DATA AUDIT — THANI¶
Overall integrity grade: C (usable with caveats)¶
Confidence the data is REAL (not hallucinated): Medium¶
The price series, ticker, and SET URL structure all resolve to the real THANI / Ratchthani Leasing entity, but two of the four SET pages returned 404 errors, news coverage of the actual company is essentially zero, and the price file is dated into the future (2026) relative to any plausible "today."
Findings by source¶
Prices — CAVEAT¶
- Date anomaly: Recent sessions run 2026-03-25 → 2026-06-22, and the package itself is stamped
Data package built: 2026-06-23T17:17:06Z. The dates are internally consistent with each other, but a downstream persona should NOT assume "today" is calendar-2025. Treat the timestamp as authoritative-as-given. - OHLC sanity: Spot-checked ~15 rows. Low ≤ Open/Close ≤ High holds throughout; no negative, zero, or impossibly wide prints.
Adj Close == Closeon all rows (no split/dividend adjustment visible in this window — flag for any analyst computing total return). - Zero-volume sessions:
2026-05-01,2026-05-04,2026-06-01,2026-06-03all show Volume=0 with High=Low=Open=Close. These look like Thai market holidays (Labour Day, Coronation Day, etc.) carried forward as flat synthetic prints rather than missing data — defensible but worth knowing. - Date continuity: Gap from
2026-04-10 → 2026-04-16(Songkran holiday week) is plausible. Gap2026-04-10 → 2026-04-16skips 13–15 Apr which are real Thai public holidays — clean. - 10y summary row internal consistency:
2016-05-26 → 2026-06-22, 2,450 rows, last_close=1.69, mean≈1.67. Annualised-return field reads0.9697(~0.97%/yr or 97% — ambiguous units), annualised-vol7.55, Sharpe38.26— Sharpe of 38 is nonsensical for any real asset. Either the field labels in the summary row are mis-mapped or the ratio is computed on the wrong scale. Do not quote the Sharpe, annualised return, or annualised vol numbers from the summary row as-is. Max drawdown-77.35%is plausible for a Thai small-cap finco over 10y. - Symbol
THANI/THANI.BKmatches config.
News — FAIL¶
- Yahoo (3 items): All three are generic "Asian Penny Stocks" listicles from Simply Wall St. None are about THANI specifically — THANI is at best a name-drop in a screen. No company-specific signal.
- Google ticker (40 items): Coverage is almost entirely wrong company. The query "THANI" pulled in:
- Multiple Dusit Thani (luxury hotels) stories (#4, #6, #7, #8, #11, #17, #18, #19)
- Muang Thong Thani (real estate / stadium / THAIFEX expo) stories (#1, #16)
- Udon Thani (province) stories (#9, #20)
- A Muay Thai fighter named Thani (#12)
- An Estonian footballer "Lasuai" (#13) and a skydiving article (#14) — pure noise
- Only #15 ("THANI - SET", Jan 2023) is plausibly the actual issuer, and it's just an SET RSS stub with no content.
- Net: Zero usable company-specific news. Any sentiment/headline reasoning on this set would be hallucination by association.
- google_company (2 items): A GuruFocus stock-list page and a 2024 article about SET50 additions that does not mention THANI. Useless.
Social — FAIL¶
google_blogs: 0 items.google_forums: 0 items.reddit_company: HTTP 403 error.reddit_ticker: HTTP 403 error.- X/Twitter: explicitly noted as unavailable.
- Thai-language venues (Pantip, Blockdit, Thairath) flagged as relevant but not fetched.
- Net: Zero social signal of any kind. Do not infer retail sentiment.
SET pages — FAIL (2 of 4 critical pages are 404s)¶
- profile (406 KB): Title
THANI - Company profile - The Stock Exchange of Thailand— page shell loaded for the correct entity. But body is a Nuxt JS-shell; the excerpt shown is head/meta/script preloads only. No business description, registered capital, board, or shareholder text is visible in the excerpt. Downstream personas should NOT quote profile facts from this excerpt — they aren't in the bytes shown. - financial (428 KB): Title
THANI - Company highlight. Same JS-shell pattern. No actual financial figures (revenue, net income, EPS, ROE, NPL, loan book, etc.) are present in the excerpt. Any numeric financial claim is fabrication. - shareholders (392 KB): Title:
ขออภัย ไม่พบข้อมูลที่คุณต้องการ("Sorry, the data you requested was not found") withog:url = /en/error/404. This is a 404 page. No shareholder data captured. The Thanachart-subsidiary claim in the config is not verified by this package. - filings (392 KB): Same 404 page (
/en/error/404). No filings list captured. - news (414 KB): Loaded as a JS-shell with correct title; no news rows visible in excerpt.
So of 5 SET fetches: profile/financial/news loaded as shells (no rendered data), shareholders/filings are explicit 404s.
Contradictions found¶
- News headlines vs. company identity: Google-news ticker results are dominated by Dusit Thani, Udon Thani, and Muang Thong Thani — none of which are Ratchthani Leasing PCL. Anyone treating these as "THANI news" would be analysing the wrong companies.
- Summary-row Sharpe = 38.26 vs. price path: The 60-session price range (~1.58–1.76) shows modest mean-reversion around 1.65. Sharpe of 38 is impossible; either field labels in the summary are misaligned or the metric is mis-scaled. The price path itself looks fine; the summary row is unreliable.
- Config says "Subsidiary of Thanachart Capital group": This claim is not corroborated by the SET shareholders page (404). Unverifiable from the package.
Missing / stale data¶
- Shareholders page: 404. No verified top-shareholder list, no free-float, no Thanachart parentage confirmation.
- Filings page: 404. No list of recent SET filings, no 56-1, no annual report links.
- Profile/financial/news pages: JS shells. Body content not in excerpts → no income statement, balance sheet, cash flow, ratios, dividend history, or management info captured.
- No insider-transaction data.
- No NPL / loan-loss / hire-purchase book metrics (critical for a leasing company — entire investment thesis hinges on these).
- No analyst-consensus / target-price data.
- No peer comparables (TIDLOR, MTC, SAWAD, etc.).
- No FX, policy-rate, or commercial-vehicle-sales macro context despite config flagging rate sensitivity.
- No Thai-language sentiment (Pantip / Blockdit not fetched).
- No X/Twitter coverage.
- Price data is stamped 2026-06-22 — if "today" is calendar-2025 in the user's wall-clock, every date is stale-in-reverse. Treat the package's own timestamp as the as-of date.
DO-NOT-FABRICATE list for downstream personas¶
Downstream personas (analyst, trader, CIO) MUST NOT invent or assume any of the following, because they are not present in this package:
- Revenue, net interest income, net profit, EPS — no income statement captured.
- Loan book size, hire-purchase receivables, NPL ratio, coverage ratio, credit cost, stage-3 loans — none present.
- ROE, ROA, NIM, cost-to-income, capital adequacy — none present.
- Dividend history, payout ratio, dividend yield — none present.
- P/E, P/B, P/PPOP or any valuation multiple — no per-share fundamentals captured.
- Shareholder structure / Thanachart ownership % / free float — shareholders page is a 404.
- Recent SET filings, MD&A commentary, opportunity day notes — filings page is a 404.
- Insider buying/selling, share buybacks.
- Beneish M-score, Altman Z-score, Piotroski F-score — would require financials that are not in the package.
- Any "news catalyst" or sentiment narrative drawn from the Yahoo/Google news lists — those items are about Dusit Thani, Udon Thani, hotels, and a MMA fighter, not the issuer.
- Retail / social-media sentiment — Reddit 403'd, X unavailable, blogs/forums empty.
- Annualised return, vol, Sharpe from the summary row — the Sharpe (38.26) is nonsensical; the row's field mapping is suspect.
- Splits / corporate actions —
Adj Close == Closethroughout the window; no adjustment data to verify. - Macro context (Thai policy rate, truck-sales data, BoT regulation on auto-HP) — not in package.
One-line instruction to the CIO¶
Trust the 60-day price series only; treat the 10y summary stats as unreliable, treat all "news" as wrong-company noise, treat shareholders/filings/financials as ABSENT (404s and JS shells) — any thesis requiring fundamentals, ownership, or sentiment must be deferred until real SET-rendered data is obtained.