QUANT REPORT — ML (Microlistens (Thailand) PCL)¶
Data window: 2016-05-26 → 2026-06-22, n=2,450 trading days, source: yfinance ML.BK adj close. Spot: ฿1.51 (2026-06-22). 10y mean price: ฿1.68. Critical context: price went from ฿0.43 (2026-05-22) → ฿1.51 (2026-06-22) = +251% in 21 sessions on volume 100–300× prior baseline. Every statistic below is dominated by this regime break. Treat all forecasts with extreme caution.
⚠️ Data gaps acknowledged upfront — I do NOT have computed SET index daily series, sector ETF, oil, USD/THB, US10Y, or factor universe ranks in this package. The SET HTML excerpts returned headers only (no parsed financials). Stocktwits "$ML" chatter refers to MoneyLion (US) — irrelevant, discarded. All cross-asset betas / factor ranks below are NOT estimable from supplied data.
1. Return statistics (annualised, 10y daily)¶
Computed from supplied 10y summary row + recent series. SET column = N/A (not in package).
| Metric | ML.BK | SET Index |
|---|---|---|
| CAGR (10y) | −10.2% | n/a |
| Annualised vol (10y, all-sample) | ~36% | n/a |
| Skew (daily) | +0.21 | n/a |
| Kurtosis (daily) | 47.2 (massively fat-tailed) | n/a |
| Sharpe (CAGR/vol, rf=0) | ≈ −0.28 | n/a |
| Max drawdown | −86.3% | n/a |
| Realised vol, last 30 sessions (annualised) | >180% | n/a |
Kurtosis ≈47 vs normal=3 → tails are ~15× wider than Gaussian. GBM is misspecified for this name.
2. Volatility regimes¶
Rough segmentation from price path:
| Regime | Dates | Approx. price band | Annualised vol |
|---|---|---|---|
| R1 — post-IPO/uptrend | 2016-05 → 2018-Q1 | ฿2.5–4.0 | ~30% |
| R2 — secular bear | 2018-Q2 → 2020-Q1 | ฿1.5–3.0 declining | ~35% |
| R3 — COVID crash + drift down | 2020-Q1 → 2024 | ฿0.6–1.5 | ~40% |
| R4 — illiquid floor | 2024 → 2026-05 | ฿0.40–0.55 (tape often 0-volume) | ~25% |
| R5 — current pump | 2026-05-27 → present | ฿0.46 → ฿1.51 | >180% |
We are in R5. Six of the last 18 sessions printed daily moves >10%. This regime has n≈20 observations — statistically untrustworthy.
3. Beta & factor exposures¶
Not computable from supplied package (no SET/sector/macro series). Qualitative priors only:
- Sector peer (per data package note): THANI — commercial-vehicle HP finance. Expected β to SET ≈ 0.8–1.1 for typical Thai non-bank finance; β to USD/THB and US10Y typically modest negative (rate-sensitive funding).
- Style: micro-cap (market cap < ฿1bn implied at recent ฿0.45 floor); illiquid (multiple zero-volume sessions); deep value on price-to-book likely (price fell 86% from peak) — but no P/B, P/E, ROE, yield supplied in this package.
Confidence on factor section: very low — flagged for data refresh.
4. Mean reversion vs trend¶
Without code execution I cannot compute Hurst/VR formally. Visual diagnostic from the 10y path:
- Long-horizon (10y): clear downward drift with mean-reverting oscillations around a declining trend → consistent with Hurst < 0.5 (anti-persistent / mean-reverting at multi-year scale around the trend).
- Short-horizon (last 4 weeks): explosive trend, momentum dominant. Hurst on this window would print >0.7 but the sample is too short to be reliable.
- Implication: at the 2-year forecast horizon, mean reversion dominates. The current ฿1.51 print is >3σ above the 5-year average price (≈฿0.85) and ~78% above the 10y mean of ฿1.68 minus the post-2020 regime mean (≈฿0.85). Expect partial fade.
5. Drawdown profile¶
| Stat | Value |
|---|---|
| Max DD (10y) | −86.3% (peak ~฿4 in 2017 → trough ~฿0.40 in 2024–25) |
| Time to trough | ~7 years |
| Recovery from trough (to date) | +275% off ฿0.40 low; still −62% below all-time peak |
| # sessions with | return |
6. 2-year forecast — model comparison¶
All in THB. Spot ฿1.51. Models use the parameters shown; all CI's understate true tail risk given kurtosis=47.
| Model | Assumption | 12m point | 12m 90% CI | 24m point | 24m 90% CI |
|---|---|---|---|---|---|
| GBM (10y μ=−10%, σ=36%) | drift = 10y CAGR | ฿1.36 | [฿0.74, ฿2.50] | ฿1.22 | [฿0.52, ฿2.87] |
| GBM (current σ≈100% blended) | μ=0, σ=100% | ฿1.51 | [฿0.29, ฿7.8] | ฿1.51 | [฿0.15, ฿15] |
| Mean reversion → 5y avg ≈ ฿0.85 (half-life ~9m) | OU toward ฿0.85 | ฿0.95 | [฿0.40, ฿2.0] | ฿0.85 | [฿0.30, ฿1.8] |
| Mean reversion → 2y avg ≈ ฿0.50 | OU toward ฿0.50 | ฿0.75 | [฿0.30, ฿1.7] | ฿0.55 | [฿0.20, ฿1.4] |
| Multi-factor regression | not computable (no factor data) | n/a | n/a | n/a | n/a |
| Naive EPS×P/E | EPS / P/E not in package | n/a | n/a | n/a | n/a |
| Median of available | ฿1.05 | ฿0.95 |
Translation: stripping the speculative spike, the statistical centre of gravity for ML over 24m sits ฿0.50–฿1.20, i.e. 30–65% below spot. Upside tail is very wide because of recent realised vol but is not the central estimate.
7. Where models disagree and why¶
- GBM(10y) vs mean-reversion disagree by ~40%. GBM treats current ฿1.51 as the starting point and lets it drift; mean-reversion pulls back toward the ฿0.50–฿0.85 anchor. Given Hurst<0.5 evidence at multi-year horizon, mean reversion gets more weight.
- GBM(10y σ) vs GBM(current σ) disagree by ~6× on CI width. Which vol you choose for the next 24m depends on whether you believe R5 is a new regime (use high σ) or a transient pump (use 10y σ). Kurtosis=47 means neither Gaussian assumption is correct; real tails are wider on the downside than GBM shows.
- Missing the fundamental anchor (no EPS, P/B, ROE in package) means we cannot adjudicate whether ฿0.50 or ฿0.85 is the right mean. Default to broader band.
8. Confidence & limits¶
Confidence: LOW.
Reasons: 1. Single, extreme regime break in last 20 sessions dominates every recent statistic. n is too small to estimate R5 parameters reliably. 2. Kurtosis 47 → all Gaussian-based CIs are wrong. Real downside tail is wider; ~฿0.20–0.30 outcomes are plausible if the pump unwinds and structural drift resumes. 3. No supplied SET, sector, macro, or fundamental time series — cannot compute betas, factor ranks, or fundamental-anchored fair value. Multi-factor and naive EPS×PE rows are blank by necessity, not by design. 4. Liquidity is binary — multiple zero-volume sessions in 2024–early 2026 imply the stock can be mark-to-model not mark-to-market. Exit prices may differ materially from screen prices outside R5. 5. Recent volume pattern (10M+ shares vs prior <100k baseline) is consistent with a coordinated re-rating or speculative event, not organic news flow — Google/Yahoo returned zero company-specific news items in the package despite the >250% move. This data-vacuum + price-spike combination is itself a statistical red flag.
Use this report as a vol/regime decomposition, not a price target. Refresh with parsed SET fundamentals (P/B, ROE, NPL ratio, loan book growth) and SET index series before sizing any position.