Quantitative Research

Statistically validated trading signals. Updated daily.

✅ Validated Signals

Study Events Return p-value
H3: Retail Panic + Foreign Buy
Retail sells >1σ below mean while foreign buys
 4,209 +0.91% <0.0001
H6: Shareholder Surge + Big Holder Exit
Holders +5% in 4 weeks while big-holder % drops
 396 -0.87% —
SITC vs Foreign Divergence
SITC wins on mid-caps (local info edge)
 8,416 +0.25% —
Supply Chain Propagation (1-day lag)
TSMC buying → MediaTek follows next day
 71 edges corr 0.18 —

❌ Rejected Hypotheses

HypothesisFinding
Follow foreign consecutive buying p > 0.3, no alpha. Foreign is contrarian, following doesn't work.
Institutional front-running revenue 51% random. No insider evidence. Institutions don't know revenue ahead.
Revenue beat → drift up Reversed! Beat stocks -1.95%. Sell-the-news effect.
News volume → institutional flow Correlation r = -0.02 (zero). Institutions don't trade on headlines.

💡 Core Insight

"Foreign institutions are contrarian accumulators. They buy when retail panics and exit when retail is euphoric. Following them doesn't work — but fading retail panic does."

🔬 Advanced Findings

Finding Effect Events
H3 strongest in Optoelectronics
Retail panic in 光電 stocks bounces 5.7x more than Computer stocks
 +4.42% 268
3σ Buy anomalies have momentum
Unusual buying continues to gain over 20 days (momentum)
 +0.83% 498
3σ Sell anomalies bounce back
Unusual selling is temporary/liquidity-driven, stocks recover
 +0.64% 510
Monday always sells
Monday is always net selling regardless of Friday. If Friday also sold, Monday selling is 6x worse.
 -104M 51 Fridays
Dividend pre-positioning
Select high-yield stocks see elevated foreign buying in May-Jul
 +2-3x 30 stocks
M&A news → foreign de-risking
On M&A/capital change news days, foreign avg sells -4M (uncertainty de-risking)
 -4.0M 1,300

📊 Retail FOMO Index

Composite of ETF retail flow, shareholder growth rate, and market-wide retail net (0-100)

0-30
Fear
30-50
Normal
50-70
Elevated
70-100
Extreme

Historically, retail fear (<30) often coincides with smart money accumulation.

📚 Academic References

• Sammon & Viceira (2025) — Harvard: Retail contrarian trading contributes to PEAD

• Barber, Lee, Liu, Odean (2009) — UC Davis: Taiwan retail loses 3.8%/yr, institutions gain 1.5%

• Lin & Hung (2012) — Emerald: Foreign institutions trade contrarian in Taiwan (buy losers, sell winners)

• TEJ (2025): TDCC shareholder data constructs predictive retail speculation factor

Disclaimer: Research for informational purposes only. Not investment advice. Past performance doesn't guarantee future results.

📅 Research Log

2026-06-04
Walk-forward validation + Options P/C + Intraday 5min K
H3+Opto Sharpe 0.77 ✅ | P/C ratio 1.36 (Computex day = extreme bearish) | 1596 candles stored
2026-06-01
Cross-data analysis (futures/margin/day-trading)
Futures short = hedge not bearish | PSMC perfect trap case | Short cover + foreign buy = squeeze
2026-05-30
Out-of-sample validation + Margin data + Sharpe ratio
H3 passes OOS (+2.77%) ✅ | Margin↓+Foreign↑ = +3.13%/5d | Transaction cost analysis
2026-05-30
6 advanced studies + FOMO Index
H3 best in Opto (+4.4%) | 3σ anomaly momentum | Monday always sells | FOMO 28/100
2026-05-29
Computex verification — Jensen keynote vs foreign flow
Pre-event +700M → event day -149M | MTK 12:25 spike (lunch break orders) | 5min precision
2026-05-27
Core 16 studies + literature alignment
H3 retail panic (p<0.0001) | H6 bubble (-0.87%) | Following foreign = no alpha | Revenue sell-the-news | News irrelevant

Total: 22 studies | 5 validated signals | 4 rejected hypotheses | Coverage: institutional flow, shareholder structure, margin, futures/options, intraday, news