Investigative Data Science Forensic Tool

Sportsbook Promo Targeting & Bankroll Decay Simulator

Based on reporting by The New York Times revealing how online gambling operators deploy machine learning to identify bettors likeliest to bet and lose in response to targeted promotions. Audit personalized incentives, house vig amplification, and bankroll decay.

Bettor & Algo Config

Preloaded Behavioral Profiles

Starting Bankroll $1,000 Initial customer deposit before sports betting activity begins.

Standard Wager Size $50 Average bet size for baseline straight wagers.

Loss-Chase Susceptibility 75% Likelihood the bettor doubles wager size or re-deposits after a losing streak.

Same-Game Parlay (SGP) Share 60% Percentage of bets routed to high-hold (20-30% hold) parlays vs. straight wagers (4.5% vig).

Algorithmic Targeting Tier Aggressive AI Determines promo velocity triggered specifically upon bettor losses.

Simulation Horizon 90 Days Total consecutive wagering days modeled.

Algorithmic Extraction Telemetry

Real-Time Computed

Expected Final Bankroll $0 -100% Decay

True House Extraction $1,850 18.4% Net Hold

Promo Illusion Value +$420 -$285 Real EV

Sportsbook Target Score 94 / 100 Tier 1: Maximum Extraction

Bankroll Trajectory: Targeted AI Nudges vs. Natural Variance

With AI Promo Nudges

Baseline Discipline

Promo Injection Event

Algorithmic Classification High-Loss Target

Vulnerability Factor	Loss-Chasing Elasticity	High (0.84)
Parlay Tax Rate	Average Synthetic Hold	21.8%
Targeting Trigger	Intervention Window	< 15 min post-loss
Turnover Playthrough	Required Play to Cash Out	3x - 5x wagered

Promotional Math vs. Reality Negative EV Trap

Sportsbooks offer "$100 Bonus Bets" that exclude stake return. On -110 odds, a $100 bonus bet has an unconditioned expectation of only $45.45. When funneled into 5-leg parlays, operator hold extracts over 80% before withdrawal.

Bonus Headline: $100 "Free Bet"

Actual Cash Realization: $31.20

Subsequent Induced Loss: -$340.00

Real-Time Behavioral Targeting Sandbox

Interactive State Machine

[00:00:00] SYSTEM READY: Monitoring real-time betting telemetry...

[00:00:02] Bettor session active. Bankroll: $1,000. No pending triggers.

How Sportsbook Machine Learning Targets Vulnerable Bettors

Former DraftKings and industry data scientists detailed to The New York Times how proprietary algorithms analyze user behavior down to the minute: wager velocity, loss reactions, bet types, and response to specific push notifications.

1. The Post-Loss Intervention Window

When a bettor suffers a tough loss, algorithms detect heightened emotional arousal ("tilt"). Push notifications with deposit matches or "risk-free" tokens are automatically timed within 5 to 30 minutes to capture the bettor before they close the app.

2. The Same-Game Parlay Trap

While standard point spreads carry a 4.5% bookmaker margin (vig), Same-Game Parlays (SGPs) carry a staggering 20% to 35% house advantage. Promotional boosts ("+30% Profit Boost on 4+ Legs") incentivize bets with overwhelmingly negative true expected value.

3. Asymmetric Profiling: Churn vs. Sharp

Bettors who demonstrate mathematical discipline (shopping lines, exploiting promotional arbitrage) are swiftly restricted with $1.50 betting limits. Conversely, bettors identified as prone to losing are rewarded with VIP hosts, lavish reloads, and targeted incentives.