Skip to main content

Algorithmic Trading & Strategy Testing

From Idea to Live Execution — One Stack

Tick-level backtesting, walk-forward optimisation, and a low-latency execution engine. The same code path from research notebook to production fills.

Request PricingTry Free Demo
strategy.py — backtest
from brokz.algo import Strategy, Order

class MeanReversion(Strategy):
    lookback = 60
    z = 2.0

    def on_bar(self, bar):
        mu, sd = self.window.stats()
        if bar.close < mu - self.z*sd:
            self.buy(qty=10)
        elif bar.close > mu + self.z*sd:
            self.sell(qty=10)

# backtest 2019-2024
result = run(MeanReversion, "EURUSD")
Backtest · EURUSD · 1m+18.4%
Sharpe2.31
Max DD-4.2%
Trades1,284
Win %58.7
Slippage0.4 bp
Calmar4.4

What Is It

A Quant Stack, Not a Trading Bot Marketplace

A complete infrastructure stack for systematic trading — data ingestion, backtesting, optimisation, and live execution. Built for desks that run their own alpha and need a platform that won't lie to them.

No fake fills. No look-ahead bias. No "EA" black boxes. Same code from research to production, deterministic replay, and execution that respects queue position.

Tick-levelBacktest resolution
Sub-msHot path latency
1 codebaseResearch → live
Multi-venueFIX / REST / WS

Pipeline

Data → Backtest → Optimise → Live

Four stages, one platform, no glue code between them.

01

Data Ingestion

Tick & OHLCV pipelines from your liquidity providers, exchanges, or vendors. Normalised, deduped, time-synced.

02

Backtest

Event-driven engine with realistic fills, slippage, and latency models. No look-ahead, no survivorship bias.

03

Optimise

Walk-forward, Monte Carlo, parameter grids, genetic and Bayesian search. Cluster-parallel runs.

04

Deploy Live

Same strategy code, same fill model. Promote to live with one config switch. No "rewriting for prod".

Features

Built for Real Money

Tick-level Backtesting

Sub-second resolution with bid/ask bookkeeping. Full L1 reconstruction; L2 optional.

Slippage & Latency Models

Per-venue latency, partial fills, queue position. Plug your own models if you have them.

Walk-Forward Analysis

In-sample / out-of-sample windows, rolling re-optimisation. Surface overfitting before capital does.

Live Execution Engine

Co-located on your venue when needed. FIX / REST / WebSocket adapters. Risk-checked at every order.

Strategy Library

Mean-reversion, trend, stat-arb, market-making templates. Starting point, not a black box.

Performance Dashboards

Sharpe, Sortino, Calmar, drawdown, regime breakdown. Export to your data warehouse.

Architecture

Technical Architecture

Python where it helps research velocity, Rust / C++ where microseconds matter. Deterministic replay for compliance and debugging. Cloud-agnostic.

  1. 1
    Languages: Python (research), Rust / C++ (low-latency execution)
  2. 2
    Data: TimescaleDB / ClickHouse for tick storage, Parquet for archives
  3. 3
    Backtest: Event-driven engine, deterministic replay, distributed runner
  4. 4
    Execution: FIX 4.4 / 5.0, REST, WebSocket. Risk gateway with kill-switch
  5. 5
    Infra: Containerised, deployable on AWS / GCP / co-lo. Your infrastructure
  6. 6
    Observability: Per-strategy P&L, fill quality, slippage attribution, alerting

Pricing

Engagement Models

Fixed-price platform delivery, or ongoing retainer. Pricing depends on data volume, venue count, and latency targets.

Request a detailed quote

FAQ

Frequently Asked Questions

Get Started

Ready to put your alpha on real rails?

Tell us about your strategies, venues, and latency targets. We respond in one business day.

Start a Conversation