Autonomous Investing
portfolio deal
← All Deals portfolio
Autonomous Investing - Deal Memo
Date Reviewed: January 8, 2026 Stage: Pre-Seed (a16z Speedrun sr005) Source: Direct / LinkedIn Investment: $10,000 via SAFE
Summary
One-liner: AI trading technology company building autonomous trading systems. Founder 10x'd his personal portfolio (2021-2025). Investment is equity in the tech company, not LP in a fund.
Founder: Heng xin Fun
- Part of a16z Speedrun sr005
- 5.5 years HFT/Quant experience at Chi-Courser Investment Management (Shanghai)
- 4.5 years at Amazon (Applied Scientist II, Web Scale Search)
- Senior ML Engineer at ByteDance (LLMs/Agents for ad targeting)
- Master's in CS/AI from USI (Magna Cum Laude), BA Economics/Statistics from UW
Round:
- Raised: $3.5M
- Investors: 3 VCs + 2 angels
- My check: $10K (SAFE, Post-money Valuation Cap)
Key Metrics:
- Personal track record: 10x portfolio (stocks + options) 2021-2025
- GitHub: Active development on earnings prediction models (AMD, PLTR, META, MSFT)
Investment Thesis
1000x Opportunity?
YES
- TAM: Global algorithmic/quant trading market is massive ($15B+ and growing); hedge fund AUM is $4T+
- Contrarian insight: Most retail traders and even many funds underperform the market; an AI system that can consistently generate alpha is extraordinarily valuable
- Timing: LLM capabilities + real-time data processing are finally mature enough for sophisticated autonomous trading
- Founder edge: Heng has the rare combination of:
- Deep quant/HFT experience (5.5 years building trading strategies)
- Modern ML/AI expertise (Amazon search, ByteDance LLMs)
- Proven personal track record (10x in 4 years) - this is the proof of concept
- Currently building at a16z Speedrun (strong institutional validation)
- Technical moat: LLM-based signals + proprietary data sources + multi-factor ML models
Kingmaker Fit?
YES - Strong fit
- Technical scaling: Strong alignment - can advise on ML infrastructure, AWS optimization
- Network: Can connect to technical talent, other founders in AI/fintech space
- GTM guidance: Prop fund model is capital-efficient; can advise on fund structure and investor relations
- Long-term optionality: If strategy proves out, could evolve into B2B API or consumer product
Green Flags
- Rare founder profile - Combines quant trading, big tech ML, and AI research
- Skin in the game - 10x'd his own money, not just theory
- a16z Speedrun sr005 - Strong institutional validation
- Active builder - GitHub shows recent work on earnings prediction models
- Technical depth - HFT strategies, LLM agents, web-scale search experience
- Strong education - Master's in AI (Magna Cum Laude) + Economics/Stats background
- VISION TEST: STRONG PASS - Clear obsession, would build this regardless of funding
- Strong round - $3.5M from quality investors validates the opportunity
- Capital-efficient model - Prop trading doesn't require massive sales/marketing spend
Red Flags
- Solo founder - No co-founder yet, but will hire as fund scales (acceptable for prop trading)
- Regulatory risk - Trading/fund management is regulated, but manageable at this stage
- Competition - Well-funded quant funds exist, but Heng's edge is modern LLM-based approaches they haven't adopted
Stealth mode- Resolved: Now have clarity on product/strategyReplicability question- Resolved: Prop fund model means he IS the product
Key Questions (Answered)
| Question | Answer |
|---|---|
| Product vision? | AI trading tech company - builds and owns the AI trading technology; may also manage capital |
| Technical moat? | LLM-based signals + proprietary data + multi-factor ML models |
| How did he 10x? | Combination of options strategies + ML-driven stock selection |
| Co-founder/team? | Solo for now, will hire quants/engineers as fund scales |
| Fundraising? | $3.5M closed from 3 VCs + 2 angels |
| Regulatory? | Starting as prop trading; will formalize fund structure as needed |
| Product state? | Active trading with his own capital; GitHub shows ongoing development |
Decision
INVESTED - $10K via SAFE
Rationale
This is a 1000x opportunity with strong kingmaker fit:
-
Founder-market fit is exceptional - Heng has the rare combination of quant trading experience (5.5 years HFT), modern ML expertise (Amazon, ByteDance), and proven results (10x personal portfolio)
-
Vision test: STRONG PASS - After multiple calls, clear that Heng has the obsession and would build this regardless of funding
-
Technical moat is real - LLM-based signals + proprietary data + multi-factor ML is a differentiated approach that legacy quant funds haven't adopted
-
Capital-efficient model - Tech company owns the IP; doesn't require massive burn on sales/marketing
-
Strong investor signal - $3.5M from 3 VCs + 2 angels validates the opportunity beyond just my conviction
-
Kingmaker fit confirmed - Strong potential to add value through technical network, ML infrastructure advice, and fund strategy guidance
Risk Acknowledgment
- Solo founder risk is real but acceptable (he IS the core technical asset)
- Regulatory complexity will increase as fund scales
- Market conditions can change; past performance doesn't guarantee future results
Conviction level: HIGH
Appendix
Contacts
- Name: Heng xin Fun
- Email: hengfun@gmail.com
- LinkedIn: https://www.linkedin.com/in/hengxinfun/
- GitHub: https://github.com/hengfun
Experience Timeline
| Period | Role | Company | Notes |
|---|---|---|---|
| Aug 2025 - Present | Founder | Autonomous Investing | a16z Speedrun sr005 |
| Jan 2025 - Jul 2025 | Senior ML Engineer | ByteDance | LLMs/Agents for monetization |
| Jan 2020 - Aug 2024 | Applied Scientist II | Amazon | Web Scale Search, Retrieval & Ranking |
| Jun 2014 - Dec 2019 | Quantitative Researcher | Chi-Courser Investment Mgmt | HFT strategies |
| Dec 2018 - Sep 2019 | Research Assistant | University of Bern | Blockchain/voting research |
Education
- USI Università della Svizzera italiana - Master's in CS/AI (2017-2019), Magna Cum Laude
- University of Washington - BA Economics, Statistics (2007-2012)
Relevant GitHub Repos
amd-q3-2025-earnings- AMD earnings analysispltr-q3-earnings-2025- Palantir earnings analysisMETA_GOOGLE_MSFT_EARNINGS- Pre-earnings prediction modelscrypto-crawler- Crypto website data collection
Documents
attachments/safe-postmoney-valuation-cap.pdf- Unsigned SAFE agreement
Contacts — Autonomous Investing
Founders
| Name | Role | Phone | ||
|---|---|---|---|---|
| — | — | — | — | — |
Relationship Timeline
No interactions logged yet.
Last Touchpoint
- Date: —
- Status: Invested
- Next Action: —
Notes - Autonomous Investing
Date: January 8, 2026 Type: Founder Calls (Multiple) Summary:
- Completed multiple calls with Heng
- Product: AI trading technology company (equity investment in tech co, NOT LP in a fund)
- The company develops and owns the AI trading systems; may also manage capital
- Round closed: $3.5M from 3 VCs + 2 angels (including $10K from me)
- Technical moat: LLM-based signals + proprietary data + multi-factor ML models
- Team: Solo founder for now, will hire as company scales
Impressions:
- VISION TEST: STRONG PASS - Clear obsession, would build this regardless of funding
- Heng's own words: "distilling himself into the AI" - encoding his 5+ years of quant intuition and trading expertise into an autonomous system
- Heng has the rare combination of quant intuition and modern AI capabilities
- His personal 10x track record is the ultimate proof of concept
- a16z Speedrun validation adds institutional credibility
- Kingmaker fit confirmed: strong potential to add value beyond capital
Decision: INVEST $10K
Date: January 8, 2026 Type: Document Added Summary:
- Added unsigned SAFE (Post-money, Valuation Cap Only) to deal folder
- Ready for execution pending final diligence
Date: January 8, 2026 Type: Initial Research Summary:
- Sourced deal via LinkedIn profile
- Founder Heng xin Fun is building an AI trading platform
- Part of a16z Speedrun sr005 cohort
- Strong founder-market fit: 5.5 years HFT quant + Amazon/ByteDance ML + 10x personal returns
Impressions:
- Rare combination of quant trading and modern ML expertise
- Active on GitHub with earnings prediction projects - shows he's building, not just talking
- a16z Speedrun validation is a strong signal
- Personal 10x track record (2021-2025) suggests real trading skill
Follow-ups:
- Reach out to schedule founder call
- Ask about product vision (B2C vs B2B)
- Understand team composition plans
- Get details on fundraising round