← All Projects
Live & Running

Algorithmic Trading Fleet

A fleet of autonomous trading bots operating 24/7 across multiple platforms, executing forex strategies with automated risk management.

The Problem

Manual forex trading is emotionally exhausting and time-intensive. Profitable strategies exist, but executing them consistently — at scale, across time zones, without fatigue — requires automation. Prop firm challenges demand strict drawdown limits that no human can monitor 24/7.

What I Built

A fleet of 5+ trading bots, each running unique strategies (supply/demand zone detection, London breakout, momentum) across multiple broker platforms. A VPS runs the signal generator and Expert Advisors around the clock. A Python backtesting pipeline validates strategies across 26+ forex pairs before going live.

Development Pipeline — Concept to Live

01DiscoverPython bulk backtesting · 26+ pairs · Fast strategy screening
02ValidateMT5 Strategy Tester on VPS · Real broker spreads · Confirms Python results
03Forward TestDeploy EA to demo account · 1–2 weeks in live market · No money at risk
04Deploy to LiveReal or prop firm challenge account · Config resolver loads credentials from .env
05MonitorDiscord alerts · Trade logs · Daily drawdown protection · 24/7 VPS

Tools & Stack

PythonMQL5MT5 APIcTraderTradeLockerVPS (Windows)dotenvDiscord WebhooksGit / GitHubCron Jobs

Key Results

  • 6 bots built: v119 SD Zones, Mochi4 SDZ, Taro1 FVG, Boba1 Gold, and two Python-based signal generators
  • v119 EA: Top performer — deployed to funded prop firm accounts
  • Mochi4 SDZ: Profitable and undergoing symbol-level optimization
  • Full diagnostic framework: 4-tier system (Deploy → Monitor → Investigate → Kill) with structured autopsy process
  • Security overhaul — full git history scrub, config resolver pattern for credentials

Challenges & Lessons

The hardest lesson was security. Early on, broker credentials were committed to git in plaintext — a dangerous mistake. I performed a full git history scrub using git-filter-repo, then rebuilt the credential architecture with environment variables and a custom config resolver. Now every bot loads secrets at runtime from .env, and account configs only contain $ENV_VAR references. Another key lesson: backtesting in Python gives fast discovery, but you must validate with the broker's own strategy tester to account for real spreads and slippage.

← PreviousImmortal Creator