Simmer Signal Sniper

Your signals, Simmer's trading intelligence.

When to Use This Skill

Use this skill when the user wants to:

Monitor RSS feeds for trading opportunities
Trade on breaking news before markets react
Configure their own signal sources and keywords
Get Trading Agent-grade safeguards on their trades

Quick Commands

# Check account balance and positions
python scripts/status.py

# Detailed position list
python scripts/status.py --positions

API Reference:

Base URL: https://api.simmer.markets
Auth: Authorization: Bearer $SIMMER_API_KEY
Portfolio: GET /api/sdk/portfolio
Positions: GET /api/sdk/positions

Quick Start (Ad-Hoc Usage)

User provides RSS feed and market directly:

User: "Watch this RSS feed for greenland news: https://news.google.com/rss/search?q=greenland"
User: "Snipe any news about trump from this feed"

→ Run with --feed and --market flags:

python signal_sniper.py --feed "https://news.google.com/rss/search?q=greenland" --market "greenland-acquisition" --dry-run

Persistent Setup (Optional)

For automated recurring scans, configure via environment:

Setting	Environment Variable	Default	Description
RSS Feeds	`SIMMER_SNIPER_FEEDS`	(none)	Comma-separated RSS URLs
Markets	`SIMMER_SNIPER_MARKETS`	(none)	Comma-separated market IDs
Keywords	`SIMMER_SNIPER_KEYWORDS`	(none)	Comma-separated keywords to match
Confidence	`SIMMER_SNIPER_CONFIDENCE`	0.7	Min confidence to trade (0.0-1.0)
Max USD	`SIMMER_SNIPER_MAX_USD`	25	Max per trade
Max trades/run	`SIMMER_SNIPER_MAX_TRADES`	5	Maximum trades per scan cycle

Polymarket Constraints:

Minimum 5 shares per order
Trades below this threshold are rejected with an error message

How It Works

Each cycle the script:

Polls configured RSS feeds
Filters articles by keywords (if configured)
Matches articles to target markets
For each match, calls SDK context endpoint for safeguards:
- Position awareness (already holding?)
- Flip-flop detection (recently changed direction?)
- Slippage estimates (is market liquid?)
- Time decay (resolving soon?)
- Resolution criteria (what actually resolves this market?)
If safeguards pass, you (Claude) analyze the signal
If confident, executes trade via SDK
Tracks processed articles to avoid duplicates

Running the Skill

Run a scan:

python signal_sniper.py

Dry run (no actual trades):

python signal_sniper.py --dry-run

Check for signals without trading:

python signal_sniper.py --scan-only

View current config:

python signal_sniper.py --config

Override for one run:

python signal_sniper.py --feed "https://..." --keywords "trump,greenland" --market "abc123"

Show processed articles:

python signal_sniper.py --history

Interpreting Context Warnings

Before trading, ALWAYS check the context warnings. The skill will show you:

Warning	Action
`MARKET RESOLVED`	Do NOT trade
`HIGH URGENCY: Resolves in Xh`	Consider if signal is timely enough
`flip_flop_warning: SEVERE`	Skip - you've been reversing too much
`flip_flop_warning: CAUTION`	Proceed carefully, need strong signal
`Wide spread (X%)`	Reduce position size or skip
`Simmer AI signal: X% more bullish/bearish`	Consider Simmer's oracle opinion

Analyzing Signals

When you find a matching article, analyze it carefully:

Read the headline and summary - What is the actual news?
Check resolution_criteria - What ACTUALLY resolves this market?
- Example: "greenland" in headline doesn't mean "acquisition complete"
- The resolution might be "US formally acquires Greenland by 2027"
- Does this signal move the needle on THAT specific criteria?
Assess confidence (0.0-1.0):
- How directly does this signal relate to resolution criteria?
- Is the source credible?
- Is this news likely already priced in?
Only trade if:
- Confidence > threshold (default 0.7)
- No severe warnings
- Signal validates against resolution criteria

Example Conversations

User: "Set up news sniping for the Greenland market" → Ask for RSS feeds they want to monitor → Configure with market ID and keywords → Enable cron for recurring scans

User: "Check this feed for trading signals" → Run: python signal_sniper.py --feed "URL" --scan-only → Show found articles and potential matches

User: "Snipe any bitcoin news from CoinDesk" → Run with CoinDesk RSS and bitcoin-related markets → Show matches and ask if they want to trade

User: "What signals have we processed?" → Run: python signal_sniper.py --history → Show recent articles and actions taken

Example Trade Flow

1. RSS poll finds: "Trump and Denmark reach preliminary Greenland agreement"
2. Keywords match: "greenland", "trump"
3. Call context endpoint for market "greenland-acquisition-2027"
4. Check warnings: none severe ✓
5. Resolution criteria: "Resolves YES if US formally acquires Greenland by 2027"
6. You analyze: "preliminary agreement" ≠ "formally acquires" but bullish signal
7. Confidence: 0.75 (positive indicator, not definitive)
8. Check slippage: 2.5% on $25 ✓
9. Execute: BUY YES $25
10. Report: "🎯 Sniped: Trump/Greenland agreement → BUY YES $25"

Troubleshooting

"No feeds configured"

Provide feeds in message: "watch this RSS: https://..."
Or set SIMMER_SNIPER_FEEDS environment variable

"No matching articles found"

Check keywords are correct
RSS feed might not have recent articles
Try --scan-only to see what's in the feed

"Skipped due to flip-flop warning"

You've been changing direction too much on this market
Wait before trading again, or find new information

"Slippage too high"

Market is illiquid
Reduce trade size or skip

"Already processed"

This article was already seen
Working as intended (dedup)

Finding Good RSS Feeds

Tips for choosing signal sources:

Google News RSS: https://news.google.com/rss/search?q=YOUR_TOPIC
Niche sources: Better than mainstream (less priced in)
Official sources: Government, company announcements
Twitter lists → RSS: Use services like Nitter or RSS.app

The skill works best when:

Feeds are relevant to your target markets
You have specific keywords to filter noise
Sources publish before mainstream coverage

simmer-signalsniper