise. They do not count against any axis above:

• Sustainability as a 56-1 section is not yet in the index at all. It is held until it is, and the SET's own ESG data tables, listed above, are a separate source that is already live.
• Opportunities and risks from the annual 56-1 overlaps the Risk factors axis we already sell, even though it comes from a different document. It is held so the same question does not return two near-duplicate answers.

How to check a BETA axis yourself before you rely on it

You do not need to trust the badge or its absence. Here is the workflow I use on any axis, BETA or not:

1. Run your search on the axis you care about, and note the badge. BETA tells you how far the axis is from the bar, not whether its rows are real.
2. Open three results at random, not the top three. Ranking pushes the cleanest passages up, so random rows tell you more about the axis as a whole.
3. Check each one on the source. Click through to the filing or the clip minute, then find the quote on the stated page or second. If it is not there, that is a bug, and I want to hear about it.
4. Ask what is missing, not only what is there. On SET news, a scanned announcement that could not be read will simply not appear. On Opportunity Day, a session whose transcript failed the word-error ceiling will not appear either. Absence on a BETA axis is weaker evidence than absence on a finished one.
5. For the decision itself, go back to the document. The Terminal finds the page. The judgment about what the page means is yours.

Five traps when reading these numbers

As someone who has sat on the validation side, these are the ones I would flag first:

1. "Unlocked" is not "complete coverage of every company". The rule measures whether what is served verifies, and whether the extraction doors are green. A company whose document never reached us is not counted as a failure, because it is not in the denominator at all.
2. A citation share measures that the quote exists, not that the interpretation is right. 99.66% on Auditor opinion means the passages are really on those pages. It says nothing about whether a summary written over them is correct. That control is described in the validator article.
3. Do not compare percentages across axes as a quality ranking. Insider trades at 100% and Business & competition at 98.85% are checking very different things: structured filing fields versus free text pulled from narrative pages.
4. A badge can come back. The rule runs both ways. If an unlocked axis drops below 98% for a day, or one of its doors goes red, the board marks it locked again and the badge returns. That is intended.
5. The numbers here are dated. Every figure in this piece is a reading from 31 October 2026. Before you quote one, check the live figure.

Overview

On the Thai market today, eleven axes are live on the Terminal. Eight carry no BETA badge. Five never did: Financials, Risk factors, and the Insider, Major-shareholder and Sustainability panels. Three earned their way out on 25 October under a rule that nobody can override by hand: Auditor opinion, Governance and Business & competition. Three still say BETA. Operating quality is four filings from a green data door. SET early-warning news sits at 91.32%, below the 98% bar. Opportunity Day carries a kind of evidence the PDF check cannot score yet. The badge will move when the numbers do, and only then.

If you run a research, risk or compliance function and want the unlock history, the per-axis verdicts and the door-by-door board behind this article for your own review, that conversation is at boomleverage.com/th/enterprise or contact@boomleverage.com.

If you want to try it yourself, sign in at terminal.boomleverage.com, run one search on an unlocked axis and one on a BETA axis, and open three rows from each before you believe either.

Disclaimer: This article explains how a data product labels the maturity of its features, and is written for education. It is not investment advice, it recommends no security, and it guarantees no outcome. · The 2S Metal passages are quoted verbatim from filings published on the SEC's website and are used only to show what each axis reads, not as a view on the company · All figures are readings from the Terminal's pipeline snapshot on 31 October 2026 and will change · Verified citations show that a passage exists on the stated page, not that the company's statement is accurate · Investing carries risk; past results do not guarantee future ones