One Red Flag Was Never Enough: I Counted 104,153 Lines Thai Companies Wrote About Themselves, and All Six Lenses Lit Together Only 9 Times

Two of your own mistakes. Identical on the page. Nowhere near each other in outcome.

The first. You open a company's accounts and hit the line that raises the hair on your arms — trade receivables growing faster than revenue, the collection period stretching out. You have read that this is the classic signature of "phantom profit." You sell.

That stock ran for another two years.

The second. You hit the exact same line at a different company. This time you remember the lesson. You tell yourself "the whole industry runs this way, don't overreact." You hold.

Eight months later that stock was carrying an SP sign.

The question worth asking is not "which time was I unlucky." It is far more direct than that:

What made those two different, when the signal you saw was identical down to the character?

The answer is short and more irritating than you want it to be — the answer is not in the signal. Stare at it as long as you like, read it as closely as you like, know as much accounting theory as you like. A single signal does not carry enough information to separate those two cases at all.

This article lays out what the data actually says. I extracted the management discussion and analysis (MD&A) that Thai listed companies write about themselves — 104,153 lines from 896 companies across FY2021–FY2026 — through six operating lenses, then counted how it distributes.

What came back killed two of my own assumptions. Both are below.

---

Your problem is not that you cannot find the signal. It is that you have no denominator

Start with the number that stopped me on the first row.

Of the six lenses I extracted, the one called "working capital" — receivables aging, payables stretched, the cash cycle lengthening, collections failing — appears in 882 of 896 companies.

That is 98% of the market.

Stop here, because almost everyone's first reaction walks straight into the wrong ditch.

That reaction is "fine, then it means nothing, if every company has it."

And that is wrong the other way — a different ditch from the one you fell into when you panicked, and just as deep.

Because inside that same lens sit 8,260 lines graded high severity (4–5 out of 5) — the most of any of the six, and nothing else is close.

Read these two sentences back to back. Same lens, same fiscal year:

"The increase in trade receivables and contract assets reflected higher working capital requirements in line with revenue and project growth. The Company therefore continues to closely monitor collections, credit terms, and the cash conversion cycle."

Read: receivables grew because the project book grew, and management is telling you it has eyes on the collection cycle. Growth working capital. Nothing here you have to act on.

— ASEFA, FY2026 Q2, page 4 · original filing, Thailand's SEC

"As of March 31, 2026, increased by 667 days compared to the previous year. This is due to an outstanding balance of uncollectible receivables, resulting in a low accounts receivable turnover ratio of 0.52 times."

Read: the collection period blew out by 667 days in a single year, and the reason the company gives is receivables it cannot collect. At 0.52x turnover the book barely moves. This is revenue already booked that has not turned into cash.

— BTW, FY2026 Q1, page 1 · original filing at the SEC

Same heading. Same category. Different planets.

The first is a company whose receivables grew because its project work grew, saying itself that it is watching. The second is a collection period that stretched 667 days in one year because the cash genuinely is not arriving.

Read only one of them and you have no idea which one you are holding.

Here is the sentence to keep from this section:

The presence of a signal tells you almost nothing. Severity does — and you only know severity once you have seen the other 895 companies.

People who read filings for a living call this the base rate: the rate across the whole population. It is the one thing that separates a reaction from an analysis.

The problem is that a base rate cannot be bought with diligence. Read every page of every quarter of every name you own and you still hold only a numerator. The denominator never arrives.

---

Six lenses management wrote down itself

Map first: what the six are, and why these six.

Form 56-1 (One Report) and the quarterly filings contain a section called management discussion and analysis (MD&A), written by management itself — what happened to the business this year, and why the numbers came out the way they did. (I mapped the whole document in the piece on 56-1 structure)

That section is not numbers. It is language, and language always arrives before the numbers do — a company has to describe the pressure it is under before that pressure becomes a line in the income statement.

The six lenses I picked are the six that hit real cash and hit operating survival — not the ones that are easy to dress up in accounting:

What the six share: the company is obliged to write them down itself under the disclosure rules. Not rumour, not a broker note, not a guess — the words management chose, in a document it signed and filed.

Before we go further: the originals filed with the SEC (Thailand's Securities and Exchange Commission) in this axis are almost entirely in English (65 Thai lines out of 104,153). Every quote in this article is therefore the English, word for word, with my read of it directly underneath — the read is mine and is never dressed up as the company's words. The text inside the box is the real text from the real file.

---

Which lens shows up most, which one hurts when it does — and they are not the same lens

This is the table that killed my first assumption.

I counted two things at once: one, how many lines each lens produced and in how many companies; two, what percentage of those lines graded high severity (4–5 out of 5).

(Measured against the live index on 27 August 2026 · FY2021–FY2026 · the 896-company base is companies with data in this axis, which is fewer than the total searchable universe, because not every company writes about all six.)

Read the left column and the right column together and you will see what I saw.

The assumption I walked in with was "rare equals important." It sounds reasonable: if companies seldom write about something, it must take an unusual event to force the disclosure.

The data says no.

Workforce restructuring is the least common of the six — 8,038 lines, five times fewer than working capital. It is also the least severe: only 3.9% reach the high band.

Customer concentration sits mid-pack on frequency (12,810 lines), but when it appears it is severe 37% of the time — the highest of the six, nearly ten times the workforce lens.

"How often it appears" and "how hard it hits when it appears" are completely different axes — and the heuristic filing readers run without noticing, that rare means important, does not work on this dataset.

Why that happens: look at the real thing.

Workforce: the rarest lens, and most of it is payroll

This is what an ordinary workforce line looks like in the index:

"The significant change was due to an increase in employee benefit expenses of Baht 213.27 million. The main reason came from an increase in employee salaries due to annual merit adjustment."

Read: ฿213.27 million more in employee benefit cost, driven by the annual merit round. Payroll doing what payroll does. It lands in the lens because the delta was material, not because anything broke.

— AOT, FY2026 Q2, page 10 · original filing at the SEC

An annual merit raise. That is the whole event.

That is why the workforce lens grades severe only 3.9% of the time — most of its mass is wages moving up in the normal course, which the healthiest companies in the country write down too.

The other 3.9% looks nothing like it:

"the board approved the 2024 business plan, which includes the termination of production and distribution of plastic sacks and flexible packaging, as well as the dismissal of all employees in production, marketing, and sales, along with certain support roles."

Read: the board did not trim a cost line. It shut the plastic sacks and flexible packaging business and dismissed everyone who made it, marketed it and sold it. A segment exit, filed as a business plan.

— NEP, FY2024, page 2 · original filing at the SEC

In the same document, page 3 gives the count outright: 108 employees terminated, in three waves.

Two poles of one lens. And if your tool can only tell you "found something about labour," it has done nothing for you.

Customer concentration: the lens that hurts most when it fires — and it is not an accusation

37% is high enough to need explaining, and the explanation is blunt: customer concentration rarely has a mild version. If a company writes about it, it is usually because it is large enough to move revenue.

The most important point about this lens, though, is that it is not an allegation. Take this one:

"Most of the company's products are sold to Provincial Electricity Authority ( PEA), Metropolitan Electricity Authority (MEA) and companies in the engineering and construction business sector."

Read: revenue runs to the two state electricity utilities and to engineering and construction contractors. About the most creditworthy customer base available in this market, and also about the narrowest.

— AI, FY2026 Q1, page 2 · original filing at the SEC

This company is not confessing to anything. It is describing the structure of its business — selling electrical insulators to the power utilities, the steadiest customers anyone could name.

But that structure is the fragility. If the utilities slow their capex, there is no other customer group to take up the slack.

And sometimes what hits has nothing to do with the business at all:

"The Thai-Cambodian border conflict since mid-2025 resulted in a significant decline in Cambodian patient revenues, down 78% YoY in 1Q26, limiting overall international patient revenue growth to 1% YoY."

Read: a border conflict, not an operating decision, took Cambodian patient revenue down 78% year on year in 1Q26 and pinned total international patient growth at 1%. Geographic concentration behaves exactly like customer concentration.

— BDMS, FY2026 Q1, page 2 · original filing at the SEC

This is one of the largest and strongest companies on the exchange, and the lens still catches it — because this lens measures dependence, not management quality.

Read the output of this lens as "a list of companies with problems" and you have misread it from the first line.

FX: the lens that exists to stop you overreacting

I like this one in particular, because its main job is to reduce alarm, not raise it.

"Excluding the impact of an unrealised FX loss of Baht (1,306.9) million, core loss in 1H2026 was Baht (439.7) million compared to core profit of Baht 453.3 million in the same period last year."

Read: two separate facts stacked in one sentence, and only one of them is non-cash. Strip the revaluation and the core business still crossed from profit into loss year on year.

— AAV, FY2026 Q2, page 7 · original filing at the SEC

Read it slowly, because there are two things in there.

First: the ฿1,306.9 million loss you see in the bottom line has not happened. It is foreign-currency liabilities remarked at the period-end rate. Not one baht of cash left the company. If the currency turns, it reverses into an accounting gain next period.

Second: even with that stripped out, the core business still swung from ฿453.3 million of profit to ฿439.7 million of loss.

Whoever looks only at the net income line overreacts. Whoever hears "accounting loss" and relaxes on the spot underreacts. The truth sits between them, and it sits in a paragraph the company wrote itself.

Inventory: watch the write-down rate, not the balance

The most common mistake in this lens is asking "is there a lot of stock," which tells you almost nothing. What tells you something is whether the company has admitted the stock is worth less than its carrying value.

"The Group recognized NRV inventory write-downs of THB 1,330.68 million across the EV and lithium-ion battery segments."

Read: ฿1,330.68 million taken off the carrying value of EV and lithium-ion battery inventory. The company has conceded in its own accounts that this stock does not clear at the price it was booked at.

— EA, FY2026 Q1, page 6 · original filing at the SEC

A write-down is an accounting confession that what sits in the warehouse will not fetch the price it was carried at. Once booked, it is margin pressure that has already happened, not margin pressure on the way.

But growing stock is not automatically bad:

"The Company's cash conversion cycle (CCC) stood at 5 days, while the average inventory turnover period increased to 59 days, primarily due to higher inventory levels to support the planned product distribution in Q3/2026, as well as additional inventory stocking for certain product categories expected to experience price increases."

Read: inventory days out to 59, and the cash conversion cycle still sitting at 5. They stocked ahead of a Q3 distribution plan and ahead of price increases they expect to pay. A decision, not a symptom.

— ADVICE, FY2026 Q2, page 12 · original filing at the SEC

Stock rose in both cases. Here the cash cycle is still 5 days and the company gave the reason in advance. Same number, opposite meaning.

Litigation and regulation: it does not mean "sued"

This lens is the most misread of the six, because the word "litigation" makes people think scandal on sight.

Most of what sits in it is nothing of the kind. The clearest examples are rule changes and contract terms:

"The operating profit for Q1/26 YoY decreased by 75% or THB 639 million, mainly due to the commencement of electricity supply under the new PSA on October 26, 2025, which the electricity sales unit price was lower than the previous contract."

Read: supply began under a new PSA on 26 October 2025 at a lower unit price than the old contract, and 75% of operating profit — ฿639 million — went with it. Nobody did anything wrong. The contract repriced, and it repriced down.

— EGCO, FY2026 Q1, page 11 · original filing at the SEC

Nobody broke a rule. The old contract ended, the new one prices lower, and ฿639 million of profit is gone — and that is a structural change in the revenue base, which a holder should know before concluding "management had a worse year."

And sometimes the company surfacing in this lens is the side that won:

"The Court of First Instance rendered its judgment ordering the associate to pay THB 37.78 million, together with interest at the rate of 15% per annum on the principal amount of THB 30.00 million, calculated from the filing date (10 April 2025) until full settlement is made to the Company."

Read: the money runs toward this company, not away from it. It is the plaintiff, it holds a first-instance judgment, and 15% a year is accruing on the principal until settlement.

— GPI, FY2026 Q1, page 10 · original filing at the SEC

"until full settlement is made to the Company" — this company is the plaintiff, and it won.

See a name come back from this lens and conclude "this stock has a legal problem" and you may be reading a judgment in the company's favour exactly backwards.

---

What appears when you look at all six at once

Now the part this article was written for.

Everything above is still one lens at a time, which is what broker notes and how-to-read-filings articles do.

The question nobody has answered: look at all six at once, in one company, in one year — does a pattern appear?

I counted. The method is plain, and here is the full definition so you can argue with me:

Definition: take every (company × fiscal year) pair with data in these six lenses — 4,729 pairs in total. Then count how many lenses in that pair carry at least one line graded high severity (4–5). A lens that clears that bar I call "lit."

Here is the whole result:

terminal.boomleverage.com

Distribution of 'lit' lenses — six operating-quality lenses

4,729 pairs (company × fiscal year) · FY2021–FY2026 · measured on the live index, 27 Aug 2026

• 🟡1 lens lit

  1,401

  an ordinary bad year

• 🟡2 lenses lit

  1,088

• 🟠3 lenses lit

  603

• 🟠4 lenses lit

  291

• 🔴5 lenses lit

  78

• 🔴All 6 lit

  9

  across 6 fiscal years

Two numbers to leave in your head.

First: 1–2 lenses lit = 2,489 pairs, more than half of everything. This is what an ordinary bad year looks like in Thai equities. A company takes one or two hard hits in a year and moves through it. Common enough to be the baseline.

Second: four lenses or more = 378 pairs, roughly 8%. All six lit at once happened 9 times across six fiscal years.

Back to the two mistakes at the top of this article.

You saw the same signal both times. What you never saw is how many lenses were lit that time — and that information is not in the line you read. It is in the five topics you did not open.

---

What this number has not proved — and I am not going to pretend it has

In the first draft of this article I wrote this sentence:

"Four or more lenses lit is the signal that the problem is structural, not just a bad year."

I pulled it, and you should know why.

Because I do not have the evidence to say it. I can count frequency. I have not run the backtest on what actually happened afterwards to companies with four lenses lit in a given year — more defaults or not, more SP flags or not, more value destroyed or not.

Leaving that sentence in would have been inventing a risk score out of thin air and selling it to you untested, which is exactly what I have attacked other people for in earlier pieces.

There is also an alternative explanation I cannot rule out yet: a company with complex operations, several business lines, sales in several countries and concessions with the state has more topics to explain by construction. The number of lenses lit may reflect business complexity as much as it reflects trouble.

So what can this number actually support?

One thing, and it is a useful thing: rarity.

One lens lit is ordinary. Four lit occurs in 8% of cases. All six lit happened nine times in six years. If you are looking at a company with four lenses lit, you are looking at something statistically abnormal, and it deserves more of your time than the queue behind it.

The correct frame for this tool is "decide whose filings to read first," not "decide who is good and who is bad."

A lit lens is not a verdict. It is a flag that management itself wrote something down that needed explaining — and reading that explanation is still your job.

---

What "read all of it" actually costs

By now you can see why I say a base rate cannot be bought with diligence.

Let me put numbers on what "read the whole MD&A" means in practice. First, the claim I will not make: I am not going to tell you humans read slower than machines. I do not have that data, and figures of the "a human needs 20 minutes per topic" variety are figures I would have to invent, which breaks the rule I set for myself.

What I do have real numbers for is coverage — a different thing from speed, and the more important one.

And here is what I will say straight to anyone who reads filings for a living:

Your heuristics work. Good analysts do not read every line. They jump to the sections where problems like to hide, sweep for familiar words, and find what they need in minutes. I am not telling you that method is bad. It is not bad at all.

But heuristics work superbly on the companies you already open, and not at all on the companies you have never opened.

That is the problem. The problem is not human speed. The problem is coverage — and coverage is the only thing that can hand you a base rate.

When the ground to cover is larger than anyone can actually cover, everybody quietly runs an abridged version, and the abridged version becomes the industry standard without anyone announcing it.

These signals are not hidden. They have simply never been on anybody's list.

---

The tool I built to do this

This is the search axis called "operating quality" on the Terminal — live now, in BETA.

---

Limits we would rather state ourselves than have you find

This section matters as much as the rest of the article.

1. This is AI extraction, not a financial opinion. What the system does is read what the company wrote and classify what it is about. The severity grade is the model's assessment — not an expert's verdict, and not investment advice.

2. The 907 lines whose page number cannot be verified, we do not serve. Not deleted — quarantined, with a count of how many were held back. Text that cannot be pointed back to a page should not count as evidence.

3. This axis covers FY2021–FY2026 only. We hold another 57,540 lines extracted from FY2016–FY2020, and not one of them has a verified page number yet. So it is not searchable, and we will not imply this axis goes back ten years.

4. "Lit" is not a score, not a ranking, and not a verdict. We do not rank companies by risk, we do not build leaderboards, and we do not paint anyone as close to failing. Counting the topics a company had to explain and judging that company to be in trouble are two different exercises — every number in this article is an aggregate distribution across the whole market.

5. FY2026 is not a complete year. In the per-fiscal-year counts, FY2026 holds 11,457 lines against 22,485 in FY2025 — not because there is less to report, but because not every quarter has been filed.

---

Count the lenses lit on your own book

You can do this by hand — genuinely. All six topics sit in Form 56-1 (One Report) and the quarterly filings, free to download from the SEC. I have written up how to read each lens in enough detail that you can open a report today and work through it yourself.

Try the real thing: Boom Leverage Terminal searches all six lenses at once from a single line of Thai, and returns the text word for word with a page number and a link to the SEC original, every time — start free, 10 credits a day, no card. The "operating quality" axis is in BETA and bills at the standard search rate.

The first query I want you to run: do not go hunting for trouble — search the position you are most confident about, and see how many lenses have something in them. That is the one you stopped checking longest ago, and the one where a base rate does the most work for you. · For how the packages differ, read the straight guide to picking one · Team and institutional (seats · Excel export · API) — the Enterprise page or contact@boomleverage.com

Back to the question at the top

Two of your mistakes, identical on the page.

The first time you sold on one lens lit — and it was one of the 1,401 most ordinary cases in the market.

The second time you held on the same single lens — and you never opened the other five topics.

What differed was not your diligence, and not your accounting. What differed is that you saw one box out of six, in one company out of 896, in one year out of six.

The data was public the entire time. The companies wrote it. They filed it with the SEC. It has always been free to download.

It has simply never been laid side by side where you could see it all at once.

---

Related reading: How to read the auditor's opinion properly · Track where the cash actually went · Catch a signal moving across an entire sector · The full map of Form 56-1 (One Report)

Disclaimer: This article is written for education and to explain how to read documents that listed companies disclose publicly. It is not investment advice, it recommends no security, and it guarantees no outcome. · All quoted text is reproduced word for word from filings submitted to the SEC, with the fiscal year and page number given so it can be checked · A company appearing in this article is not an indication that it has financial problems, has broken the law, or has dressed up its accounts — every company here is cited to show how one lens reads, one lens per company, and several are examples of routine disclosure or of the side that won its case · "Lit" is not a risk score, not a ranking, and has not been backtested against real outcomes. Every figure here is an aggregate distribution across the whole market · Classification and severity grading are the output of AI extraction, not the opinion of a licensed professional · All statistics were measured against the index as of 27 August 2026, which changes as new documents enter the archive · Investing carries risk; past results do not guarantee future ones