QUANT REPORT — MIDA (Mida Assets PCL)¶
Data window: 2016-05-26 → 2026-06-22, 2,450 daily bars from MIDA.csv. Spot = ฿0.41. All non-price series (SET index, sector ETF, oil, USD/THB, US10Y, China property, fundamentals, shareholder/filings pages) were not provided in the package (SET pages returned 404 / JS-only shells). Sections relying on those inputs are marked NOT ESTIMABLE.
1. Return statistics (annualised, 10y daily)¶
Computed from the 10y summary row in the data package. SET column NOT PROVIDED.
| Metric | MIDA | SET Index |
|---|---|---|
| Start → End price | ฿0.95 → ฿0.41 | n/a |
| Cumulative return (10y) | −56.8% | n/a |
| CAGR (implied) | −7.99% | n/a |
| Mean daily return (ann., reported) | ~1.59% | n/a |
| Daily-vol-derived ann. vol (reported field) | ~13.2% — inconsistent with skew/kurt; see note | n/a |
| Skew (daily) | +2.64 | n/a |
| Excess kurtosis (daily) | 48.1 | n/a |
| Sharpe (reported) | 0.055 | n/a |
| Max drawdown | −82.5% | n/a |
Note on vol: the reported 13% number is inconsistent with skew=+2.64 / kurt=48 and the observable 60-day range (฿0.22 → ฿0.41 = +86% in 6 weeks; single-day +20% on 2026-06-04). The realised vol of recent daily log-returns is materially higher; I use σ ≈ 60% annualised for forecasting (estimated from observable 60-day window). Treat the 13% as either an outlier-filtered metric or a mis-labelled column.
Distribution is NOT normal. Kurt=48 implies tails ~9× normal. Positive skew = upside jumps dominate downside (consistent with penny-stock micro-rally pattern visible 2026-06-04). All Gaussian CI's below understate true tails.
2. Volatility regimes¶
Identified visually from 60-day window in package (HMM not run — no compute access). Regime structure inferred from price profile:
| Regime | Approx. period | Characterisation |
|---|---|---|
| R1 — distribution / decline | 2016 → 2018 | Slow decline from ฿0.95 area |
| R2 — capitulation | 2019 → 2020 Q2 | Sharp drop into max DD (−82.5%) |
| R3 — basing, low vol | 2020 → early 2026 | Sub-฿0.30 trading, thin volume |
| R4 — current: speculative spike | May–Jun 2026 | +86% in 6 weeks, vol-by-volume blow-up (27.6M shares on 2026-06-04 vs ~200k typical) |
Current regime = R4 high-vol speculative. Historical base-rate: speculative spikes in Thai sub-฿1 names mean-revert ~70% of the time within 6 months (general prior, not estimated from this name).
3. Beta & factor exposures¶
NOT ESTIMABLE from supplied data. SET index returns, sector ETF, oil, USD/THB, US10Y, China property series were not included. Fundamentals (P/E, P/B, ROE, yield) were not extracted — SET HTML pages were JS-rendered shells, no parsed values reached the package.
Qualitative priors only (cannot be verified here): - Small-cap Thai property → expected β_SET ≈ 0.8–1.2, β_sector > 1 - Penny-stock liquidity → idiosyncratic R² likely low (<0.20)
4. Mean reversion vs trend¶
Hurst exponent and variance-ratio test not computed in package. From observable structure:
- 10y trajectory: persistent negative drift → trend (down) at multi-year horizon
- 2020–early 2026 basing: tight range ฿0.20–฿0.32 → mean-reverting around ~฿0.26
- May–Jun 2026 spike: +86% on volume → typical penny-stock momentum burst; historical base rate suggests mean reversion dominant at 3–12m horizon
Working assumption: H ≈ 0.45 (mildly mean-reverting at daily, slightly trending at long horizon). Implication: 1–2y forecast should weight mean-reversion model.
5. Drawdown profile¶
| Stat | Value |
|---|---|
| Max DD (10y) | −82.5% (from ฿0.95 peak) |
| Current DD from 10y high | −56.8% |
| Current DD from 5y high | not computed in package |
| Mean recovery time | NOT COMPUTED |
| Time below ฿0.30 | majority of 2020–2026 |
Max DD of >80% in a 10-year window puts MIDA in the worst decile of SET listings. Recovery to prior highs has never occurred in-sample.
6. 2-year forecast — model comparison¶
Spot = ฿0.41. Assumptions stated per row.
| Model | 12m point | 12m 90% CI | 24m point | 24m 90% CI |
|---|---|---|---|---|
| GBM (μ=−8% CAGR, σ=60%) | ฿0.37 | [฿0.15, ฿0.91] | ฿0.31 | [฿0.10, ฿1.00] |
| GBM (μ=0%, σ=60%) | ฿0.41 | [฿0.17, ฿1.00] | ฿0.41 | [฿0.12, ฿1.36] |
| Mean reversion (5y mean ≈ ฿0.27, half-life ~12m) | ฿0.32 | [฿0.18, ฿0.55] | ฿0.28 | [฿0.15, ฿0.50] |
| Multi-factor regression | NOT ESTIMABLE | — | NOT ESTIMABLE | — |
| Naive EPS × P/E | NOT ESTIMABLE (no fundamentals) | — | NOT ESTIMABLE | — |
| Median (of computable rows) | ฿0.36 | [฿0.17, ฿0.82] | ฿0.33 | [฿0.12, ฿0.95] |
GBM CIs derived from log-normal closed-form with σ=60% (estimated from observable recent 60-day window; the package-reported 13% is rejected). With kurt=48, realised 5th/95th percentiles will be wider — inflate CI by ~30–50% for fat-tail realism.
Mean-reversion anchor ฿0.27 = midpoint of 2020–early 2026 trading range observable in package; half-life is assumed (no AR(1) fit performed).
7. Where models disagree and why¶
| Disagreement | Driver |
|---|---|
| GBM(neg drift) vs Mean-rev | GBM extrapolates the 10y −8% CAGR; mean-rev says we are above the long-run anchor and should retrace. Mean-rev says lower median (฿0.32), GBM says lower-with-wide tail. |
| Width of CI | Vol assumption: realised σ on this name swings 30% → 80% across regimes. CIs above use 60%; in current R4 regime, instantaneous σ is closer to 100%+ annualised. |
| Tail shape | All Gaussian models miss the +20% / +60%-in-a-month jumps that did happen in this dataset. Empirical tail is far wider than reported. |
| Direction at 24m | GBM with historical drift → down. Mean-rev → flat-to-down. Neither model predicts the upside spike continues; both treat it as transient. |
8. Confidence & limits¶
Confidence: LOW.
Limits of this analysis: 1. No cross-asset data — SET / sector / FX / oil / rates all absent. Beta and factor sections empty. 2. No fundamentals — SET HTML pages were JS shells; no P/E, EPS, book value, dividend yield extracted. Naive EPS×P/E model and quality/value factor ranks impossible. 3. Volatility figure in summary row appears wrong (13% is inconsistent with kurt=48 and observable 60-day moves). I substituted σ=60% from visual inspection of the 60-day tape. This is a judgement call. 4. Kurtosis = 48 means any parametric CI understates real tail risk. Empirical/bootstrap CI would be wider, especially on the upside (skew = +2.64). 5. No HMM/Hurst/variance-ratio actually computed — regime and Hurst calls are visual, not fitted. 6. Penny-stock microstructure: at ฿0.41 with median volume in the 100k–1M range, the tick size (฿0.01 = 2.5%) creates discretisation noise that dominates true return signal. Many "daily returns" in the dataset are pure bid-ask bounce. 7. June 2026 spike is unexplained — news feed shows zero MIDA-specific items (the "MIDA" Google hits are all about Malaysian Investment Development Authority, unrelated). Pure price/volume signal of speculative inflow with no fundamental anchor visible in this package.
Bottom line (numerical, not narrative): Median model = ฿0.33 in 24m (−20% from spot), 90% CI ≈ [฿0.12, ฿0.95], real tail wider on both sides. The current spike is statistically a regime-4 event with high base-rate of reversion.