Every 13F since 2014, free and account-free, from a Rails app you can run yourself.
FreeOpen source
Self-hosted stock market tools
Tools that run on your own machine or server — including six that are self-hosted and not open source at all.
Last updated
Self-hosted is a deployment fact, not a licence
Most of this list is also open source, which makes it easy to assume the two words mean the same thing. Six tools here are the reason they do not: kdb+, Highcharts Stock, QuantRocket, ArcticDB, vectorbt and vectorbt PRO all run on hardware you control under terms that are commercial, source-available, or both.
ArcticDB is BUSL-1.1: each release converts to Apache-2.0 two years after it ships, so the version you are running today is source-available and the one from two years ago is open. vectorbt carries Apache-2.0 with the Commons Clause, which removes the right to sell it and with that the OSI definition. Neither is a trap — both are on the cards — but neither is what a reader who filtered for "self-hosted" is likely to have assumed.
Three reasons people end up here
The positions should not sit on somebody's server. This is the portfolio-tracker case, and it is why Ghostfolio and Wealthfolio exist at all. Note what self-hosting does not buy: both still need a market-data key, and every holding you track is a request that leaves the machine.
The data is too big to send anywhere. The tick data storage category is self-hosted almost by definition — once a year of quotes is measured in terabytes, the question stops being which vendor and starts being which disk.
The vendor is the risk. A hosted backtester that changes its pricing, its API or its mind takes your research with it. NautilusTrader, Zipline Reloaded and QuantRocket are all answers to that, at three very different levels of assembly required.
What it costs you instead
Upgrades, backups and the debugging. That is a real trade and the cards do not pretend otherwise: a self-hosted tracker is a database you now own, and a self-hosted tick store is a machine whose read throughput is your problem. Where a tool has a hosted twin — QuantConnect's cloud beside the open LEAN engine, Timescale's managed service beside the source — the card says so, because for most people the right answer is the hosted one and the point of the licence is that it does not have to stay that way.
All 58 of them
Showing 58 of 58
LLM investor personas score a ticker, deterministic code sizes the book, nothing trades.
FreeOpen source
Linkware — free in commercial products as long as the amCharts link stays on the chart.
$64/yrFree tier
Per-product, per-domain commercial licensing with a watermark until you pay.
$49/yr
Real-time OLAP cluster built for high-concurrency dashboards, not for research joins.
FreeOpen source
Apache-2.0 general-purpose charting with a candlestick series and nothing else financial.
FreeOpen source
GARCH and the rest of the volatility-model family, plus the tests you need around them.
FreeOpen source
Versioned Pandas frames written straight onto S3, with no server to run.
Free tier only
A single-instrument Python backtester — one OHLC series, one strategy, no live trading.
FreeOpen source
An event-driven Python backtester with 122 indicators, frozen since April 2023.
FreeOpen source
Python backtesting for allocation and rebalancing rules, not entries and exits.
FreeOpen source
Columnar OLAP database that keeps years of ticks on disk cheaply and scans them fast.
$66.52/moFree tierMCP
Macro series from 93 statistics offices and central banks behind one open API.
€1500/yrFree tier
Closed-source tick database with a vector language, as-of joins and streaming engines.
Free tier onlyMCP
In-process analytical SQL over Parquet tick files, with no server to run.
FreeOpen source
The maintained fork of Quantopian's empyrical — risk statistics, no plots.
FreeOpen source
Scrapes SEC, FINRA, FRED and CFTC data into your own Postgres and serves it over MCP.
$9.99/moFree tierMCP
Sessions, minutes and holidays for 69 exchanges, keyed by ISO-10383 code.
FreeOpen source
A one-call wrapper over Backtrader, dormant since 2023 and broken on PyPI.
FreeOpen source
Performance stats, drawdowns and portfolio weights from a DataFrame of prices.
FreeOpen source
Financial LLM adapters, instruction datasets and training notebooks, all MIT.
FreeOpen source
Open-source agents that turn a ticker and your own API keys into an analyst report.
FreeOpen source
Self-hosted open source portfolio tracking, or a hosted account for USD 48 a year.
FreeOpen sourceMCP
Commercial JS financial charting — 48 indicators, a navigator, a per-seat licence.
$370/yrFree tier
The community continuation of ib_insync — same API, new maintainers, TWS still required.
FreeOpen source
Line-protocol time-series database rewritten on Arrow, DataFusion and Parquet.
Free tier onlyOpen sourceMCP
The tick database trading desks have run for 25 years, queried in q rather than SQL.
$1822.92/moFree tierMCP
Apache-2.0 canvas candlesticks with 27 indicators built in and zero dependencies.
FreeOpen source
The chart and nothing else — six series types on canvas, attribution required.
FreeOpen source
One strategy class for backtest and live, plus a built-in LLM agent runtime.
$24.50/moFree tier
Thirty-seven stock-analysis tools for any MCP client, on Yahoo data and no key.
FreeOpen sourceMCP
Rust core, Python API, one strategy that backtests and trades live unchanged.
Free tier onlyOpen source
Enterprise tick capture, storage and streaming analytics, sold only by quote.
—
One Python API, REST app and MCP server over thirty other people's data feeds.
$500/yrFree tierMCP
The DataFrame indicator library whose repo is gone and whose last release is a year old.
FreeOpen source
Candlestick and OHLC traces in an MIT library the notebook already has installed.
FreeOpen source
Prices in, weights out — efficient frontier, Black-Litterman and HRP.
FreeOpen source
An institutional-shaped Python backtester whose last commit to master was June 2024.
FreeOpen source
The open-source LEAN engine, plus a hosted cloud that runs it against real brokers.
$84/moFree tierMCP
The open-source derivatives pricing library banks actually use, reachable from Python.
FreeOpen source
Zipline and Moonshot in Docker on your own hardware, wired to Interactive Brokers.
Free tier only
HTML tearsheets and about eighty risk metrics from one pandas series of returns.
FreeOpen source
Open-source time-series SQL built for tick data — ASOF JOIN, SAMPLE BY, LATEST ON.
Free tier onlyOpen sourceMCP
Twenty-six convex risk measures, four objectives, one cvxpy-backed optimiser.
FreeOpen source
WebGL and WebAssembly speed. The free build is watermarked and times out every 6 months.
$1349.66/yrFree tier
SEC filings and XBRL financials as MCP tools, under AGPL or a paid commercial licence.
FreeOpen sourceMCP
Portfolio optimisation as scikit-learn estimators — fit, predict, cross-validate.
FreeOpen source
The C indicator library everything else wraps — and pip now ships the C part with it.
FreeOpen source
Hypertables, columnar compression and continuous aggregates bolted onto plain Postgres.
$30/moFree tier
Analyst, bull, bear, risk and portfolio-manager agents argue over one ticker.
FreeOpen source
Free, closed-source and companies-only — the licence is the whole decision.
Free
Screeners, technical analysis and backtests for any MCP client. No TradingView account.
$9/moFree tierMCP
Vectorised backtesting — thousands of parameter combinations in one NumPy pass.
Free
Paid vectorbt — native Rust simulators, streaming indicators and a built-in MCP server.
$25/moMCP
A local finance research agent with 74 MCP tools, nine backtest engines and 14 brokers.
FreeOpen sourceMCP
Local-first portfolio and net-worth tracker for desktop, iOS and your own Docker host.
FreeOpen sourceMCP
The Python library everyone starts with — free, unofficial, and not something to build on.
FreeOpen source
The maintained fork of Quantopian's Zipline. Bring your own data — it ships almost none.
FreeOpen source
FAQ
Is self-hosted the same as open source?
No, and the difference costs money. Self-hosted is about where the process runs; open source is about what the licence lets you do with the code. kdb+, Highcharts Stock, QuantRocket, ArcticDB and vectorbt PRO all run on your own hardware under commercial or source-available terms. You get the deployment independence without the right to fork, and in several cases with a per-seat or per-core fee attached.
Does self-hosting a tracker mean my data never leaves the machine?
Your positions do not, which is usually the point. The prices still have to come from somewhere — a self-hosted tracker needs a market-data provider behind it, and every symbol you hold is a request to that provider. Ghostfolio and Wealthfolio both make you supply that key. What self-hosting removes is a vendor holding a standing copy of your portfolio, not every outbound request.
What does self-hosting actually cost?
For a tracker or a notebook-scale backtest, a small VPS and the hour it takes to set up. For tick storage the arithmetic changes — a year of US equity trades and quotes is terabytes, and the bill moves from the software to the disk and the machine that has to read it. The tick storage cards carry that calculation rather than a licence price, because the licence is the small number.
Which of these can run entirely offline?
The libraries and the databases — ClickHouse, QuestDB, TimescaleDB, ArcticDB, the charting libraries, and any backtesting framework once you have the bars on disk. Anything that quotes a live price cannot, by definition. Self-hosted means the software is yours to run, not that the market comes to you.