P2A-C5 · Behavioral Finance (6 Biases)¶
Core Takeaway
Your biggest opponent isn't the market — it's yourself. Recognizing 6 biases = dismantling 90% of retail investor traps.
Universal Investment Model — Applicable to Any Industry
P2A-C5 (Part 2.A, final chapter). After this chapter, you'll be able to identify 6 major behavioral biases in investment decisions and hack them with explicit tools.
1. The Problem: Your Thesis Is Perfect, Yet You Still Fall into Traps When Deciding¶
In P2A-C1 through C4, you learned valuation / mental models / historical comparisons / sizing — the tools are all there.
But in practice: - You panic-sell during the DeepSeek selloff (1/27) - You FOMO-buy NVDA after it's up 50% (1/15) - Your starter position is down 20%, and you're anchored, refusing to cut losses (sunk cost) - You follow Druckenmiller's public trim and sell alongside him (authority)
Tools ≠ Execution. Behavioral biases render tools ineffective.
Daniel Kahneman's 2002 Nobel Prize work: The human brain has 25+ systematic biases. In investing, the 6 most lethal are:
| # | Bias | Manifestation in Investing |
|---|---|---|
| 1 | Confirmation bias | You only seek evidence that supports your thesis |
| 2 | Anchoring | You anchor to your entry price and refuse to cut losses when it drops |
| 3 | Loss aversion | You avoid losses (pain of losing $100 vs. pleasure of gaining $100) by 2x |
| 4 | Sunk cost fallacy | You've already lost, yet you still hold waiting to break even |
| 5 | Herd behavior / FOMO | You buy because others buy, sell because others sell |
| 6 | Recency bias | Recent events are overweighted (DeepSeek selloff = AI is doomed?) |
2. The Solution: 6 Biases + 6 Explicit Hacks¶
| Bias | Hack |
|---|---|
| Confirmation bias | Mandatory anti-thesis (P3-C5) — write 5 counterarguments for every long position |
| Anchoring | Ignore entry price — look at fair value range + invalidation_triggers |
| Loss aversion | Explicit stop loss — write it in thesis yaml, execute without hesitation when triggered |
| Sunk cost | Monthly reassessment — If you didn't have a position today, would you buy? No → exit |
| Herd behavior | Decision cooldown — wait 24 hours after seeing a recommendation before acting |
| Recency bias | Base rate (P2A-C3) — calibrate using historical paradigms |
Write these 6 hacks into your process — don't rely on willpower.
3. How It Works: Detailed Explanation of 6 Biases + Real Examples¶
3.1 Confirmation Bias¶
Manifestation: After going bull on NVDA, you only read bullish NVDA reports and ignore bearish ones.
Real Example: In 2024 hyperscaler capex reports, bulls read "MSFT capex +45%", bears read "MSFT FCF sharp turn". You only read one side.
Hack: - For every long thesis, force yourself to write an anti-thesis (P3-C5) - Subscribe to 1 opposing source (e.g., Jim Chanos Twitter) - Find 1 strong short report each month
3.2 Anchoring¶
Manifestation: You bought NVDA at $130, it drops to $115, and you don't sell ("waiting to get back to $130 to sell"). $130 has nothing to do with fair value.
Real Example: Before the DeepSeek selloff on 2025/01/27, NVDA was $146. After the selloff, it was $121. Holders who bought at $146 psychologically "wait to get back to $146". But fair value (Damodaran reverse DCF) might be $110-130.
Hack: - In thesis yaml, write the fair value range, not the entry price - When reviewing, only look at "price vs. fair value", not "price vs. entry" - Write an explicit stop loss (e.g., if it breaks below $100, trigger a review)
3.3 Loss Aversion¶
Manifestation: The psychological pain of losing $100 ≈ the pleasure of gaining $200. This makes you overly avoid losses → you prematurely trim winners and hold onto losers.
Real Example: You buy 5 AI stocks. 3 are up 20%, and you trim all of them (lock in profits). 2 are down 20%, and you hold all of them (waiting for a comeback). 12 months later, the 3 you trimmed continue up 100%, and the 2 you held continue down 40%. Net result: big loss.
Hack: - Explicit thesis review — if losers meet invalidation_triggers, exit immediately - Don't trim winners unless valuation is extreme (>5-year high + growth slowing) - Write "ride winners, cut losers" as your mantra
3.4 Sunk Cost Fallacy¶
Manifestation: Your NVDA position is down 30%. You think, "I've already lost, I can't sell and realize the loss." In reality, you're holding because of sunk cost.
Real Example: In 2025, suppose your META position is down 40% (hypothetical). The real question isn't "how much have I lost," but "If I didn't have a position today, would I buy at the current price?" If no → you should exit.
Hack: - Monthly thesis review: "If I didn't have a position today, would I buy at the current price?" Answer No → exit - When a thesis fails (invalidation_triggers are hit), exit without hesitation, ignoring sunk cost
3.5 Herd Behavior / FOMO¶
Manifestation: You see NVDA up 50% and FOMO-buy. You see a recommendation on X and buy. You see Druckenmiller trim and sell alongside him.
Real Example: - In 2023, NVDA at $100 — retail investors thought it was too expensive - In 2024, NVDA at $130 (up 30%) — retail investors FOMO-bought - In 2025/01, NVDA at $146 — retail investors added - On 2025/01/27, -17% — panic sell - → Buy high, sell low
Hack: - 24-hour cooldown — wait 24 hours after seeing a recommendation or news before deciding - Only make trades on weekends (avoid intraday impulses) - Write a paragraph on "Why am I buying now?" — if it's just "because everyone else is buying," don't buy
3.6 Recency Bias¶
Manifestation: Recent events are overweighted. After the DeepSeek selloff, you think "AI is doomed." After NVDA hits new highs, you think "it will go up forever."
Real Example: - On 2025/01/27, -17% — retail investors think "AI bubble burst" - In February, after recovery — retail investors think "AI is unstoppable" - In April, at new highs — retail investors FOMO-buy (at the top) - → Every time, recency bias leads to mistakes
Hack: - Base rate thinking (P2A-C3) — calibrate using historical paradigms - Monthly thesis review, not daily - After a big price move, take 1 week of no action (cooldown)
4. vs. Retail Investor Behavior¶
| Dimension | Retail Investor (Falling for Biases) | What You Can Change |
|---|---|---|
| Buy immediately after seeing a recommendation | 24-hour cooldown | ✓ Psychological discipline |
| Anchor to entry price | Look at fair value, not entry | ✓ Add fair value to thesis yaml |
| Hold due to sunk cost | Monthly "Would I buy today?" test | ✓ Monthly review |
| FOMO buy high | Wait 1 week before deciding | ✓ Weekend decisions |
| Panic sell low | Invalidation_triggers + wait 1 week | ✓ Write explicitly |
| React to recency | Calibrate with base rate (history) | ✓ P2A-C3 tool |
5. Try It: Run a 6-Biases Audit on Your Portfolio¶
Task (~30 minutes): Look at your current holdings and answer:
| Bias | Manifestation in Your Portfolio | Hack Action |
|---|---|---|
| Confirmation | When was the last time you actively sought an anti-thesis? | Find 1 opposing source this week |
| Anchoring | Can you state the fair value of each stock without mentioning entry price? | Add fair value to thesis yaml |
| Loss aversion | What's your ratio of trimming winners vs. holding losers? | Reassess — should losers be exited? |
| Sunk cost | Test each stock: "If I didn't have a position today, would I buy?" | Exit those with a No answer |
| Herd / FOMO | Was your last purchase a calm decision or FOMO? | Next time, use 24-hour cooldown |
| Recency | Have you overweighted news from the past month? | Calibrate with base rate |
Self-check (3 items checked → Part 2.A complete):
- You can list the 6 biases + the 2-3 you most commonly fall into
- You've added fair value (not just entry price) to your thesis yaml
- You've instituted at least 1 hack: 24-hour cooldown / monthly review test
6. Part 2.A Complete¶
🎉 Completed the 5 chapters of the Universal Investment Model. You now have:
- ✅ Valuation basics (DCF / multiples / reverse DCF + margin of safety, P2A-C1)
- ✅ 5 Major Mental Models (Buffett / Munger / Graham, P2A-C2)
- ✅ Historical base rates (dotcom / mobile / industrial revolution, P2A-C3)
- ✅ Portfolio construction (sizing + concentration + risk parity, P2A-C4)
- ✅ Behavioral finance (6 biases + 6 hacks, P2A-C5)
Part 2.A is a universal toolkit for any industry. AI-specific investing tools are in Part 2.B (4-dimension thesis / KPIs / glossary / walkthrough / self-written thesis).
After finishing Part 2 → Move to Part 3 for real analysis workflows (hedge fund / Buffett / finding bottlenecks / multi-PM / anti-thesis).
7. Deep Dive (optional): Kahneman / Thaler / Munger Behavioral Finance Reading¶
Click to see 5 core readings
Daniel Kahneman 《Thinking, Fast and Slow》 (book): - System 1 (fast) vs. System 2 (slow) thinking - All investment biases stem from System 1 automatically taking over - Recommended: Chapters 11-19 (anchoring / availability / framing)
Richard Thaler 《Misbehaving》 (book): - Founder of behavioral economics - Many specific investment cases
Charlie Munger 《Poor Charlie's Almanack》Chapter 11: - "Psychology of Human Misjudgment" — Munger's long speech on 25 biases - Public PDF available online
Howard Marks memos: - 2008 "The Tide Goes Out" — investor psychology cycles - 2017 "There They Go Again... Again" — bubble vs. healthy
Annie Duke 《Thinking in Bets》: - Poker mindset + investment decisions - Emphasizes "process > outcome" — good decisions can have bad results
Read 1 book per month, finish in 1 year — your behavioral decision-making will transform.