QuantRocket: Python research, backtesting and customer-run trading

The company describes its professional audience as portfolio managers and hedge funds, alongside technically capable individual traders, especially software developers. The practical fit is a team willing to maintain Python code and its deployment. Purchasing the platform does not buy an outsourced quantitative research team or a managed investment strategy.[2]

MoonshotML builds on Moonshot’s Pandas conventions. The documented live workflow reruns a strategy on current historical data to generate a batch of orders. That is a useful distinction when choosing between batch signal production and a strategy that depends on continuously reacting to individual market events.[4]

The data catalogue distinguishes included datasets, separately purchased datasets, third-party API subscriptions and custom imports. Paid software includes US stock and ETF daily and one-minute history from 2007, with delisted securities and nightly updates. Live subscriptions are a separate layer: Interactive Brokers market data is subscribed through the brokerage account. An included historical bundle therefore does not imply an included real-time feed.[5]

The Sharadar integration spans US fundamentals, insider holdings, institutional ownership, SEC Form 8-K events, historical S&P 500 membership and end-of-day prices. These datasets cover different historical periods. A research design should choose the actual field and history needed rather than assume that every series starts on the same date or represents the same publication timestamp.[6]

QuantRocket’s international proposition uses Interactive Brokers access and multiple data providers, with handling for currencies and market timezones. That is relevant to teams applying a common strategy across countries. The published market totals describe the vendor’s broad reach; actual trading still depends on the selected instrument, broker account and data subscription.[7]

The API reference documents Python and command-line operations, including account balances and positions. Balance queries normally read a database refreshed each minute and offer a forced refresh; outputs can be CSV or JSON. These details matter when another monitoring system consumes the results: a stored account snapshot and a fresh broker request are different data paths.[16]

Learning and Evaluation is publicly offered without charge or a credit card, with sample data, but excludes both live and paper trading. Full Access prices require login or account creation and selection of a licence. The public page offers monthly and annual billing and displays fields for user count, licence and maximum account size. No current paid amount is visible without that selection.[8]

The account guidance says upgrades take effect immediately with a credit for unused time, while cancellation leaves access active until the current billing period ends. Lower plan limits and annual-to-monthly changes require contacting the company and take effect at period end. The account page provides a preview of the credit and amount due before an upgrade is confirmed.[9]

The terms provide automatic renewal and no refund for ending a subscription before its period expires. Professional licensing counts people accessing outputs or data as users, as well as those operating the software directly.[10]

Installation routes cover Windows, macOS, Linux and customer-selected cloud infrastructure. QuantRocket recommends beginning locally because it is simpler than cloud deployment. The software is installed into the customer’s environment; choosing a cloud guide does not turn it into a vendor-operated hosting service. The operating owner must still decide how the deployed system will be maintained.[11]

The Mac walkthrough uses a Docker Compose definition to download images and create the application’s containers, then opens JupyterLab in a browser. This gives a concrete installation sequence and a built-in Quickstart. The walkthrough’s installation estimate is not a promise that data collection, strategy validation or live trading will be ready within the same interval.[13]

The operating guide describes IB Gateway credentials stored within the deployment and mobile two-factor authentication for live accounts. Weekly Sunday restarts require authentication; a fresh login may require it as well. It also documents local exports and Amazon S3 database backup/restore, including service restarts after some restores. These recurring responsibilities should be assigned before leaving a strategy unattended.[14]

Basic support uses a public forum for software bugs, problems and pointers beyond the documentation. It excludes writing or debugging the customer’s strategy, teaching novice programming and other extensive assistance. The stated typical response is within one business day, rather than a contractual restoration commitment. Private forum messages are available; paid email retainers are subject to availability, particularly when ongoing support demand exceeds occasional questions.[15]

The agreement identifies a software licence for personal nonprofessional use or internal professional business use. It prohibits sublicensing and use on behalf of third parties. Nonprofessional installations are limited to two; professional installations to licensed users plus one. Broker execution is supplied by third parties. Data redistribution is restricted, and included data cannot continue to be used after the software licence ends.[10]

QuantRocket’s published materials describe research and technology rather than an offer of investment advice. A brokerage or fund should consequently distinguish its own regulated activity and client obligations from the software subscription. The vendor’s professional licence is a commercial usage classification, not a financial-services authorisation.[2]

QuantRocket LLC publishes an address in Asheville, North Carolina. Brian Stanley is its founder and principal developer; the company history dates launch to early 2018 and describes reliance on automation, outsourcing and partnerships. That account provides operating context without establishing the current size of a support or engineering team.[2]

The public update history’s leading release is 2.11.0, dated 28 January 2025, adding features including CFD support and fee modelling. Its update process downloads a new Compose file and redeploys containers; individual services can have different patch versions. This is evidence of the published release baseline, not a commitment to a particular future update cadence.[17]

Operator

QuantRocket LLC

Service scope

Python research, historical data, backtesting and broker-connected strategy execution.

Customer types

Portfolio managers; Hedge funds; Technically experienced individual traders

Delivery

On premises; API

Pricing publication

Public and quoted pricing

Headquarters

Asheville, North Carolina, United States (published company address)

Platforms

QuantRocket; Moonshot; MoonshotML; Zipline; JupyterLab; Pipeline; Alphalens

Integrations

Interactive Brokers; Alpaca; Sharadar; Python API; Command-line interface

Pricing scope

Free evaluation; paid monthly or annual prices require account and licence selection. Data and infrastructure costs vary.

Implementation

Customer-run Docker deployment on local hardware or selected cloud infrastructure.

Support

Public support forum and private messages; paid email retainers subject to availability.

Permissions and regulatory scope

Software and research tools; orders execute through third-party brokers under separate account permissions.

Provider metadata checked 21 Oct 2026Source 1Source 10Source 2Source 11Source 8Source 15Source 4Source 3Source 13Source 18Source 6Source 16

The code library supplies examples that can be cloned into a deployment: introductory Moonshot and Zipline strategies, Pipeline screening, MoonshotML walk-forward work and adapted quantitative-finance lectures. Several use free sample data. The FX Business Day example separately requires an Interactive Brokers account even where the backtest needs no paid QuantRocket subscription. Tutorial availability therefore needs to be read alongside its data and account prerequisites.[18]

Zipline, Pipeline, Alphalens and Pyfolio are related open-source components incorporated into the research workflow. The Zipline product page connects research, historical testing and broker execution, but those stages still have different inputs and conditions. A reproducible deployment needs its strategy, data version, account allocation and execution settings recorded together.[3]