Skip to content

Quickstart

1. Local CSV to audited backtest

import asrquant as asr

lab = asr.open_lab("prices.csv", date_column="Date")
result = lab.backtest("sma", fast=20, slow=100, costs_bps=5)
print(result.metrics)
result.plot("dashboard")
result.report("report.html")

2. Remote data

lab = QuantLab.from_provider("yahoo", ["SPY", "QQQ"], start="2018-01-01")

lab = QuantLab.from_provider("binance", ["BTCUSDT", "ETHUSDT"], interval="1h", limit=1000)

Alpha Vantage and FRED require API keys. Yahoo requires the data extra.

3. Monte Carlo

sim = lab.monte_carlo("gbm", drift=0.05, volatility=0.20, paths=10_000, random_state=7)
print(sim.summary)
sim.plot("fan")

mc = lab.option(
    "monte_carlo", strike=100, maturity=1, rate=0.03,
    volatility=0.20, paths=100_000, antithetic=True, random_state=7,
)
print(mc.summary)

4. Closed-form and tree pricing

bsm = lab.option("black_scholes", strike=100, maturity=1, rate=0.03, volatility=0.20)
bach = lab.option("bachelier", strike=100, maturity=1, rate=0.03, normal_volatility=10)
tree = lab.option("crr", strike=100, maturity=1, rate=0.03, volatility=0.20, steps=1000)

5. Martingale diagnostics

diagnostic = lab.martingale_test(rate=0.03, lags=10)
print(diagnostic.statistics)
diagnostic.plot()

6. Regression

fit = lab.regress("SPY", ["QQQ", "TLT"], covariance="HAC", maxlags=5)
print(fit.coefficients)
fit.plot("residuals")

7. Walk-forward ML

import asrquant as asr

X = lab.ml_features("SPY").shift(1)
y = asr.forward_target(lab.prices["SPY"], horizon=5)
wf = lab.ml_walk_forward("ridge", X, y, train_size=504, test_size=63, gap=5, model_params={"alpha": 1.0})
print(wf.aggregate_metrics)

Input contract

  • index: unique, sortable timestamps;
  • values: finite positive prices for QuantLab;
  • columns: unique asset identifiers;
  • missing observations: rejected by default, or explicitly dropped/forward-filled;
  • target weights: same index and a subset of the same asset columns;
  • execution: one-bar delay by default.

Built-in strategies

buy_hold, sma, momentum, mean_reversion, vol_target, breakout, bollinger, rsi, and pairs.

Built-in strategies are examples and reusable primitives, not investment recommendations.

8. N-dimensional parameter surface

surface = lab.parameter_surface(
    experiment,
    {
        "gamma": [0.5, 1, 2, 4],
        "cost_bps": [0, 5, 10],
        "hedge_every": [1, 5, 20],
        "volatility": [0.15, 0.30],
    },
    x="gamma",
    y="cost_bps",
    animate_by=["hedge_every", "volatility"],
    metric="metrics.utility",
)
surface.save_animation("parameter_landscape.html")

The HTML output includes a frame slider. Use .gif or .mp4 for video-style exports, and surface.best("max") to retrieve the best finite point.