Research Thread Tables — Place Your Bets
Each thread is a table at the casino. Composite scores are the house edge. Higher scores = better odds for the house (more reliable returns). Kelly fraction tells you optimal allocation.
🪙 The Kelly Criterion Laboratory
A biased coin demonstrates why optimal bet sizing matters. This same math determines how much of our research budget goes to each GLIM thread.
Biased Coin Flip
0 flips
H
T
Click the coin to flip
0%
Kelly Optimal Fraction: f* = (bp − q) / b
Bankroll Evolution
$1,000
0
Wins
0
Losses
0%
ROI
💡 The Connection to GLIM
Each GLIM research thread has a "win probability" (its composite score normalized) and "payout" (potential for breakthrough). The Kelly Criterion tells us the optimal fraction of research budget to allocate to each thread. Threads with high scores AND high novelty deserve larger bets. Threads that improved dramatically between waves (like coordinator_synthesis at +1.25) are "hot streaks" — momentum signals. The Allocation Engine tab computes these fractions for all 9 threads simultaneously using the generalized Kelly formula.
📊 Generalized Kelly Allocation
Multi-asset Kelly criterion applied to GLIM research threads. Each thread's allocation is proportional to its edge (composite score − threshold) divided by variance (score spread).
Research Budget Allocation
Kelly Optimal
⚡ Cross-Pollination Breakthroughs
Wave 3 synthesis discovered 8 actionable cross-links between research threads. These are where real breakthroughs live.
🏆 Sprint Superlatives
9.05
Top Score
Bayesian Active Learning
+1.25
Biggest Improver
Coordinator Synthesis
650K
Chars Generated
~162 pages of research
8
Cross Links
All actionable
🏘️ Village Agent Roster
The Hermitage Agent Manager orchestrates these specialist agents across both initiatives.
Dispatch Log
0 dispatches
Waiting for agent dispatches...