Andrew Rose  /  automated trading system  /  2026

Three bots.
One server. Fully automated.

A self-hosted algorithmic trading stack built over two days — each bot with its own strategy, isolated Alpaca paper account, and real-time Telegram monitoring. Target: real money by day 14.

All systems active
Paper trading since Jun 13, 2026 DigitalOcean · Ubuntu 24 · root@134.122.122.5

The bots

DAYBOT-V2 Paper · PAPER1
The Patient Hunter
VWAP reclaim · Intraday · EOD flat
2.64
Backtest PF
52%
Win rate
0.31%
Max DD

Waits for price to dip below VWAP, then reclaim it on volume. Enters only when the market regime agrees. Sells half at +1R, stops to breakeven, trails the rest. Comfortable sitting out entire sessions — many silent days are correct.

Universe~100 liquid symbols
Risk per trade1% equity
Daily halt−6% equity
Max positions5 concurrent
Time stop45 min
EOD flatten3:50 pm ET
Backtest Jan–Jun 2026: 25 trades · avg win $135 · avg loss $55 · PF 2.64 · max DD 0.31%
SWINGBOT-V2 Paper · PAPER2
The Trend Follower
150-day SMA gate · 3× ETF sleeves · Daily
35.1%
CAGR (15yr)
$90k
Total deployed
3
Sleeves

Holds leveraged ETFs when their underlying index is above its 150-day SMA. Steps to cash when it breaks below. Evaluates once daily at 4:15 pm ET, executes at next open. No intraday noise — pure trend.

TQQQ · QQQ 150d gateCash
SOXL · SOXX 150d gateCash
SPXL · SPY 150d gateIN since Jun 11
15-year backtest (2011–2026): TQQQ sleeve 35.1% CAGR · max DD −58.7% · validated walk-forward
AGGROBOT-V1 Paper · PAPER3
The Momentum Chaser
Gap-and-go · HOD continuation · VWAP reclaim
40+
Symbol universe
30
Max trades/day
Wk 1
Calibration

Rides high-momentum intraday moves in leveraged ETFs and high-beta names. Dynamic movers screener runs every 2–3 min. Exits aggressively — 1/3 at +1R, 1/3 at +2R, trail the rest. Anti-parabolic veto prevents chasing vertical tops.

Core universeSOXL TQQQ SPXL NVDA AMD TSLA + more
Tier A (≥$5, avg vol ≥3M)$15k max
Tier B (penny, RVOL ≥5×)$3k max
Risk per trade1% equity · $1k
Daily halt−$5,000
Time stop20 min
Status: Live paper · week 1 calibration · target 50+ trades with PF > 1.3 for real-money approval

Infrastructure

Server

ProviderDigitalOcean
OSUbuntu 24 LTS
RAM1 GB + 2 GB swap
Disk5.9 GB / 24 GB used
BrokerAlpaca (3 isolated paper accounts)
DataAlpaca IEX + yfinance fallback
LanguagePython 3 · systemd services

Services

daybot-v2.serviceactive
swingbot-v2.serviceactive
aggrobot-v1.serviceactive
dip-alert.serviceactive
trading-dashboard.serviceactive
DashboardFlask · SSH tunnel · localhost:8050
Codebase~15,000 lines Python
Watchdog
Cron every 5 min. Detects all 4 services going stale. Telegram alert with 30-min cooldown. Market-hours gated.
Telegram alerts
Morning brief, hourly digest, EOD postmortem. Immediate alerts for halts, crashes, orphan positions.
Nightly backup
All journals, state files, and bot code tarballed nightly. 14-day retention. Pushed as Telegram document for off-site copy.
Weekly drift check
Fridays 4:45 pm ET. Computes 7-day PF, win rate, drawdown across all bots. Alarms if performance drifts below floor.

Gates to real money

AGGROBOT-V1
Fastest path — target day 14
  • 50+ paper trades with PF > 1.3
  • No single day loss > $3k
  • EOD flatten confirmed clean every session
  • Start small — $10–25k real, not $100k
Target: June 27, 2026
DAYBOT-V2
Methodical — needs live sample
  • 100+ live trades (months at 0.3/day)
  • PF > 1.5 in live journal AND backtest
  • Max DD < 10% live
  • Alpaca Algo Trader Plus for SIP data
  • $25k minimum (PDT rule)
Target: Fall 2026
SWINGBOT-V2
No rush — let it cycle
  • 30+ days paper across all 3 sleeves
  • Each sleeve IN and OUT at least once
  • No SIP data needed (daily bars only)
Target: Post-July 2026
Andrew Rose
Videographer · Creative Director · Algo trader in progress
Built in 2 days · June 13–14, 2026
3 bots · 5 services · ~15,000 lines of Python
Paper trading → Real money