Threshold Trader

Automatically trades prediction markets when prices deviate beyond your configured probability threshold with Kelly Criterion sizing and safety checks.

Security Scan

The skill's code, runtime instructions, and declared requirements are coherent with a prediction-market trading template that uses a single Simmer API key; nothing in the bundle suggests it is trying to do unrelated or secret-harvesting work.

Purpose & Capability

The skill claims to trade prediction markets and requires a Simmer API key and the simmer-sdk; the Python code calls SimmerClient methods (get_markets, trade, auto_redeem, etc.) which are appropriate for that purpose. There are no unrelated credentials or unexpected binaries requested.

Instruction Scope

SKILL.md instructions and the Python code stick to market discovery, edge calculation, safety checks, sizing, execution, and redemption. Publish scripts call npx and git, but those are for publishing the package (not runtime trading). The runtime code does not read system files or unrelated environment variables beyond the declared configuration.

Install Mechanism

There is no download-from-URL install step. The package expects a pip dependency (simmer-sdk) and contains Python code; that is proportionate to its function. No extract-from-arbitrary-URL or non-standard binary installs are present.

Credentials

The primary credential is SIMMER_API_KEY (declared in clawhub.json and used by the code). Other env vars are non-sensitive configuration (thresholds, venue, sizing). Note: top-level metadata in the evaluation header said 'Required env vars: none' which contradicts clawhub.json and the SKILL.md; the actual code requires SIMMER_API_KEY.

Persistence & Privilege

clawhub.json configures an automaton with cron '*/15 * * * *' and an entrypoint, so when installed the agent runtime may schedule scans every 15 minutes. By default the script is dry-run unless run with --live; automated scheduling combined with enabling live mode could execute real trades (financial impact), so users should be deliberate about running with live enabled.

Assessment

This skill appears to be what it says: a trading template that requires only your SIMMER_API_KEY and the simmer-sdk. Before installing: 1) Confirm you trust the Simmer service and the source of this skill; 2) Keep the API key secret (the skill will use it to place trades); 3) Run in 'sim' (paper) mode first — the code defaults to dry-run but can place real trades with --live; 4) Note the automaton cron will run every 15 minutes when managed by the platform, so disable automation or avoid enabling live mode until you’ve tested; 5) If you have concerns, review the threshold_trader.py code (it’s contained in the bundle) or run it in an isolated environment first. Also be aware of a small metadata inconsistency: the top-level summary in the provided manifest claims 'no required env vars' but the code and clawhub.json require SIMMER_API_KEY.

A simple probability-threshold trading skill for prediction markets. Trades when market prices diverge from your configured probability threshold by a minimum edge amount.

This is a template. The default signal is a fixed probability threshold — remix it with your own model predictions, external data feeds, or custom signals. The skill handles all the plumbing (market discovery, trade execution, safeguards). Your agent provides the alpha.

How It Works

1. Monitor Markets: Scans active markets on your configured venue (Polymarket or sim)
2. Check Threshold: Compares current market price against your configured probability
3. Calculate Edge: Only trades when edge exceeds minimum threshold (default 3%)
4. Position Sizing: Uses Kelly Criterion-based sizing for optimal stake amounts
5. Safety Checks: Reviews flip-flop warnings, slippage estimates, and edge analysis before executing
6. Auto-Redeem: Automatically collects payouts from resolved markets

Configuration

Configure via environment variables:

• THRESHOLD_PROBABILITY: Your probability estimate (0.0-1.0). Default: 0.70
• TRADE_SIDE: Which side to trade ("YES" or "NO"). Default: "YES"
• SIMMER_VENUE: Trading venue ("sim" for paper, "polymarket" for real). Default: "sim"
• SIMMER_MIN_EV: Minimum edge to trade (0.0-1.0). Default: 0.03 (3%)
• SIMMER_KELLY_MULTIPLIER: Kelly fraction (0.0-1.0). Default: 0.25
• MAX_POSITION_SIZE: Maximum position size in USD. Default: 100.0

Usage

Install the skill:

npx clawhub@latest install threshold-trader

Set your credentials and configuration:

export SIMMER_API_KEY="your-api-key-here"
export THRESHOLD_PROBABILITY="0.65"
export TRADE_SIDE="YES"
export SIMMER_VENUE="sim"  # or "polymarket" for real money

Run manually:

python threshold_trader.py

Or let the automaton run it every 15 minutes (configured in clawhub.json).

Remixing

The core logic is in should_trade() — swap the fixed threshold for:

• Your ML model: Replace THRESHOLD_PROBABILITY with model predictions
• External APIs: Fetch signals from news feeds, social sentiment, or market data
• Multi-market strategies: Scan correlations across related markets
• Event-driven: React to external triggers (sports scores, election results)

The skill handles position sizing, safety checks, execution, and redemption. You focus on the signal.

Safety Features

• Flip-flop detection: Skips trades if you've been reversing positions too frequently
• Slippage protection: Avoids illiquid markets with high slippage
• Edge validation: Only trades when expected value exceeds threshold
• Dry-run default: Paper trades unless you explicitly pass --live

Examples

Paper trade YES when probability is above 70%:

export THRESHOLD_PROBABILITY="0.70"
export TRADE_SIDE="YES"
python threshold_trader.py

Real money trade NO when probability is below 30%:

export THRESHOLD_PROBABILITY="0.30"
export TRADE_SIDE="NO"
export SIMMER_VENUE="polymarket"
python threshold_trader.py --live

Requirements

• simmer-sdk >= 1.0.0
• Python >= 3.8
• SIMMER_API_KEY from simmer.markets/dashboard

---

name: threshold-trader description: Simple probability-threshold trading skill. Trades when market prices diverge from your configured threshold. Remixable template for custom signals. metadata: author: "Simmer Community" version: "1.0.0" displayName: "Threshold Trader" difficulty: "beginner"