ALGORITHMIC
INTELLIGENCE

Autonomous agents built to find and exploit mispriced binary contracts on the world's most regulated prediction exchange.

Meet Razorbill → All Models LinkedIn
BTC · 15-MIN · ETH · LIVE · SOL · ACTIVE · XRP · MONITORING · RAZORBILL · IN TESTING · LOGICULL · PREDICTION INTELLIGENCE · CFTC-REGULATED · 97% · ACCURACY · BTC · 15-MIN · ETH · LIVE · SOL · ACTIVE · XRP · MONITORING · RAZORBILL · IN TESTING · LOGICULL · PREDICTION INTELLIGENCE · CFTC-REGULATED · 97% · ACCURACY ·
About Logicull

WHERE MATH MEETS MARKETS

We built Logicull to exploit a simple truth: prediction markets are systematically mispriced, and algorithms find edge better than humans.

An autonomous algorithmic trading system built for Kalshi — the CFTC-regulated prediction exchange. Trading 15-minute binary options across four major crypto assets with an architecture rebuilt three times from the ground up.

Each generation was born from disciplined loss analysis. Every failure became a systematic fix. The result is a machine that classifies windows into mathematically validated tiers and only deploys capital when the edge is real.

01
Tier Classification
Every window scored into five tiers. No guessing — only certainty triggers execution.
02
Multi-Asset Engine
BTC, ETH, SOL, XRP — four parallel coroutines each with own signal stack and regime.
03
AI Decision Layer
Claude-powered reasoning validates every non-trivial trade — a second layer of verification.
04
Compounding Accuracy
90.4% → 92.3% → 94.7% → 95.6%. Each generation a structural fix on the last.
CULLING THE NOISE Thousands of signals. Only certainty moves capital.
Razorbill
Active · In Testing · Razorbill

MEET
RAZORBILL

The most architecturally advanced generation yet. Built to eliminate the systematic failure modes of its predecessor and push accuracy to a new ceiling.

01
Stop-Loss Protection
Exits positions at 20% loss instead of holding to resolution. Worst single loss capped at $112, down from $487.
02
Signal Stability Filter
Requires 3 consecutive same-direction reads before allowing a trade. Blocks flip-flopping signals entirely.
03
Polymarket Integration
Cross-exchange divergence signal comparing Polymarket momentum vs Gate.io spot for smarter entry decisions.
04
82% Break-Even
Break-even win rate drops from 95% to 82% with stop-loss. 13.7% margin of safety above break-even.
Previous Models

Each a complete architectural rethink, driven by the losses the previous model could not prevent.

NUTHATCH
90.4% Accuracy

The foundation. BTC-only signal classification with Kelly Criterion sizing, ADX regime detection, and the execution infrastructure all subsequent models build upon.

BTC OnlySignal CascadeKelly SizingRegime Detection
HERON
92.3% Accuracy

Complete rewrite. Five-tier classification system scaled to four simultaneous markets with independent async coroutines and fair value models.

4 MarketsTier SystemAsync EngineClaude Brain
ORIOLE
94.7% Accuracy

Independent alt signals with anti-correlation detection and a direction-conflict guard. Per-asset volatility regimes computed from native price history.

Independent SignalsAnti-CorrelationConflict GuardPer-Asset ATR

Read Our
Blog

Deep dives into algorithmic trading, prediction market mechanics, and the engineering decisions behind our autonomous agents. Research notes, model updates, and market analysis.

View All Posts
April 7, 2026
Crypto Prediction Markets: Where the Edge Lives
Binary contracts on BTC and ETH move differently than spot. We break down where mispricings form, why they persist, and what makes crypto prediction markets uniquely exploitable.
April 2, 2026
Understanding Binary Contracts
A primer on binary options contracts — how they work, how they're priced, and why regulated prediction exchanges create unique opportunities for algorithmic traders.
March 20, 2026
Welcome to the Logicull Blog
Introducing our blog — a space for deep dives into algorithmic trading, prediction market mechanics, and the engineering decisions behind our autonomous agents.
Contact Us
Name
Email
Subject
MSG