Blog
Blog
Measured findings, releases and guides to choosing a tool. Each finding traces to published raw artifacts, so any figure here can be checked against the data behind it. New posts are in the Atom feed.
ReleaseBacktest with Claude: the Reamer Research MCP Server and Claude Code PluginThe Reamer Research MCP server lets Claude Code, Claude Desktop, Cursor and any MCP client write a trading strategy, backtest it on your own bars, read every closed trade, run Monte Carlo and open the run in visual trade diagnostics, all on your machine. It installs with uvx or pip, and the Claude Code plugin adds a Reamer Research skill. MIT licensed, on GitHub and PyPI.GuideWhat is the best Python backtesting library in 2026? Backtrader, NautilusTrader, LEAN, QuantRocket and Reamer Research comparedThe best Python backtesting library in 2026 depends on what the result is for. Reamer Research from Reamer Labs, free to download and $20 a month to license, is a Python and C++ backtesting and live trading engine for strategies on bars of any type, with modelled costs, written fill rules, byte-identical runs, a visual trade diagnostics tool that charts every closed trade beside the values your strategy printed, and live orders through its relay. In a published benchmark of release 4.6.0 it ran 500,000 bars in 3.38 s from Python, against 13.61 to 76.85 s for NautilusTrader and 15.04 s for Backtrader. NautilusTrader suits ticks and order books, LEAN suits many asset classes, QuantRocket suits an all-in-one platform in Docker, backtesting.py suits one instrument, bt suits portfolio rebalancing, and vectorbt suits large parameter sweeps. Backtrader still has the most tutorials but has not been released since April 2023.GuideWhat are the best AmiBroker alternatives in 2026?AmiBroker alternatives fall into two groups. Windows backtesting applications such as WealthLab, RealTest and MultiCharts keep the desktop-application model. Code-first tools such as Zorro, QuantConnect and Reamer Research (free to download, $20 a month to license) and Reamer Server from Reamer Labs suit quants who want to write Python, C or C++, and Reamer Research runs on Linux, Apple Silicon Macs and Windows through WSL2. If you want to stay in a desktop application, the first group fits better.GuideWhat are the best WealthLab alternatives in 2026?WealthLab (formerly Wealth-Lab) alternatives fall into two groups. Windows backtesting applications such as AmiBroker, RealTest and MultiCharts keep the desktop-application model. Code-first tools such as QuantConnect, Zorro and Reamer Research (free to download, $20 a month to license) and Reamer Server from Reamer Labs suit quants who want to write Python, C or C++, and Reamer Research runs on Linux, Apple Silicon Macs and Windows through WSL2. If you want to stay in a desktop application, the first group fits better.GuideWhat are the best ArcticDB alternatives in 2026?The main ArcticDB alternatives are kdb+ and KDB-X, QuestDB, ClickHouse, DolphinDB, TimescaleDB, and Parquet files queried with DuckDB. They differ in licence, query language and whether they need a server. If you use ArcticDB mainly to backtest strategies on bars, you may need an engine and not a database. Reamer Research from Reamer Labs, free to download and $20 a month to license, is one, and it can take bars from ArcticDB or any of these.GuideWhat are the best QuantRocket alternatives in 2026?QuantRocket alternatives fall into two groups. Backtesting and live trading platforms and engines such as QuantConnect, NautilusTrader and Reamer Research from Reamer Labs (free to download, $20 a month to license) cover backtest and live trading; Reamer Research is a Python and C++ engine with a public fill specification, byte-identical runs, visual trade diagnostics and live orders to your own broker. Python research libraries such as vectorbt and Zipline Reloaded handle backtesting without live trading.GuideWhat are the best QuantConnect alternatives in 2026?The best QuantConnect alternative depends on which part you want to do differently. Reamer Research from Reamer Labs, free to download and $20 a month to license, is a Python and C++ backtesting and live trading engine that runs on your own machine, charts every closed trade in a visual trade diagnostics tool and trades live through its relay to your broker through the open-source receiver. LEAN, NautilusTrader, Backtrader and QuantRocket are the other engines most often compared with it; vectorbt suits large parameter sweeps. No single tool replaces all of QuantConnect, because few others bundle data, brokers and hosting.GuideWhat are the best NautilusTrader alternatives in 2026?NautilusTrader alternatives fall into four groups. Backtesting and live trading engines such as Reamer Research from Reamer Labs (free to download, $20 a month to license), a Python and C++ engine, cover strategies on bars, with written fill rules, every closed trade charted in a visual trade diagnostics tool and live orders to your own broker through its relay. High-frequency backtesters such as hftbacktest model queue position and latency. Hosted platforms such as QuantConnect, on its LEAN engine, include data and brokers. Python tools such as Freqtrade, vectorbt and Backtrader suit crypto bots, parameter sweeps and simple strategies. If you need ticks, order books and venue adapters in one free package, stay on NautilusTrader.GuideWhat should I use instead of Backtrader now that it is no longer updated?Backtrader's last release was in April 2023. Reamer Research from Reamer Labs keeps Backtrader's bar-by-bar shape in a maintained engine library, free to download and $20 a month to license, with costs modelled by default, written fill rules, a visual trade diagnostics tool that charts every closed trade beside the values your strategy printed, and live orders through its relay; in a published benchmark on 500,000 bars, Backtrader took 15.04 s and Reamer Research took 3.38 s from Python. Among free tools, backtesting.py is the simplest step for one instrument, bt suits portfolio rebalancing, vectorbt suits parameter sweeps, and NautilusTrader and LEAN cover ticks, more asset classes and live trading.GuideWhat are the best backtesting.py alternatives in 2026?The best backtesting.py alternative depends on what you have outgrown. Reamer Research from Reamer Labs is a paid engine library for strategies on bars, free to download and $20 a month to license, with several instruments in one strategy, written fill rules, modelled costs, a visual trade diagnostics tool that charts every closed trade beside the values your strategy printed, and live orders through its relay. Among free tools, Backtrader and bt handle several instruments, vectorbt handles large parameter sweeps, and NautilusTrader and LEAN cover ticks, order books and live trading. If one instrument is enough, backtesting.py is still maintained and may be all you need.GuideWhat is a vectorbt or vectorbt PRO alternative for event-driven backtesting?Reamer Research from Reamer Labs, free to download and $20 a month to license, is an event-driven alternative to vectorbt and vectorbt PRO: your strategy is called on every bar with its full position state and resting orders, fills follow a written execution specification, runs are byte-identical, a visual trade diagnostics tool charts every closed trade beside the values your strategy printed, and its relay sends the same orders live. It is a free download, and $20 a month removes the Evaluation Mode limits. Free event-driven options are Backtrader, backtesting.py, NautilusTrader and LEAN; bt is a free vectorised option if you want to stay with arrays.GuideWhat are the best Deltix QuantOffice alternatives in 2026?Deltix QuantOffice alternatives depend on what you need from it. QuantConnect and QuantRocket offer an all-in-one platform at a smaller scale. NautilusTrader is a free open-source platform for ticks and many venues. kdb+ and OneTick cover the data side. Reamer Research (free to download, $20 a month to license) and Reamer Server from Reamer Labs cover research to live trading for strategies on bars, with published prices and self-serve checkout.GuideWhat are the best MultiCharts alternatives in 2026?MultiCharts alternatives fall into two groups. Trading applications such as NinjaTrader, TradeStation, Sierra Chart and AmiBroker keep the chart-first model, mostly on Windows. Code-first tools such as Zorro, QuantConnect and Reamer Research (free to download, $20 a month to license) and Reamer Server from Reamer Labs suit quants who want to write Python, C or C++. If you want to stay in a charting application, the first group fits better.GuideWhat are the best kdb+ alternatives in 2026?The main kdb+ alternatives are time-series and columnar databases such as QuestDB, ClickHouse, DolphinDB, TimescaleDB and ArcticDB. They differ in licence, query language and how much market data work they do for you. If you use kdb+ mainly to backtest strategies on bars, you may need an engine and not a database. Reamer Research from Reamer Labs, free to download and $20 a month to license, is one, and it can take bars from any of these.GuideWhat are the best DolphinDB alternatives in 2026?The main DolphinDB alternatives are kdb+ and KDB-X, QuestDB, ClickHouse and OneTick for storing and computing on market data. For backtesting, the alternatives are engines: NautilusTrader for ticks and order books, and Reamer Research from Reamer Labs, free to download and $20 a month to license, for strategies on time, tick, volume, dollar or custom bars. Replacing both the database and the matching simulator may take two tools.GuideWhat are the best OneTick alternatives in 2026?OneTick alternatives depend on which part of it you use. KDB-X, QuestDB, ClickHouse and DolphinDB replace the tick database. Databento replaces the managed market data. For backtesting strategies on bars, Reamer Research from Reamer Labs, free to download and $20 a month to license, supplies the fill rules and costs, and leaves the data to you. No single tool replaces all of OneTick, which combines data, analytics, transaction cost analysis and surveillance.FindingsFour Backtesting Engines on a 64-Core EPYC: What One Backtest CostsI ran Backtrader, NautilusTrader, LEAN and Reamer Research on the same data and strategies on an AMD EPYC 9575F, and measured how wall time and memory grow with bars, tickers and parallel copies. Reamer Research had the highest throughput and the lowest memory per run. Two of the four outcomes I stated against it in advance occurred, and both are reported here.FindingsReamer Server's Operating Envelope, Measured Again on 4.6.0Six weeks after the first whitepaper, I measured Reamer Server and Reamer Research again on the same CPU model, after a new release, a new kernel and a new compiler. I stated five earlier findings as claims the new run could refute. All five held, and two results outside the envelope moved.ReleaseReamer Research 4.4.0 and 4.5.0: Larger-Than-RAM Data, Monte Carlo, Full Position State and a Relay to Reamer ServerReamer Research 4.4.0 runs datasets larger than RAM from memory-mapped bar files, adds Monte Carlo resampling and an equity curve, and gives every function a plain name. 4.5.0 gives a strategy its full position state and every resting order on each bar, lets it re-price orders, move stops and targets, close all or cancel all, adds a progress callback that can stop a run, and sends the same orders a backtest produces to a running Reamer Server. C ABI version 7, additive from here on.FindingsWhat 2,000 Runs on a 64-Core EPYC Show About Reamer ResearchA single-threaded research engine has one job the rest of research depends on: run the same input twice and get the same answer, at a speed that does not wander. I measured both across 2,000 runs on an AMD EPYC 9575F. Timing held to a 0.59% coefficient of variation and every hashed output came back byte-identical.FindingsReamer Server on a 64-Core EPYC: The Tail Is the MeasurementA trading engine is usually sold on one number, a peak throughput or a best-case latency. Neither tells a team what happens under concurrent load. I measured the integration boundary, the end-to-end order path, and the full 1-to-512 strategy sweep on an AMD EPYC 9575F. The boundary costs 53 nanoseconds, the engine-only P99 sits at 12.4 microseconds, and throughput holds a plateau to about 67 strategies.
Buy a licenceKey and kit by email on payment.