DCA-Bot Optimizer

Back-test & parameter-search engine for 3Commas-style “Deal-Start / DCA / Trailing-TP” bots.
• Indicator-based smart entries (Bollinger %B + RSI) or fixed-delay stupid re-entries
• Fast Numba core (--jobs 0 = all CPU cores)
• Optuna triple search: BEST (profit), SAFE (drawdown), FAST (deal frequency)
• PNG equity curves, JSON summaries, auto-panel comparison
• Works on Windows / Linux / macOS, Python 3.8+

---

1 Features

Module	What it does
loader.py	Downloads / caches Binance klines (1‑min)
strategies/	dca_ts.py = pure‑Python back‑tester dca_ts_numba.py = Numba‑JIT back‑tester (20–50× faster)
simulator.py	ROI, APR, drawdown, trade metrics (handles open deals)
plotting.py	Down‑sampled 3 000 × 4 000 px equity + panel PNGs
optuna_search.py	Three Optuna studies with duplicate‑guard & DB storage
scripts/optuna.py	CLI wrapper: runs search → four showcase back‑tests
run_coins.ps1	Example PowerShell batch to optimise many symbols

Entry modes

• Smart / signal (default)
  Bollinger %B (20, 2) crosses up 0 and previous 3‑min RSI‑7 < 30
• Stupid / delay
  --use-sig 0 --reopen-sec N → open a new deal N seconds after the last closed.

---

2 Installation

git clone https://github.com/GiladR1979/dca-bot-optimizer.git
cd dca-bot-optimizer
python -m venv .venv && source .venv/bin/activate       # Windows: .venv\Scripts\activate
pip install -r requirements.txt

Dependencies: numpy, pandas, numba, ta, matplotlib, optuna, requests

---

3 Quick start

# optimise SOLUSDT on 1‑min candles, 2021‑01‑01 → 2025‑03‑01
python -m dca_bot.scripts.optuna SOLUSDT 2021-01-01 2025-03-01 \
       --trials 400  --jobs 0            \  # 400 trials, all CPU cores
       --storage sqlite:///dca.sqlite    \  # resume across runs
       --use-sig 1                       \  # indicator mode
       -v                                   # verbose logs

Outputs → results/

SOLUSDT_best.png     SOLUSDT_safe.png     SOLUSDT_fast.png
SOLUSDT_default.png  SOLUSDT_triple.png   SOLUSDT_opt_summary.json

---

4 Command‑line flags

Flag	Default	Meaning
--trials	200	trials per study
--jobs	0	Optuna workers (0 = all cores)
--storage	sqlite:///dca.sqlite	RDB url (none → in‑memory)
--use-sig	1	1 = use indicator, 0 = ignore
--reopen-sec	60	delay when --use-sig 0
-v	off	verbose logging

---

5 Batch run

Edit run_coins.ps1 and then:

.
un_coins.ps1

---

6 Metrics

Key	Meaning
total_pl	realised profit on $1 000 base
annual_pct	APR (linear) = ROI / years
max_drawdown_pct	worst % below $1 000
avg_deal_min	average closed‑deal duration

---

7 Contributing

• Fork → feature branch → PR against dev
• Run black + flake8 before push
• Include test or before/after screenshot

---

8 License

MIT © 2025 Gilad R.