driftbrake

A chain-agnostic simulation and drift-based halt guard for algorithmic trading strategies.

driftbrake pre-flight simulates transactions against local chain state, tracks simulated versus realized profit upon on-chain confirmation, and halts execution the moment realized profit drifts from prediction. It prevents simulation drift, revert bursts, and silent capital bleed across algorithmic trading strategies.

Installation

pip install driftbrake

Quickstart

from driftbrake import ReconcileHistory, ReconcilePolicy, StrategyHaltedError

# Configure institutional multi-guard policy
policy = ReconcilePolicy(
    fast_window=3,
    fast_threshold=0.50,
    slow_window=20,
    slow_threshold=0.70,
    max_consecutive_reverts=3,
    max_revert_gas_budget=500_000,
    volume_weighted_window=15,
    volume_weighted_threshold=0.65,
)
history = ReconcileHistory()

# In your execution loop:
# Append predicted vs realized profit pairs (in wei or native units)
history.append(predicted_profit=1_000_000, realized_profit=950_000)

# Evaluate halt decision:
decision = policy.evaluate(history)
if decision.is_halted:
    raise StrategyHaltedError(f"Strategy halted: {decision.reason}")

Vectorized Backtesting Sweep

Run parameter sweeps over millions of historical trade rows in microsecond compiled Rust time:

from driftbrake import run_sweep_raw

# pairs: list of (predicted_profit, realized_profit)
results = run_sweep_raw(
    pairs,
    fast_windows=[2, 3, 4],
    fast_thresholds=[0.40, 0.50, 0.60],
    slow_windows=[15, 20, 25],
    slow_thresholds=[0.65, 0.70, 0.75],
)

for r in results:
    if not r.halted:
        print(f"Optimal parameters: fast={r.fast_window}/{r.fast_threshold}, slow={r.slow_window}/{r.slow_threshold}")

Documentation & Whitepaper

• Whitepaper & Mathematical Invariants: xtley001.github.io/driftbrake/whitepaper.html
• Architecture & System Design: xtley001.github.io/driftbrake/ARCHITECTURE.html
• GitHub Repository: github.com/Xtley001/driftbrake

License

Released under the MIT License.