← Back to list
prediction-market-strategy Checked
Venue- and market-type-agnostic strategy, sizing, and backtesting layer for binary prediction mar…
8.5Overall rating
Installation
🤖 Install via AI
Copy the prompt below and send it to your AI assistant (e.g. Claude Code) — it will follow the instructions and install automatically.
Install the "prediction-market-strategy" skill by following the instructions at https://skill123.me/install/prediction-market-strategy.
⌨️ Command line install
Run in your terminal — downloads and installs to ~/.claude/skills/.
curl -fsSL https://skill123.me/install/prediction-market-strategy.sh | bash
📦 Download ZIP
Download the zip and extract it into your skills directory (e.g. ~/.claude/skills/), then restart your session.
⬇ Download v1.0.0 · 19 KBAbout this skill
Overview
Venue- and market-type-agnostic strategy, sizing, and backtesting layer for binary prediction markets (Kalshi, Polymarket, ForecastEx). Covers the durable edge thesis, fee-aware selection, fractional-Kelly sizing, and leak-free validation methodology.
Source
- Repo: https://github.com/agiprolabs/claude-trading-skills
- Path: skills/prediction-market-strategy
Score breakdown
Trigger
7.0
What/venues/synonyms (Kelly, favorite-longshot, backtesting) covered well; no explicit When-Not section, no trigger-phrase variants; English-only but this is a public English skill so no penalty.
Structure
8.7
199-line SKILL.md with 4 well-scoped references and one script; clean separation; Files section annotates each reference but does not state when to load each during a task (minor).
Workflow
8.4
Task-oriented selection procedure (net_edge > theta gate, ladder walk, slippage guard, exposure caps) with concrete low-freedom formulas in script and high-freedom strategy discussion; strong error handling via phantom-edge hall of fame; validation loop via forward paper-trading and Brier comparison. Slightly knowledge-heavy in the thesis/literature sections.
Content
9.0
Concise imperative instructions; formulas with concrete defaults and worked example; tables used effectively; minimal filler.
Engineering
10.0
Frontmatter valid; script has shebang, stdlib-only, self-documented; all referenced paths exist.
Security
9.0
6.0 manifest: no network egress, no file writes outside cwd, no credential access, no persistence — script is pure computation. Capabilities match declared scope ('runs offline' claim verified). No hardcoded keys, no injection text. Minor: no investment-risk disclaimer for a skill that outputs concrete position-sizing thresholds.
Strong strategy/sizing/backtesting knowledge layer for prediction markets with fee-aware edge gates, quarter-Kelly sizing, and an excellent phantom-edge bug catalog. Sizing script is pure stdlib and offline. Deductions: description lacks explicit When-Not/boundary language and trigger variants; references listed but no explicit load-when guidance; no financial-risk disclaimer despite giving concrete trading thresholds (theta gates, position sizes).