My AI Trading Bot Treats 4 Coins Equally. Today I Found One Has Been Dragging the Portfolio Down.

My AI Trading Bot Treats 4 Coins Equally. Today I Found One Has Been Dragging the Portfolio Down for 2 Months.

I almost made my AI trading system worse today. The data stopped me.

The Near-Mistake

Started the day planning to raise a risk-reward threshold. The single-day analysis looked compelling: signals with 1.5:1 R:R were negative expected value at my 38% win rate.

Then I queried all 126 production trades.

The band I was about to cut: 40% win rate, + profit across 30 trades. Sprint cancelled. Threshold stays.

Per-Coin Analysis

With the production database open, I asked a question I should have asked months ago: how does each coin perform individually?

BTC: 45% win rate, + across 42 trades ETH: 39% win rate, + across 36 trades XRP: 44% win rate, + across 16 trades SOL: 34% win rate, + across 32 trades

SOL has taken almost as many trades as BTC but generated one-sixth the profit. SOL LONGs specifically: 31% win rate, net -.

The Root Cause

SOL higher volatility means wider stops (more expensive losses) but the same win rate. Winners need 24 hours to reach take-profit. Losers stop out in 8 hours. The system is right about direction but SOL intraday noise kills the trade before it works.

The Fix

Sprint 125 adds per-coin performance tracking and targeted SOL restrictions. Not removing SOL entirely, just requiring stronger conviction.

Stats

+,000 (+33%) on K starting capital. 126 trades.

#buildinpublic #algotrading #solofounder