Alpha Thesis
Live Dashboard →Bet Screening →

Cognitum + Foresight-v3

Prediction markets are inefficient for sports. The crowd is smart, but beatable. We combine the #1 ranked forecasting model with an agentic intelligence harness to extract alpha where retail traders cannot.

0
Games Analyzed
0.1810
Kalshi Brier Score
-13.24%
Consensus ROI
#1
ProphetArena Rank
Scroll to explore
01 / The Problem

Prediction Markets Are Inefficient for Sports

Kalshi attracts retail traders, not sharp sports bettors. The prices are decent but far from efficient. Our analysis of 21,039 real games with resolved outcomes proves there is room for a superior model.

Brier Score Comparison (lower is better)Brier score measures calibration. 0 = perfect, 1 = worst possible.

Perfect = 0.000
Based on 302 Kalshi outcomes resolved via API + 21,039 matched sportsbook games

The Naive Strategy Loses Money

The obvious approach: find where sportsbook consensus disagrees with Kalshi, and bet the discrepancy. This loses money. We tested it on 21,039 real games.

Metric Sportsbook Consensus Kalshi Price
Accuracy 70.0% 72.7%
Brier Score 0.1901 0.1810
Edge ROI (3%+ threshold) -13.24% --

Conclusion: Kalshi is more accurate than sportsbook consensus. Betting the discrepancy is a losing strategy. The edge must come from a superior forecasting model, not from arbitraging existing price sources.

02 / The Prediction Engine

Foresight-v3 by Lightning Rod AI

The #1 ranked sports forecasting model on ProphetArena, the UChicago prediction benchmark. A 32-billion parameter model trained not on human labels, but on actual outcomes.

#1 on ProphetArena

UChicago's benchmark for prediction quality. Foresight-v3 beats GPT-5, Gemini 3 Pro, and all other frontier models on sports forecasting tasks.

32B Parameters

Trained with Future-as-Label methodology: outcome-based reinforcement learning. The model learns from what actually happened, not what humans predicted.

Calibrated Probabilities

Brier score optimized. When Foresight says 70%, it means 70%. Calibration is the foundation of profitable prediction market trading.

What Foresight-v3 Does Not Have

No Execution
Cannot place orders on Kalshi or Polymarket
No Risk Management
No sizing, limits, or circuit breakers
No Learning Loop
Does not adapt from its own trading outcomes
API Calls Made
15
Real predictions, live API
Cost for 15 Calls
$0.22
Extremely capital-efficient
03 / The Agentic Harness

Cognitum Intelligence Layer

A production-grade intelligence harness built on 11 native Rust modules. Cognitum does not predict outcomes. It turns predictions into profitable, risk-managed bets.

RuVector Sidecar

  • 11 native Rust modules via NAPI (arm64)
  • Mixture-of-Experts signal fusion (4 experts)
  • Dendritic coincidence coherence gate
  • Split conformal prediction intervals
  • VIX-based 6-regime classifier

HNSW Similar States

  • Vector search: "What happened last 10 times markets looked like this?"
  • 150x-12,500x faster than brute-force search
  • Historical context informs every decision
  • Powered by @ruvector/rvf and @ruvector/edge

Risk Controls

Max Bet
5%
of bankroll
Daily Loss
3%
circuit breaker
Kelly
0.5x
half-Kelly sizing
Min Edge
5%
to place any bet

SONA Learning

  • Records every prediction, bet, and outcome
  • Adapts sizing based on win/loss patterns
  • Identifies regime shifts in market efficiency
  • Continuous improvement loop from real results

What Cognitum Does Not Have

No Prediction Model
Without a superior forecasting source, the harness has nothing to amplify. Our data proves sportsbook consensus edges lose money (-13.24% ROI).
04 / The Combined System

How Foresight + Cognitum Create Alpha

Each component solves the other's weakness. Foresight generates calibrated probabilities. Cognitum validates, sizes, executes, and learns. Together they form a complete trading system.

🧠
Predict
Foresight-v3 generates
calibrated probability
🔍
Edge Detect
Compare to market
price on Kalshi
🛡
Coherence
Multi-signal gate
validates the edge
📊
HNSW
Historical context
from similar states
📈
Kelly Size
Half-Kelly with
5% max risk
Execute
Place order on
Kalshi / Polymarket

The Learning Loop

Every outcome feeds back through SONA. The system learns which market conditions produce the highest edge, which coherence thresholds filter noise vs. signal, and how to adjust sizing over time. This is not a static model. It is a system that improves with every resolved bet.

Execute Outcomes SONA Learn
05 / Why Combined

Neither Component Works Alone

Foresight Alone

  • Calibrated predictions
  • #1 ranked model
  • No execution
  • No risk limits
  • No edge validation
  • No learning from outcomes
A prediction without execution is just a number.
vs.

Cognitum Alone

  • Execution infrastructure
  • Risk management
  • Coherence gate
  • SONA learning
  • No prediction model
  • Consensus edges lose (-13.24%)
A harness without a signal source has nothing to amplify.

Foresight + Cognitum

Superior Prediction
Foresight's calibrated probabilities
Intelligent Execution
Coherence gate, Kelly sizing, limits
Continuous Learning
SONA adapts from every outcome
Selective Betting
Only when model + gate agree
06 / The Path Forward

From Prototype to Leaderboard

We follow a disciplined validation path. No real money until the data proves positive edge. Every phase has measurable gates that must be passed before proceeding.

Complete

Market Analysis

Analyzed 21,039 games. Resolved 302 Kalshi outcomes via API. Proved consensus edges lose money. Established Brier benchmarks. Identified the opportunity.

Complete

Foresight Integration

Foresight-v3 API connected and tested. 15 real predictions made at $0.22 total cost. OpenAI-compatible API. Partnership with Lightning Rod AI established.

In Progress

Edge Validation

Running Foresight on 50+ Kalshi sports markets daily. Tracking predictions vs. outcomes. Computing real Brier score after 200+ resolved predictions. Gate: Foresight Brier must be below 0.181.

Predictions tracked 15 / 200 target
Pending Validation

Paper Trade

Small bets ($1-5 each) on Kalshi. 10 bets/day maximum. Track every prediction, bet, and outcome. Gate: Positive P&L after 100 bets.

Pending Paper Results

Scale

Increase to $10-50 per bet. Automated scanning and execution via PM2. Gate: Sharpe ratio above 1.0 over 4 consecutive weeks.

Target

Kalshi Leaderboard

Most Kalshi traders are retail. An AI system with calibrated probabilities, intelligent risk management, and a continuous learning loop has a structural advantage. Selective betting -- only when Foresight + coherence gate agree -- is our edge.

Kalshi Balance
$10.00
Ready for Phase 2
Foresight API
LIVE
$50 credit available
Structural Advantage
AI vs. Retail
Calibrated model vs. gut feel