mirror of
https://github.com/manifoldbt/manifoldbt.git
synced 2026-11-02 07:56:11 +00:00
93 lines
3.0 KiB
Python
93 lines
3.0 KiB
Python
"""CSV Import -- load your own OHLCV data from a CSV file.
|
|
|
|
Demonstrates:
|
|
- bt.import_csv() -- auto-detects standard / MetaTrader 4 / MetaTrader 5
|
|
- Backtesting on the imported data, exactly like a built-in connector
|
|
- Free on all tiers (no Pro license required)
|
|
|
|
The standard format is a header row + `timestamp,open,high,low,close,volume`
|
|
where timestamp is Unix milliseconds. MT4/MT5 exports are auto-detected.
|
|
|
|
Data: synthetic (seed 18) — a sample generated by this file, reproducible
|
|
|
|
Usage:
|
|
python examples/18_csv_import.py
|
|
"""
|
|
import os
|
|
import tempfile
|
|
|
|
import numpy as np
|
|
|
|
import manifoldbt as mbt
|
|
from manifoldbt.indicators import close, ema
|
|
from manifoldbt.helpers import time_range, Interval
|
|
|
|
# -- 1. A sample CSV ----------------------------------------------------------
|
|
# In practice you'd point `import_csv` straight at your own file. Here we
|
|
# synthesize a small one so the example runs out of the box.
|
|
tmp = tempfile.mkdtemp()
|
|
csv_path = os.path.join(tmp, "SAMPLE_1m.csv")
|
|
|
|
# 120 days of 1-minute bars. The span is the point: annualised metrics (CAGR,
|
|
# Calmar) divide by the length of the window, so a fixture covering two days
|
|
# reports a growth rate with fifty digits in it and teaches the reader nothing
|
|
# except that nobody read the output.
|
|
N = 120 * 1440
|
|
rng = np.random.default_rng(18)
|
|
|
|
steps = rng.normal(1.4e-6, 4.0e-4, N) # ~+20% over the window
|
|
px = 100.0 * np.exp(np.cumsum(steps))
|
|
o = px
|
|
c = np.roll(px, -1)
|
|
c[-1] = px[-1]
|
|
amp = np.abs(rng.normal(0.0, 2.0e-4, N))
|
|
hi = np.maximum(o, c) * (1 + amp)
|
|
lo = np.minimum(o, c) * (1 - amp)
|
|
vol = rng.uniform(1_000, 5_000, N)
|
|
|
|
base_ms = 1_704_067_200_000 # 2024-01-01 00:00 UTC
|
|
with open(csv_path, "w") as f:
|
|
f.write("timestamp,open,high,low,close,volume\n")
|
|
f.write("\n".join(
|
|
f"{base_ms + i * 60_000},{o[i]:.4f},{hi[i]:.4f},{lo[i]:.4f},{c[i]:.4f},{vol[i]:.0f}"
|
|
for i in range(N)
|
|
))
|
|
f.write("\n")
|
|
|
|
# -- 2. Import into the store (free, all tiers) -------------------------------
|
|
store = mbt.import_csv(
|
|
csv_path,
|
|
symbol="SAMPLE",
|
|
symbol_id=1,
|
|
interval="1m",
|
|
data_root=os.path.join(tmp, "data"),
|
|
metadata_db=os.path.join(tmp, "meta.sqlite"),
|
|
asset_class="crypto_spot",
|
|
)
|
|
print("Imported:", store.list_symbols())
|
|
|
|
# -- 3. Backtest on it like any other data ------------------------------------
|
|
strategy = (
|
|
mbt.Strategy.create("ema_cross")
|
|
.signal("fast", ema(close, 10))
|
|
.signal("slow", ema(close, 30))
|
|
.size(mbt.when(ema(close, 10) > ema(close, 30), 0.5, 0.0))
|
|
.describe("EMA(10/30) crossover on CSV-imported data")
|
|
)
|
|
|
|
start, end = time_range("2024-01-01", "2024-05-01")
|
|
config = mbt.BacktestConfig(
|
|
universe=[1],
|
|
time_range_start=start,
|
|
time_range_end=end,
|
|
bar_interval=Interval.minutes(1),
|
|
initial_capital=10_000,
|
|
warmup_bars=30,
|
|
)
|
|
|
|
if __name__ == "__main__":
|
|
result = mbt.run(strategy, config, store)
|
|
print("The series carries a mild upward drift, so the figures below")
|
|
print("describe the fixture. The mechanism on show is the import.\n")
|
|
print(result.summary())
|