Trade history tracking, sync, and performance analytics
56
70%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
Fix and improve this skill with Tessl
tessl review fix ./src/skills/bundled/history/SKILL.mdFetch, sync, and analyze trade history from Polymarket and Kalshi with detailed performance metrics.
/history fetch # Fetch all trades from APIs
/history fetch poly # Fetch Polymarket only
/history fetch --from 2024-01-01 # From specific date
/history sync # Sync to local database/history list # Recent trades
/history list --limit 50 # Last 50 trades
/history list --platform poly # Polymarket only
/history list --market <id> # Specific market/history stats # Overall statistics
/history stats --period 30d # Last 30 days
/history stats --platform kalshi # Platform-specific/history daily-pnl # Daily P&L
/history weekly-pnl # Weekly P&L
/history monthly-pnl # Monthly P&L
/history by-market # P&L by market category/history export # Export to CSV
/history export --format json # Export as JSON
/history export --from 2024-01-01 # Date range/history filter --side buy # Only buys
/history filter --pnl positive # Only winners
/history filter --pnl negative # Only losers
/history filter --min-size 100 # Min $100 tradesimport { createTradeHistoryService } from 'clodds/history';
const history = createTradeHistoryService({
polymarket: {
apiKey: process.env.POLY_API_KEY,
address: process.env.POLY_ADDRESS,
},
kalshi: {
apiKey: process.env.KALSHI_API_KEY,
},
// Local storage
dbPath: './trade-history.db',
});// Fetch all trades from exchange APIs
const trades = await history.fetchTrades({
platforms: ['polymarket', 'kalshi'],
from: '2024-01-01',
});
console.log(`Fetched ${trades.length} trades`);
// Fetch from specific platform
const polyTrades = await history.fetchTrades({
platforms: ['polymarket'],
limit: 100,
});// Sync fetched trades to local database
await history.syncToDatabase();
console.log('Trades synced to database');// Get trades from local storage
const trades = await history.getTrades({
platform: 'polymarket',
from: '2024-01-01',
to: '2024-12-31',
limit: 100,
});
for (const trade of trades) {
console.log(`${trade.timestamp}: ${trade.side} ${trade.market}`);
console.log(` Size: $${trade.size}`);
console.log(` Price: ${trade.price}`);
console.log(` P&L: $${trade.pnl?.toFixed(2) || 'open'}`);
}// Get comprehensive statistics
const stats = await history.getStats({
period: '30d',
platform: 'polymarket',
});
console.log(`=== Trading Statistics (30d) ===`);
console.log(`Total trades: ${stats.totalTrades}`);
console.log(`Winning trades: ${stats.winningTrades}`);
console.log(`Losing trades: ${stats.losingTrades}`);
console.log(`Win rate: ${(stats.winRate * 100).toFixed(1)}%`);
console.log(`\nP&L:`);
console.log(` Total: $${stats.totalPnl.toLocaleString()}`);
console.log(` Gross profit: $${stats.grossProfit.toLocaleString()}`);
console.log(` Gross loss: $${stats.grossLoss.toLocaleString()}`);
console.log(` Profit factor: ${stats.profitFactor.toFixed(2)}`);
console.log(`\nTrade sizes:`);
console.log(` Average: $${stats.avgTradeSize.toFixed(2)}`);
console.log(` Largest win: $${stats.largestWin.toFixed(2)}`);
console.log(` Largest loss: $${stats.largestLoss.toFixed(2)}`);
console.log(`\nRisk metrics:`);
console.log(` Sharpe ratio: ${stats.sharpeRatio.toFixed(2)}`);
console.log(` Max drawdown: ${(stats.maxDrawdown * 100).toFixed(1)}%`);// Get daily P&L breakdown
const dailyPnl = await history.getDailyPnL({
days: 30,
platform: 'polymarket',
});
console.log('=== Daily P&L ===');
for (const day of dailyPnl) {
const sign = day.pnl >= 0 ? '+' : '';
const bar = day.pnl >= 0
? '█'.repeat(Math.min(Math.floor(day.pnl / 10), 20))
: '▓'.repeat(Math.min(Math.floor(Math.abs(day.pnl) / 10), 20));
console.log(`${day.date} | ${sign}$${day.pnl.toFixed(2).padStart(8)} | ${bar}`);
}// Get performance breakdown by market category
const byMarket = await history.getPerformanceByMarket({
period: '30d',
});
console.log('=== Performance by Market Category ===');
for (const [category, data] of Object.entries(byMarket)) {
console.log(`\n${category}:`);
console.log(` Trades: ${data.trades}`);
console.log(` Win rate: ${(data.winRate * 100).toFixed(1)}%`);
console.log(` P&L: $${data.pnl.toLocaleString()}`);
console.log(` Avg trade: $${data.avgTrade.toFixed(2)}`);
}// Export to CSV
await history.exportCsv({
path: './trades.csv',
from: '2024-01-01',
to: '2024-12-31',
columns: ['timestamp', 'platform', 'market', 'side', 'size', 'price', 'pnl'],
});
// Export to JSON
const json = await history.exportJson({
from: '2024-01-01',
});CREATE TABLE trades (
id TEXT PRIMARY KEY,
platform TEXT NOT NULL,
market_id TEXT NOT NULL,
market_question TEXT,
side TEXT NOT NULL, -- 'buy' or 'sell'
outcome TEXT, -- 'YES' or 'NO'
size REAL NOT NULL,
price REAL NOT NULL,
fee REAL DEFAULT 0,
pnl REAL,
timestamp INTEGER NOT NULL,
created_at INTEGER DEFAULT (strftime('%s', 'now'))
);
CREATE INDEX idx_trades_platform ON trades(platform);
CREATE INDEX idx_trades_timestamp ON trades(timestamp);
CREATE INDEX idx_trades_market ON trades(market_id);alsk1992/CloddsBot
c930628
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.