Research & Methodology
Original frameworks and data-driven insights from Ethan Cho (조여준) and TheVentures (더벤처스)
Research Philosophy
At VentureOracle, we believe in data over narrative and frameworks over intuition. Our research combines 20+ years of cross-market experience (US, Korea, Asia) with rigorous quantitative analysis.
🔍 Our Approach
- Primary data sources: We analyze actual market data, not just media narratives
- Reproducible findings: All research includes methodology and data sources
- Real skin in the game: Frameworks tested with $17B+ portfolio outcomes
- Korea-global lens: Unique perspective bridging Korean and global markets
Original Frameworks
📊 The Four Lenses Framework
Developed: 2018-2022 (Qualcomm Ventures, Google, Samsung, KB Investment)
Status: Proprietary methodology used in all TheVentures investments
Every investment decision should be analyzed through four distinct perspectives. The best opportunities light up all four lenses simultaneously.
1️⃣ Finance & Accounting Lens
What it reveals: Unit economics, cash flow dynamics, margin structure
Key question: "Do the numbers actually work?"
Red flags: Burning cash to buy growth, negative unit economics, "we'll make it up in volume"
2️⃣ Global Lens
What it reveals: Market arbitrage opportunities, pattern transfer from other markets
Key question: "What works elsewhere that could work here?"
Example: Toss = Venmo + Alipay for Korea (2017 insight)
3️⃣ Big Tech Lens
What it reveals: Platform dynamics, API dependencies, defensibility against tech giants
Key question: "Could Google/Meta/Amazon copy this in 6 months?"
Survival test: Regulatory moat, proprietary data, or network effects
4️⃣ Venture Capital Lens
What it reveals: Fundability, narrative strength, exit paths
Key question: "Can this raise the next round + eventual exit?"
Reality check: Markets change, but exit multiples are finite
✅ Investment Rule:
Minimum threshold: Pass at least 2 lenses
Best opportunities: Light up all 4 lenses
Example (Toss, 2017): Finance ✅ | Global ✅ | Big Tech ✅ | VC ✅ → Result: $7B+ unicorn
💸 The Optimism Tax
Research basis: Analysis of 72 million prediction market trades, $18.26B volume
Published: February 2026
Data source: Jon Becker's Kalshi dataset
Core finding: Markets systematically transfer wealth from emotional participants (takers) to patient market makers. The patterns mirror venture capital LP/GP dynamics exactly.
📊 Key Data Points
| Metric | Takers (Emotional) | Makers (Patient) | Gap |
|---|---|---|---|
| Average Return | -1.12% | +1.12% | 2.23pp |
| Finance Category | -0.08% | +0.08% | 0.17pp |
| Entertainment | -2.40% | +2.40% | 4.79pp |
| YES Bias (1¢ prices) | -41% EV | +23% EV | 64pp |
🎯 VC Application
- LPs = Takers: Chase hot sectors, deploy at peaks, overpay for consensus deals
- Top GPs = Makers: Patient capital, contrarian timing, structured downside protection
- Emotional sectors tax harder: Consumer social, climate tech, AI hype = maximum wealth transfer
- YES bias in VC: Founders pitch optimistic case; investors who structure around NO case win
Methodology: Analyzed maker/taker spreads across 5 prediction market categories, correlated emotional attachment with wealth transfer magnitude. Extended analysis to VC LP behavior patterns.
📱 The MAU Trap
Identified: February 2026 (OpenAI hardware delay analysis)
Definition: When massive user numbers (MAU) hide shallow engagement and strategic vulnerability
Core insight: Monthly Active Users (MAU) is the most commonly gamed metric in startups. High MAU + low engagement = fundable but not defensible.
📊 Case Study: AI Chatbots (2026)
| Product | MAU | Time/Day | Market Share | Verdict |
|---|---|---|---|---|
| ChatGPT | 810M | 12.4 min | 69.1% → 45.3% | MAU Trap (falling share despite user growth) |
| Claude | Lower | 34.7 min | Growing | Deep engagement (2.8x ChatGPT time) |
| Gemini | Lower | Higher | Growing | Google integration advantage |
⚠️ Red Flags (You're in the MAU Trap)
- Founders lead with MAU, not engagement depth or monetization
- User growth accelerates but revenue/engagement stagnates
- Success metrics shifted from "daily active" to "monthly" (hiding retention issues)
- Comparison to competitors uses MAU, not time spent or transaction value
- Virality without stickiness (everyone tries it once, few return)
Korea VC application: Korean startups especially prone to MAU Trap due to:
- Small addressable market (51M population) makes MAU growth hard → founders optimize for breadth over depth
- Kakaotalk integration = easy to get initial users (viral loops) but hard to build true dependency
- Fundable metrics culture (VCs ask for MAU first) incentivizes gaming the metric
TheVentures investment rule: We don't invest in MAU. We invest in DAU/MAU ratio (stickiness),time spent (engagement depth), and monetization per user (economic value). 2M users worth nothing < 20K users who pay.
🤖 AI Native VC
Implementation: "Vicky" system at TheVentures
Status: Production deployment, actively managing deal flow
Definition: AI Native VC = using AI systems not just for analysis, but for end-to-end VC operations with learning loops from actual investment outcomes.
🔄 Vicky System Architecture
1. Deal Sourcing
Monitor 1,000+ sources (news, demo days, LinkedIn, GitHub, Product Hunt, Korean VC channels)
→ Flag companies matching TheVentures thesis
2. Initial Screening
Apply Four Lenses Framework automatically to pitch decks, websites, public data
→ Score 0-10 on Finance/Global/Big Tech/VC lenses
3. Due Diligence Support
Extract unit economics, comp analysis, market sizing, technical feasibility checks
→ Generate diligence memo drafts for human review
4. Portfolio Monitoring
Track KPIs, news mentions, hiring signals, product updates across 15 active companies
→ Alert on inflection points (positive or negative)
5. Learning Loop
Feed actual outcomes (funding rounds, revenue growth, exits, failures) back into model
→ Improve pattern recognition for future deals
💡 Key Difference from Generic AI Tools
Generic pitch deck analyzer: Gives generic advice, no skin in game, no learning from outcomes
AI Native VC (Vicky): Trained on TheVentures' actual investment data, learns from $17B+ portfolio outcomes, improves with every deal
The edge: Most VCs use AI as a tool. We use AI as infrastructure. The system gets smarter with every investment.
📊 Data Sources & Reproducibility
Primary Data Sources
- Prediction Market Analysis: Jon Becker's Kalshi dataset (72M trades, $18.26B volume)
- Korea Market Data: KOSDAQ/KOSPI listings, Ministry of SMEs and Startups, Korea Venture Capital Association (KVCA)
- Global VC Data: Crunchbase, PitchBook, Preqin
- Portfolio Performance: Proprietary TheVentures data (Toss, Dunamu, 15 active investments)
- AI/Tech Trends: GitHub activity, Product Hunt launches, academic papers (arXiv)
🔬 Reproducibility Commitment
All VentureOracle research includes:
- Data sources with links
- Analysis methodology
- Sample sizes and time periods
- Limitations and assumptions
If you can't reproduce our findings, we didn't do our job. Contact: ethan@theventures.co.kr
📝 How to Cite VentureOracle Research
APA Format
MLA Format
Chicago Format
BibTeX
author = {Cho, Ethan},
title = {The Optimism Tax: What \$18B in Prediction Market Data Reveals About VC},
journal = {VentureOracle},
year = {2026},
month = {Feb},
url = {https://ventureoracle.kr/insights/the-optimism-tax}
}
🤝 Research Collaboration
Interested in collaborating on Korea VC research, accessing our datasets, or discussing our methodology?
Contact Ethan Cho (조여준):
Email: ethan@theventures.co.kr
LinkedIn: linkedin.com/in/ethan-yj-cho
We're particularly interested in: LP psychology research, Korea unicorn pattern analysis, AI-native fund operations, and cross-border VC dynamics.