polymarket-fast-loop-improved

Trade Polymarket BTC/ETH/SOL 5-minute and 15-minute fast markets using multi-signal CEX momentum. Adds funding rate confirmation, order book imbalance, time-of-day filtering, volatility-adjusted sizing, win-rate calibration, and fee-accurate EV math. Use when user wants to trade sprint/fast markets with a more rigorous edge filter.

Safety Notice

This listing is from the official public ClawHub registry. Review SKILL.md and referenced scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "polymarket-fast-loop-improved" with this command: npx skills add oryselias/fastloop

Polymarket FastLoop Trader — Improved

An enhanced version of the Simmer FastLoop skill with rigorous edge filtering, multi-signal confirmation, and real calibration tracking.

Default is paper mode. Use --live for real trades. Always run 100+ paper trades first to validate your win rate before going live.

⚠️ Fast markets carry Polymarket's 10% fee. Your signal needs to be right 63%+ of the time to profit. This skill will tell you your actual win rate.

Key Improvements Over Original

FeatureOriginalImproved
Fee mathApproximateExact breakeven with configurable buffer
SignalBinance momentum onlyMomentum + funding rate + order book
Momentum threshold0.5% (too low)1.0% default, calibration-driven
Time filteringNoneSkips low-liquidity hours
Position sizingFixedVolatility-adjusted
Win rate trackingNoneLogs outcomes, reports calibration
Market selectionSoonest expiryConfigurable sweet-spot window
StatsNoneFull P&L, win rate, signal breakdown

Quick Start

# Install dependency
pip install simmer-sdk

# Set API key
export SIMMER_API_KEY="your-key-here"

# Paper mode — see what would happen (default)
python fastloop_improved.py

# Go live
python fastloop_improved.py --live

# Check calibration stats (win rate, P&L, signal accuracy)
python fastloop_improved.py --stats

# Resolve any expired paper trades against real outcomes
python fastloop_improved.py --resolve

# Quiet mode for cron
python fastloop_improved.py --live --quiet

How to Run on a Loop

OpenClaw native cron:

openclaw cron add \
  --name "FastLoop Improved" \
  --cron "*/5 * * * *" \
  --tz "UTC" \
  --session isolated \
  --message "Run improved fast loop: cd /path/to/skill && python fastloop_improved.py --live --quiet. Show output summary." \
  --announce

Linux crontab:

*/5 * * * * cd /path/to/skill && python fastloop_improved.py --live --quiet

Configuration

# Raise momentum threshold (recommended: 1.0–2.0%)
python fastloop_improved.py --set min_momentum_pct=1.5

# Require order book confirmation
python fastloop_improved.py --set require_orderbook=true

# Set sweet-spot window for market selection (seconds remaining)
python fastloop_improved.py --set target_time_min=90 --set target_time_max=180

# Disable time-of-day filter (trade 24/7)
python fastloop_improved.py --set time_filter=false

All Settings

SettingDefaultDescription
entry_threshold0.05Min divergence from 50¢
min_momentum_pct1.0Min % BTC move (raised from 0.5)
max_position5.0Max $ per trade
signal_sourcebinancebinance or coingecko
lookback_minutes5Candle lookback window
min_time_remaining60Skip if less than N seconds left
target_time_min90Prefer markets with ≥ N seconds left
target_time_max210Prefer markets with ≤ N seconds left
assetBTCBTC, ETH, or SOL
window5m5m or 15m
volume_confidencetrueSkip low-volume signals
require_fundingfalseRequire funding rate confirmation
require_orderbookfalseRequire order book imbalance confirmation
time_filtertrueSkip low-liquidity hours (02:00–06:00 UTC)
vol_sizingtrueAdjust size by recent volatility
fee_buffer0.05Extra edge required above fee breakeven
daily_budget10.0Max spend per UTC day
starting_balance1000.0Paper portfolio starting balance

Signal Logic

Three signals are evaluated independently. The momentum signal is always required. Funding and order book are optional confirmation layers.

1. Momentum (always on)

  • Fetch N one-minute Binance candles
  • momentum = (close_now - open_then) / open_then * 100
  • Must exceed min_momentum_pct

2. Funding Rate (optional, require_funding=true)

  • Fetch Binance perpetual funding rate for the asset
  • Positive funding + upward momentum = longs crowded, signal is weaker → SKIP
  • Negative funding + upward momentum = confirmation → TRADE
  • Logic inverted for downward momentum

3. Order Book Imbalance (optional, require_orderbook=true)

  • Fetch top 20 levels of Binance L2 book
  • imbalance = (bid_depth - ask_depth) / (bid_depth + ask_depth)
  • Imbalance > 0.1 confirms upward momentum
  • Imbalance < -0.1 confirms downward momentum

Fee-Accurate EV

entry_price  = market price of chosen side
win_profit   = (1 - entry_price) × (1 - fee_rate)
breakeven    = entry_price / (win_profit + entry_price)
required_div = (breakeven - 0.50) + fee_buffer

Trade only fires if actual_divergence ≥ required_div.

Time-of-Day Filter

Skips 02:00–06:00 UTC by default. US session (13:00–21:00 UTC) is the highest-liquidity window for crypto prediction markets.

Volatility-Adjusted Sizing

24h_vol = std(hourly_returns_last_24h) × √24
size    = max_position × min(1.0, 0.02 / 24h_vol)

High volatility → smaller position. Low volatility with strong trend → full size.

Win Rate Calibration

The skill tracks every paper and live trade in fastloop_ledger.json. After market expiry, run --resolve to fetch the actual Polymarket outcome and log it. After 50+ trades, --stats shows your real win rate broken down by momentum threshold, time of day, and asset — so you can tune settings based on actual data rather than guessing.

Troubleshooting

All troubleshooting from the original skill applies. Additional:

"Funding rate fetch failed"

  • Binance futures API may be rate-limited. Skill falls back to momentum-only.

"Order book imbalance: neutral"

  • Market is balanced, signal is ambiguous — skipped if require_orderbook=true.

"Time filter: low liquidity window"

  • Current UTC hour is in the 02–06 block. Set time_filter=false to override.

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

Dlazy Kling Audio Clone

Generate customized speech that highly restores the timbre by uploading reference audio using Kling Audio Clone.

Registry SourceRecently Updated
General

Dlazy Keling Sfx

Generate matching scene sound effects based on text descriptions or video frames using Kling SFX.

Registry SourceRecently Updated
General

Dlazy Keling Tts

Convert text into high-quality, emotional speech reading using Kling TTS.

Registry SourceRecently Updated
General

Dlazy Jimeng T2i

Text-to-image generation with Jimeng, quickly converting text to high-quality images.

Registry SourceRecently Updated
polymarket-fast-loop-improved | V50.AI