0/70 completed
Behavioral Model Interactive

Prospect Theory

People feel losses 2x more than equivalent gains. Understanding loss aversion enables better payout structures and promotion design.

๐Ÿง  The Core Insight

๐Ÿ˜ข
-$100
Feels like losing
-225 utils
๐Ÿ˜Š
+$100
Feels like gaining
+100 utils

Loss aversion coefficient ฮป โ‰ˆ 2.25: losses hurt more than twice as much as equivalent gains please.

Value Function Parameters

ฮฑ (Sensitivity) 0.88
0.5 1
ฮป (Loss Aversion) 2.25
1 4
Reference Point 0
-50 50

๐Ÿ“Š Current Settings

Gain sensitivity 0.88
Loss multiplier 2.25x
Break-even ratio Win $225 to offset $100 loss

Prospect Theory Value Function

Note: Curve is steeper for losses (left) than gains (right). This asymmetry = loss aversion.

Gamble Evaluation

Gamble Amount ($) 100
10 500
Win Probability 0.5
0.1 0.9
Expected Value
+$0
Rational: Take it
Prospect Value
-36
Behavioral: Pass

โš ๏ธ Loss aversion makes people reject this +EV gamble!

๐Ÿ“š Key Concepts

Loss Aversion

Losses hurt 2-2.5x more than equivalent gains

Losing $100 feels worse than winning $100 feels good

Implication: Users avoid risky bets even with +EV

Diminishing Sensitivity

Value function is concave for gains, convex for losses

$100โ†’$200 feels bigger than $1000โ†’$1100

Implication: Small wins feel proportionally better

Reference Point

Gains/losses evaluated vs reference, not absolute

Winning $50 after expecting to win $100 = loss

Implication: Frame outcomes relative to expectations

๐Ÿ’ฐ Pricing Applications

Payout Design

  • โ†’ Smaller, more frequent wins feel better than rare big wins
  • โ†’ Insurance/cashout options appeal to loss-averse users
  • โ†’ Frame bonuses as gains, not reduced losses

Promotion Design

  • โ†’ "Win $20 free" beats "Get 20% back on losses"
  • โ†’ Loss rebates feel 2x as valuable as equivalent bonuses
  • โ†’ Streak bonuses leverage reference point shifts

R Code Equivalent

# Prospect Theory value function
prospect_value <- function(x, alpha = 0.88, lambda = 2.25, ref = 0) { 
  adjusted <- x - ref
  ifelse(adjusted >= 0,
         adjusted^alpha,
         -lambda * (-adjusted)^alpha)
}

# Evaluate a gamble
evaluate_gamble <- function(win_amount, lose_amount, win_prob,
                            alpha = 0.88, lambda = 2.25) { 
  ev <- win_prob * win_amount - (1 - win_prob) * lose_amount
  pv <- win_prob * prospect_value(win_amount, alpha, lambda) +
        (1 - win_prob) * prospect_value(-lose_amount, alpha, lambda)
  
  list(expected_value = ev, prospect_value = pv,
       rational_choice = ev >= 0, behavioral_choice = pv >= 0)
}

# Example: 50/50 bet for $100
result <- evaluate_gamble(100, 100, 0.5)
cat(sprintf("EV: $%.0f | PV: %.0f | Take it: %s\n",
            result$expected_value, result$prospect_value,
            result$behavioral_choice))

โœ… Key Takeaways

  • โ€ข Losses hurt ~2.25x more than gains please
  • โ€ข People reject +EV gambles due to loss aversion
  • โ€ข Frame outcomes relative to reference points
  • โ€ข Frequent small wins > rare big wins psychologically
  • โ€ข Insurance/cashout taps into loss aversion
  • โ€ข Loss rebates have outsized perceived value

Pricing Models & Frameworks Tutorial

Built for mastery ยท Interactive learning