Interactive backtesting walkthrough

From strategy definition to walk-forward validation, a complete Python backtesting workflow with Manifold-BT.

getting_started.ipynbPython 3 (manifoldbt)

Code

In [ ]:

import manifoldbt as bt

from manifoldbt.indicators import close, sma, rsi, volume

In [ ]:

# Ingest data from Binance (parallel fetch, Arrow IPC storage)

store = bt.ingest(

provider="binance",

symbols=[("BTCUSDT", 1), ("ETHUSDT", 2), ("SOLUSDT", 3)],

start="2020-01-01T00:00:00Z",

end="2026-01-01T00:00:00Z",

interval="1m",

)

store.list_symbols()

In [ ]:

# Define a momentum strategy with the expression DSL

fast = sma(close, 20)

slow = sma(close, 50)

momentum = (fast - slow) / slow

signal = bt.when(

(momentum > 0.01) & (rsi(close, 14) < 70),

1.0,

0.0,

)

strategy = (

bt.Strategy.create("momentum_crossover")

.signal("entry", signal)

.size(signal * 0.25)

)

In [ ]:

# Configure: Binance perps, realistic execution

config = bt.BacktestConfig(

universe=[1],

initial_capital=10_000,

time_range_start=bt.date_to_ns("2020-01-01"),

time_range_end=bt.date_to_ns("2025-01-01"),

bar_interval={"Minutes": 1},

fees=bt.FeeConfig.binance_perps(),

execution=bt.ExecutionConfig(

signal_delay=1,

allow_short=True,

slippage={"VolumeImpact": {"coefficient": 0.1}},

warmup_bars=50,

)

In [ ]:

# Run backtest

result = bt.run(strategy, config, store)

result.summary()

In [ ]:

# Generate tearsheet

bt.plot.tearsheet(result)

In [ ]:

# 2D parameter sweep, EMA crossover t-stat(alpha)

fast_values = list(range(5, 1000, 3))

slow_values = list(range(10, 5000, 3))

batch = bt.run_sweep_lite(

strategy,

{"fast": fast_values, "slow": slow_values},

config,

store,

)

bt.plot.heatmap_2d({

"x_param": "fast",

"y_param": "slow",

"x_values": fast_values,

"y_values": slow_values,

"metric": "t-stat(alpha)",

"metric_grid": metric_grid,

})

In [ ]:

# Same data as 3D surface

bt.plot.surface_3d({

"x_param": "fast",

"y_param": "slow",

"x_values": fast_values,

"y_values": slow_values,

"metric": "t-stat(alpha)",

"metric_grid": metric_grid,

})

In [ ]:

# Walk-forward validation

wf = bt.run_walk_forward(

strategy,

wf_config={

"method": "Anchored",

"n_splits": 5,

"train_ratio": 0.7,

"optimize_metric": "sharpe",

"param_grid": {

"sma_fast": [10, 15, 20, 25],

"sma_slow": [40, 50, 60, 70],

config=config,

store=store,

)

bt.plot.walk_forward(

wf,

mode="stitched",

full_result=result,

)

In [ ]:

# Monte Carlo robustness check

mc = bt.plot.monte_carlo(

result,

n_paths=10_000,

method="bootstrap",

)

That's the full loop. Strategy → backtest → analysis → validation, all under 5 seconds.

Back to home Read the docs →