QUANT REPORT — THANI¶
Data window: 2016-05-26 → 2026-06-22, 2,450 daily observations. Current price ฿1.69. Source: yfinance THANI.BK adjusted close. Macro/factor series (SET, oil, USD/THB, UST10Y) not provided in package — beta estimates below are flagged as un-computable from this data alone.
1. Return statistics (annualised, 10y daily)¶
Parsed from summary row (count=2450, mean px=1.674, last=1.69, max=7.55, min=1.43, maxDD=-77.4%, kurt≈16.2, skew≈0.20). Daily-return moments recomputed from spot levels in tail; SET comparators not in package.
| Metric | THANI | SET Index |
|---|---|---|
| Mean log return (ann.) | ≈ -8% (px fell 5.27 → 1.69 over 10y) | n/a (not in pkg) |
| Volatility (ann.) | ≈ 31% (est.) | n/a |
| Skew (daily) | +0.20 | n/a |
| Kurtosis (daily, excess) | +16.2 (very fat-tailed) | n/a |
| Sharpe (10y, ann., rf≈2%) | ≈ -0.32 | n/a |
| Max drawdown | -77.4% (peak ~฿7.55 → trough ~฿1.43) | n/a |
Kurtosis of 16 means tails are ~5× normal. GBM-implied bands below will systematically understate true tail risk.
2. Volatility regimes¶
Identified by structural break in price level + rolling 60d vol (qualitative from full series):
| Regime | Dates (approx) | Price range | Character |
|---|---|---|---|
| R1 — Bull/rerating | 2016 → 2018 H1 | ฿3 → ฿7.5+ | Trending, low vol |
| R2 — Top/distribution | 2018 H2 → 2019 | ฿6 → ฿4 | Elevated vol |
| R3 — COVID crash | 2020 Q1–Q2 | ฿4 → ฿2 | Vol spike, gapping |
| R4 — Range/grind down | 2021–2023 | ฿3–฿4.5 | Mid vol |
| R5 — Capitulation | 2024–2025 | ฿4 → ฿1.4 | Sustained decline |
| R6 — Basing (current) | 2025 H2 → now | ฿1.5–฿1.75 | Low realised vol (~14% ann. from last 60d), tight range |
Current 60-day realised vol = ~14% ann. vs 10y avg ~31%. We are in a low-vol basing regime, not characteristic of long-run distribution.
3. Beta & factor exposures¶
Cannot compute from package — SET Index, sector ETF, oil, USD/THB, UST10Y series were not supplied. Reporting prior expectations for a Thai hire-purchase lender, to be validated:
| Factor | Expected sign | Rationale | Computed? |
|---|---|---|---|
| β to SET | +0.7 to +1.0 | Mid-cap financial | NO |
| β to BoT policy rate (Δ) | negative | NIM compression on hikes, but funding cost lag | NO |
| β to USD/THB | mildly negative | THB-only book | NO |
| β to diesel/oil | slightly negative | Trucking customer cashflow | NO |
| β to China property/freight | negative | Cross-border trucking demand | NO |
Style ranks vs SET universe — financials data not in package. Last 5y price action (-65%) is consistent with deep value, negative momentum (12-1), low quality (rising NPLs in HP sector industry-wide).
4. Mean reversion vs trend¶
Tests not directly computed; inferred from price path.
- Hurst exponent (est.): ~0.55–0.60 on log-price (trending), but 0.40–0.45 on returns at 1–20d horizon → short-horizon mean-reverting, long-horizon trending down.
- Variance ratio (informal): VR(20)/VR(1) appears < 1 at short horizons (mean-reverting day-to-day, as evidenced by tight 60d range) but > 1 at multi-year (persistent drift).
- Implication: at 2y horizon, trend dominates — historical drift since 2018 is -16% CAGR. Recent 12m drift is flat (~0%) suggesting a regime change may be in progress, but statistically insufficient evidence (one year of basing vs 6y of decline).
5. Drawdown profile¶
| Stat | Value |
|---|---|
| Max DD | -77.4% (from ฿7.55 peak ~2018 to ฿1.43 trough ~2025) |
| Time to trough | ~7 years |
| Recovery to peak | not recovered; current price 22% of peak |
| Current DD from 10y high | -77% |
| Current DD from 5y high | ~-60% |
| Mean DD (rough) | -45% |
Drawdown distribution heavily skewed toward deep, slow drawdowns — atypical of a mean-reverting financial; behaves like a structurally impaired name.
6. 2-year forecast — model comparison¶
Anchor: spot ฿1.69. Inputs: μ_10y ≈ -8%/yr, μ_2y ≈ 0%/yr, σ_10y ≈ 31%, σ_60d ≈ 14%. Fundamental P/E, EPS, BVPS not provided in package → Model D uses placeholder noted as un-anchored.
| Model | Assumption | 12m point | 12m 90% CI | 24m point | 24m 90% CI |
|---|---|---|---|---|---|
| GBM (10y drift/vol) | μ=-8%, σ=31% | ฿1.56 | [฿0.94, ฿2.60] | ฿1.44 | [฿0.73, ฿2.85] |
| GBM (recent regime) | μ=0%, σ=20% | ฿1.69 | [฿1.21, ฿2.36] | ฿1.69 | [฿1.04, ฿2.74] |
| Mean-rev to 5y avg (~฿2.50) | half-life ~18m | ฿2.00 | [฿1.30, ฿3.05] | ฿2.30 | [฿1.40, ฿3.70] |
| Multi-factor regression | cannot run — no factor series | — | — | — | — |
| Naive EPS×P/E | cannot run — no EPS/PE in pkg | — | — | — | — |
| Median of available | ฿1.75 | ฿1.81 |
CIs from GBM assume log-normal; understated given kurtosis 16. Real left-tail at 24m likely extends to ฿0.50–0.60 (consistent with continuation of -16% post-2018 drift).
7. Where models disagree and why¶
- GBM-10y vs GBM-recent (₿1.44 vs ₿1.69 at 24m): the 10y series embeds a one-way de-rating; recent 12m is flat. If basing is real, drift is reset → +18% gap. This is the key parameter risk.
- Mean-reversion vs GBM: MR model anchors to 5y avg ~฿2.50 — assumes the decline was disequilibrium. If the decline reflected permanent NPL/credit-cost re-rating, the anchor is wrong. No statistical test in package can distinguish.
- Naive EPS×P/E cannot be computed — fundamental time series absent. This is the biggest gap.
8. Confidence: LOW¶
Limits of this analysis: - No factor series (SET, rates, FX, oil) → no β, no factor decomposition, no multi-factor projection. - No EPS / book value / NIM / NPL time series → no fundamental anchor; mean-reversion target is purely statistical (5y avg price), which is circular for a name in secular decline. - Kurtosis 16 invalidates Gaussian CIs; the ฿0.94 lower bound at 12m GBM-10y is optimistic. - Regime ambiguity: 12 months of basing is insufficient sample (~250 obs) to reject the prior 6y downtrend at any reasonable significance. - News flow in package is noise (Dusit Thani hotel, Muang Thong Thani place names — wrong entity). Zero signal on company fundamentals from news scrape.
Bottom line numbers: 24m median ≈ ฿1.80, 90% range [฿0.75, ฿3.70]. Range is ~5×; do not treat point estimate as informative.