6.6.7Factor & Behavioral Finance

Understand anchoring and confirmation bias

1,812 words8 min readdifficulty · medium1 backlinks

WHAT are these biases?


WHY do these biases exist?


HOW anchoring distorts an investment decision (derive it)

Let's build a simple model of rational vs. anchored valuation.

A rational investor should value a stock only by its fair value VV^*, based on fundamentals (cash flows, growth). But an anchored investor's perceived value V^\hat{V} is a blend of the true value and the anchor AA:

V^=(1w)V+wA\hat{V} = (1-w)\,V^* + w\,A

Why this form? If the investor fully adjusted away from the anchor, weight w=0w=0 and V^=V\hat V = V^* (rational). If they were totally stuck on the anchor, w=1w=1 and V^=A\hat V = A. Real people sit in between, so 0<w<10<w<1 is the "stickiness" of the anchor.

The decision error is:

ε=V^V=w(AV)\varepsilon = \hat V - V^* = w\,(A - V^*)

Why this step? Subtract the correct value from the perceived value. The error is proportional to (a) how sticky the anchor is, ww, and (b) how far the anchor sits from truth, (AV)(A - V^*).

The "won't sell at a loss" trap

Suppose you bought at price A = \100(youranchor),butfairvaluedroppedto(your anchor), but fair value dropped toV^* = $70.Withstickiness. With stickiness w=0.5$:

V^=0.5(70)+0.5(100)=$85\hat V = 0.5(70) + 0.5(100) = \$85

You think the stock is worth 85,soyourefusetosellat85, so you refuse to sell at 70 — even though the market agrees it's worth $70. Your error is:

ε=0.5(10070)=+$15\varepsilon = 0.5(100 - 70) = +\$15

You are overvaluing by $15 purely because of your purchase price.

Figure — Understand anchoring and confirmation bias

HOW confirmation bias compounds it (Bayesian view)

Rational updating uses Bayes' rule. For a hypothesis HH ("this stock will rise") given evidence EE:

P(HE)=P(EH)P(H)P(E)P(H\mid E) = \frac{P(E\mid H)\,P(H)}{P(E)}

Why this matters: A rational investor updates on all evidence EE. A confirmation-biased investor filters the evidence set — they only let in evidence where P(EH)P(E\mid H) is high (supportive), and mentally discard disconfirming evidence.

So instead of updating on the full set {E1,E2,...,En}\{E_1, E_2, ..., E_n\}, they update only on the supportive subset. The result: their posterior belief P(HE)P(H\mid E) stays artificially high — the numerator keeps getting boosted, disconfirming evidence never lowers it.


Worked examples


Steel-manned mistakes


Flashcards

What is anchoring bias?
Over-relying on the first number seen (e.g. purchase price, 52-week high) and adjusting away from it insufficiently.
What is confirmation bias?
Seeking/interpreting/remembering only evidence that supports what you already believe, ignoring contradicting evidence.
In the model V^=(1w)V+wA\hat V=(1-w)V^*+wA, what does ww represent?
The "stickiness" of the anchor — how much the perceived value is dragged toward the anchor AA instead of the true value VV^*.
What is the decision error caused by anchoring?
ε=w(AV)\varepsilon = w(A-V^*) — proportional to anchor stickiness and the distance of the anchor from fair value.
Why does confirmation bias inflate belief in Bayesian terms?
It filters evidence to only supportive signals (likelihood ratio > 1), so the posterior keeps rising and never gets corrected downward.
Best test to beat the purchase-price anchor?
"Would I buy this stock today with fresh cash?" If no, sell — your past price is irrelevant.
One antidote to confirmation bias before buying?
A pre-mortem: write the 3 strongest reasons the thesis could be wrong.
Why is anchoring an "efficient" but bad shortcut?
Fully adjusting a number is mentally costly, so the brain stops adjusting too early — fast but inaccurate.

Recall Feynman: explain to a 12-year-old

Imagine you buy a toy for 10.Latereveryonesaysitsreallyonlyworth10. Later everyone says it's really only worth 6. But because you *paid 10,yourbrainkeepswhispering"no,itsworthlike10*, your brain keeps whispering "no, it's worth like 8!" That sticky first number is anchoring. Now say you think this toy is the best toy ever — so you only listen to friends who agree and you ignore anyone who says it's junk. That's confirmation bias: you only collect "you're right!" and throw away "you're wrong." Together they trick you into keeping a bad toy way too long.


Connections

  • Loss Aversion and the Disposition Effect — why anchoring on buy price makes you hold losers
  • Overconfidence Bias — confirmation bias is its main fuel
  • Bayesian Updating in Investing — the rational benchmark biases violate
  • Behavioral Finance Overview — parent framework
  • Efficient Market Hypothesis — biases explain persistent inefficiencies
  • Mean Reversion vs. Anchoring — distinguishing valid reference points from false anchors

Concept Map

produces

produces

grabs first number

blends into

blends into

minus V-star gives

causes

filters evidence in

biased update ignores

reinforces

leads to

leads to

Lazy brain shortcuts

Anchoring bias

Confirmation bias

Anchor A: purchase price or 52wk high

Perceived value V-hat

Fair value V-star

Decision error = w times A minus V-star

Won't sell at a loss trap

Bayesian updating P of H given E

Contradicting evidence

Hold losers and chase stories

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Dekho, do bahut common mental traps hain jo har investor phasata hai. Pehla hai anchoring — matlab tumhara dimaag pehle number ko pakad leta hai aur chhodta nahi. Jaise tumne stock 100pekhareeda,abwoh100 pe khareeda, ab woh 70 ka ho gaya, phir bhi dimaag bolta hai "yaar 85tohaihi!"kyunki85 to hai hi!" — kyunki 100 wala buy price ek "anchor" ban gaya. Reality yeh hai ki market ko pata bhi nahi tumne kitne ka khareeda; future value sirf fundamentals se decide hoti hai, tumhare purchase price se nahi.

Dusra trap hai confirmation bias — jab tum ek stock kharid lete ho, to sirf woh news padhte ho jo bolti hai "tum sahi ho." Bearish articles ko mute kar dete ho, debt footnote skip kar dete ho. Bayes ke terms mein, tum sirf supportive evidence ko andar aane dete ho, isliye tumhara belief artificially high rehta hai aur kabhi correct nahi hota. Phir jab bura news aata hai, tum totally blindsided ho jaate ho.

Simple model yaad rakho: V^=(1w)V+wA\hat{V}=(1-w)V^*+wA. Yahan ww anchor ki "chipakne ki taakat" hai. Jitna zyada ww, utna zyada tumhari valuation galat — error hota hai w(AV)w(A-V^*). Yeh dono biases milke tumhe losers ko hold karne pe majboor karte hain.

Ilaaj? Anchoring ke liye poochho: "Aaj agar mere paas cash hota, to kya main abhi yeh stock kharidta?" Agar nahi, to bech do. Confirmation bias ke liye pre-mortem karo — buy karne se pehle 3 sabse strong reasons likho ki tum galat kyun ho sakte ho. Yeh discipline hi tumhe emotional galtiyon se bachayega.

Test yourself — Factor & Behavioral Finance

Connections