"Vibe-Trading: Your Personal Trading Agent"
Python ai-agentalgorithmic-tradingbacktestingfintechllmmcpmulti-agentpythonquantitative-financetrading
fix(skills): use API-native decimal funding rate in perp-funding-basis
Haozhe Wud55de48d4203
diff --git a/agent/src/skills/perp-funding-basis/SKILL.md b/agent/src/skills/perp-funding-basis/SKILL.mdindex 0b0b5df67..61842e4c5 100644--- a/agent/src/skills/perp-funding-basis/SKILL.md+++ b/agent/src/skills/perp-funding-basis/SKILL.md@@ -24,12 +24,18 @@ If perp price < spot price → funding rate negative → shorts pay longs**Annualized funding rate:**```python-# OKX funding rate is per 8-hour period-# Annualized = rate × 3 (per day) × 365-funding_rate_8h = 0.01 # 0.01% per 8h-annualized = funding_rate_8h * 3 * 365 # = 10.95% annualized+# Funding rate is a per-8h DECIMAL, exactly as OKX/Binance return it+# (the API sends "0.0001" for 0.01%). 3 funding windows/day → × 3 × 365.+funding_rate_8h = 0.0001 # 0.01% per 8h, as a decimal+annualized = funding_rate_8h * 3 * 365 # 0.1095 → 10.95% annualized```+> **Units convention (whole skill):** all *code* treats the funding rate as a+> per-8h **decimal** (`0.0001` = 0.01%), matching the raw OKX/Binance API value.+> The tables below show the equivalent **percentages** for readability — divide+> a table's `%` by 100 to get the decimal a comparison expects+> (`+0.05%` → `0.0005`).+### 2. Funding Rate Signal Framework| Funding Rate (8h) | Annualized | Market State | Signal |@@ -49,13 +55,13 @@ def funding_regime(rates_7d):consecutive_positive = all(r > 0 for r in rates_7d[-3:])consecutive_negative = all(r < 0 for r in rates_7d[-3:])- if avg > 0.03 and consecutive_positive:+ if avg > 0.0003 and consecutive_positive: # > +0.03% per 8hreturn "overheated_long" # High risk of long squeeze- elif avg > 0.01 and consecutive_positive:+ elif avg > 0.0001 and consecutive_positive: # > +0.01% per 8hreturn "bullish_carry" # Good carry trade environment- elif avg < -0.02 and consecutive_negative:+ elif avg < -0.0002 and consecutive_negative: # < -0.02% per 8hreturn "overheated_short" # High risk of short squeeze- elif avg < -0.005 and consecutive_negative:+ elif avg < -0.00005 and consecutive_negative: # < -0.005% per 8hreturn "bearish_carry" # Inverse carry tradeelse:return "neutral"@@ -108,8 +114,8 @@ def carry_trade_pnl(spot_entry, funding_rates, position_size):return total_funding_collected-# Example: $100,000 position, avg funding +0.015% per 8h, 30 days-# Revenue: 0.015% × 3 × 30 × $100,000 = $1,350 (16.2% annualized)+# Example: $100,000 position, avg funding +0.015% (0.00015 decimal) per 8h, 30 days+# Revenue: 0.00015 × 3 × 30 × $100,000 = $1,350 (16.4% annualized)```**Carry trade execution on OKX:**@@ -131,15 +137,15 @@ Different exchanges have different funding rates for the same asset. Arbitrage t```python# Example: BTC-USDT perpetual funding rates-exchange_rates = {- "OKX": 0.015, # +0.015% per 8h- "Binance": 0.020, # +0.020% per 8h- "Bybit": 0.025, # +0.025% per 8h+exchange_rates = { # per-8h decimals (API-native)+ "OKX": 0.00015, # +0.015% per 8h+ "Binance": 0.00020, # +0.020% per 8h+ "Bybit": 0.00025, # +0.025% per 8h}# Strategy: short on highest funding (Bybit) + long on lowest funding (OKX)-# Net carry = 0.025% - 0.015% = 0.010% per 8h-# Annualized: 0.010% × 3 × 365 = 10.95%+# Net carry = 0.00025 - 0.00015 = 0.00010 per 8h (0.010%)+# Annualized: 0.00010 × 3 × 365 = 0.1095 → 10.95%# Risk: execution cost + potential for rates to converge/flip```@@ -164,15 +170,15 @@ exchange_rates = {```python# Combined OI + Funding signaldef oi_funding_matrix(oi_change_24h_pct, funding_rate):- if oi_change_24h_pct > 5 and funding_rate > 0.03:+ if oi_change_24h_pct > 5 and funding_rate > 0.0003: # funding > +0.03%return "leveraged_long_buildup" # High risk, squeeze potential- elif oi_change_24h_pct > 5 and funding_rate < -0.01:+ elif oi_change_24h_pct > 5 and funding_rate < -0.0001: # funding < -0.01%return "leveraged_short_buildup" # Short squeeze potentialelif oi_change_24h_pct < -5 and funding_rate > 0:return "long_liquidation" # Forced long closingelif oi_change_24h_pct < -5 and funding_rate < 0:return "short_liquidation" # Forced short closing- elif abs(oi_change_24h_pct) < 2 and abs(funding_rate) < 0.005:+ elif abs(oi_change_24h_pct) < 2 and abs(funding_rate) < 0.00005: # |funding| < 0.005%return "quiet_market" # Low conviction, waitelse:return "mixed"