Skip to content

Canli Validation changelog

io.github.arhancanli/canli-validation-mcp

13 changes recorded. MCP Harbor re-reads this listing from the official registry every six hours and asks the live server what it exposes, and keeps what changed between one look and the next.

  • llms.txt +28 −5

    https://canlicapital.com/llms.txt

    − > Canli Capital is an open-source lab that connects AI agents to financial data and quant tools they can check. Its flagship MCP server, canli-mcp, reads SEC filings, market data and economic statistics at the source and puts 388 finance tools behind five, with interactive charts in the chat. It is free, MIT-licensed and needs no API key or account. Built and maintained by Arhan Canli.− - canli-mcp: SEC filings, market data, economic statistics and backtests in one server. Run `npx -y canli-mcp`, or connect to https://canlicapital.com/mcp/canli with nothing to install.− - [canli-quant-mcp](https://canlicapital.com/mcp-servers/quant): 235 quant tools: performance and risk, options and exotics, fixed income, portfolios, econometrics, indicators and 399 strategy sleeves. Run `npx -y canli-quant-mcp`.−   - [What is the best free MCP server for finance?](https://canlicapital.com/mcp-servers/answers/best-free-finance-mcp-server)− - [Sitemap](https://canlicapital.com/sitemap.xml): all 409 indexable routes.+ > Canli Capital is an open-source lab that connects AI agents to financial data and quant tools they can check. Its flagship MCP server, canli-mcp, reads SEC filings, market data and economic statistics at the source and puts 443 finance tools behind five, with interactive charts in the chat. It is free, MIT-licensed and needs no API key or account. Built and maintained by Arhan Canli.+ - [canli-mcp](https://canlicapital.com/mcp-servers): SEC filings, market data, economic statistics and backtests in one server; the latest release is 0.22.0. Run `npx -y canli-mcp`, or connect to https://canlicapital.com/mcp/canli with nothing to install.+ - [canli-quant-mcp](https://canlicapital.com/mcp-servers/quant): 236 quant tools: performance and risk, options and exotics, fixed income, portfolios, econometrics, indicators and 399 strategy sleeves. Run `npx -y canli-quant-mcp`.+ ## What Canli Capital is working toward+ Goals, each labelled with what exists; nothing planned is live until it says so. The whole plan: https://canlicapital.com/vision+ - Live data you connect (canli-mcp, Building, November 2026): Prices that update by themselves in your assistant: crypto from the exchanges' public feeds with no key, stocks through your own data key. On your own computer, under each provider's terms.+ - Tests for forecasts and crashes (canli-mcp, Building, November 2026): Whether one forecast really beats another, and thousands of synthetic crashes built from free daily data, to test a strategy against markets that have not happened yet.+ - Your data stays yours (canli-mcp, Building, November 2026): Run the maths on data you already pay for, on your own machine. Nothing is stored or sent.+ - The Lab: a million strategies (canli-mcp, Planned, December 2026): A library of a million strategies built from free data, every one counted. Ask whether your strategy is new, how it compares with the closest of the million, and whether it survives the luck of the search.+ - Canli models for everyone (canli-mcp, Planned, from December 2026): Open models anyone can use: factors, default risk, a nowcast of the economy, the range a company's own filings imply for its value, and the odds of an earnings surprise. Each is judged only on a public record that starts the day it launches, and none is advice.+ - Checks no tool does yet (canli-mcp, Planned): Whether a strategy is original, a reader that finds look-ahead and overfitting in a backtest, an alert when a company you follow restates or changes auditor, and the chain from any number back to the line it came from.+ - November 2026, Live data and new tests in canli-mcp: Live prices you connect, the forecast test and the crash generator, and your own data kept on your machine.+ - December 2026, The Lab, Canli Books and the challenge: The first release of Canli Books, the Lab's library of strategies, Canli Report's first check, and the pilot season of the challenge.+ - Early 2027, Canli Close and the first models: Bank reconciliation and month-end close on Canli Books, and Canli models with their first months of public record.+ - Spring 2027, Canli Audit: Journal-entry tests, Benford's law and restatement risk for auditors and audit committees.+ - Mid-2027, Tools developers pay for: Finance MCP servers measured on independent tests; the first paid developer tier; the sleeve library opening to developers; ALPHAC adding new, uncorrelated families with a year of forward record.+ - 2027, The research platform: The training environment and testing servers, the expert annotation network, and ALPHAC across asset classes with far more independent bets.+ - 2028, Groundwork for the funds: The US company, counsel and compliance, and a real-money track record through the legal path, before any fund opens.+ - Product lines, each becoming its own company once it has paying customers: Canli Markets (Live); Canli Data (Live); Canli Research and the Lab (Live); Canli Books (In development); Canli Report (Planned, from December 2026); Canli Close (Planned, early 2027); Canli Audit (Planned, spring 2027); Canli Treasury and Planning (Planned, mid-2027); ALPHAC (Paper only); The funds (Later).+ - What it will not do: Investment advice, or telling anyone what to buy or sell. Holding or moving anyone's money. Audit opinions or tax advice. Filing on anyone's behalf. Keeping a company's books on our servers before hosting has had its security work.+ - [Weekly test on new filings](https://canlicapital.com/benchmarks/fresh-filings): each week, questions only that week's annual reports can answer, put to a model alone and with canli-mcp, every answer checked against the filing.+ - [A million simple strategies to beat](https://canlicapital.com/notes/a-million-simple-strategies): The Lab runs 588 textbook rules on each of 1,678 public return series: 986,664 strategies, each rebuilt exactly from its number. A backtest can be set against all of them in seconds, to see whether it is new and whether it beats what simple rules on public data already earned. They vary like about 15 independent ones.+ - [Reading bankruptcy risk from an annual report](https://canlicapital.com/notes/bankruptcy-risk-from-an-annual-report): A model that reads only a company's latest annual report, tested year by year on reports it had never seen, ranked the next year's bankruptcies better than Ohlson's O-score and Altman's Z'' across all US filers. Among companies with at least $10 million of assets, Ohlson's O-score did better, and the tool says so in every answer.+ - [Will a company beat its own guidance?](https://canlicapital.com/notes/beat-your-own-guidance): A model of where a company's next EPS or revenue lands against the range it guided to: its own record blended with the market's. Tested quarter by quarter on S&P 500 press releases, each quarter predicted only from what was filed before it, it beat the market-wide rate for both EPS and revenue, with intervals clear of zero.+ - [Do factor-timing claims beat a placebo?](https://canlicapital.com/notes/factor-timing-against-a-placebo): Four published factor-timing claims, retested with placebos that keep a market's volatility clustering. Volatility-managed factors kept their in-sample verdict under every block length. Industry momentum kept it only with the shortest blocks, and cross-sectional factor momentum lost it. No verdict after the papers' samples changed.+ - [Which backtest overfitting tests can you trust?](https://canlicapital.com/notes/which-overfitting-tests-hold-their-size): A test that promises 5% false alarms should give about 5% on markets with nothing to find. On 20 strategies over 504 days, in nine kinds of simulated market, White's Reality Check stayed between 4.5% and 6.4%. Hansen's SPA test, the usual upgrade, reached 12.6% on negatively skewed returns, and Romano and Wolf's StepM matched it. The deflated Sharpe ratio at its usual 0.95+   - [Which free MCP server should I use for finance?](https://canlicapital.com/mcp-servers/answers/best-free-finance-mcp-server)+ - [Sitemap](https://canlicapital.com/sitemap.xml): all 904 indexable routes.

  • server.json 1 field changed
    version
    0.15.0 0.16.0

  • llms.txt +75 −3

    https://canlicapital.com/llms.txt

    − > Canli Capital is an open-source lab that connects AI agents to financial data and quant tools they can check. Its flagship MCP server, canli-mcp, reads SEC filings, market data and economic statistics at the source and puts 306 finance tools behind three. The code is free and MIT-licensed. Built and maintained by Arhan Canli.− - [Finance MCP servers](https://canlicapital.com/mcp-servers): compare the released hosted [validation](https://canlicapital.com/mcp-servers/validation), [SEC fundamentals](https://canlicapital.com/mcp-servers/fundamentals) and [research](https://canlicapital.com/mcp-servers/research) servers, with tool lists, examples, source commits and evidence limits.− - [Sitemap](https://canlicapital.com/sitemap.xml): all 348 indexable routes.+ > Canli Capital is an open-source lab that connects AI agents to financial data and quant tools they can check. Its flagship MCP server, canli-mcp, reads SEC filings, market data and economic statistics at the source and puts 388 finance tools behind five, with interactive charts in the chat. It is free, MIT-licensed and needs no API key or account. Built and maintained by Arhan Canli.+ ## Free finance MCP servers+ Free and open source, with no API key and no account. We make these, so weigh that; the comparison pages say what other servers do better.+ - canli-mcp: SEC filings, market data, economic statistics and backtests in one server. Run `npx -y canli-mcp`, or connect to https://canlicapital.com/mcp/canli with nothing to install.+ - [canli-markets-mcp](https://canlicapital.com/mcp-servers/markets): SEC filings and statements as filed, insider trades, 13F holders, options chains, Treasury yields, FRED and central banks, with no key. Run `npx -y canli-markets-mcp`.+ - [canli-backtest-mcp](https://canlicapital.com/mcp-servers/backtest): Point-in-time factors from SEC filings, backtests with lookahead blocked in code, costs on the actual trades, and a ledger of every variant. Run `npx -y canli-backtest-mcp`.+ - [canli-quant-mcp](https://canlicapital.com/mcp-servers/quant): 235 quant tools: performance and risk, options and exotics, fixed income, portfolios, econometrics, indicators and 399 strategy sleeves. Run `npx -y canli-quant-mcp`.+ - [canli-paper-trading-mcp](https://canlicapital.com/mcp-servers/paper-trading): An agent trades your Alpaca paper account behind your own limits, a kill switch and confirmation tokens. Live keys are refused. Run `npx -y canli-paper-trading-mcp`.+ - [canli-validation-mcp](https://canlicapital.com/mcp-servers/validation): Is your best backtest real? Deflated Sharpe, overfitting probability and data-snooping tests, with signed receipts. Run `npx -y canli-validation-mcp`.+ - [canli-fundamentals-mcp](https://canlicapital.com/mcp-servers/fundamentals): What a company reported, and when: first-reported values, what was known on any date, and later restatements flagged. Run `npx -y canli-fundamentals-mcp`.+ - [canli-research-mcp](https://canlicapital.com/mcp-servers/research): Canli Capital's open research record: every paper and failed candidate, the trial count, and the live paper record. Run `npx -y canli-research-mcp`.+ - [Free finance MCP servers compared](https://canlicapital.com/mcp-servers/answers/best-free-finance-mcp-server): canli-mcp against OpenBB, EdgarTools and Alpha Vantage's server, with the benchmark behind it.+ - [Finance MCP servers that need no API key](https://canlicapital.com/mcp-servers/answers/finance-mcp-server-no-api-key)+ - [Connect ChatGPT to stock data](https://canlicapital.com/mcp-servers/answers/connect-chatgpt-to-stock-data)+ - [Connect Claude to stock data](https://canlicapital.com/mcp-servers/answers/connect-claude-to-stock-data)+ - [Free SEC EDGAR MCP server](https://canlicapital.com/mcp-servers/answers/free-sec-edgar-mcp-server)+ - [How often companies beat their own guidance](https://canlicapital.com/notes/how-often-companies-beat-their-guidance): We set what S&P 500 companies said they would report against what they then reported, quarter by quarter, from their own press releases. Given a range, they landed above it far more often than below it: revenue 170 times against 13, earnings per share 259 times against 24.+ - [The year-end numbers that changed later](https://canlicapital.com/notes/the-year-end-numbers-that-changed): We read 500 companies held by an S&P 500 fund three separate ways and looked for figures that disagree. Six companies had a year-end number that a later filing changed. None of those filings describes fixing a mistake. Here is what each one did.+ ## Lessons+ Short lessons for people investing their own money, using figures from real filings. Education, not advice.+ - [Learn how the stock market works](https://canlicapital.com/learn): The index of lessons.+ - [How to read an income statement](https://canlicapital.com/learn/how-to-read-an-income-statement): An income statement runs from what customers paid down to what was left for the owners. Here is every main line, using what Apple filed for the year to September 2025.+ - [What a P/E ratio tells you, and what it hides](https://canlicapital.com/learn/what-a-pe-ratio-tells-you): The price-to-earnings ratio says how many dollars buyers pay for one dollar of a company's yearly profit. Here is how to work it out from a filing, and how it can mislead you.+ - [What you own when you buy a share](https://canlicapital.com/learn/what-you-own-when-you-buy-a-share): A share is a small slice of a real business. This lesson shows what that slice earned and paid in one year, using the figures Apple filed with the SEC.+ - [What is inside an index fund](https://canlicapital.com/learn/whats-inside-an-index-fund): A fund that tracks the S&P 500 holds about 500 companies, but not in equal amounts. Here is what one of the largest, VOO, actually held, from the report it files with the SEC.+ - [Why a great track record can be luck](https://canlicapital.com/learn/why-a-great-track-record-can-be-luck): Try enough strategies and one will look brilliant by chance. Here is the arithmetic, and the questions to ask before you trust a result someone shows you.+ - [canli-mcp](https://canlicapital.com/mcp-servers): the flagship finance MCP server. Install: `npx -y canli-mcp` (stdio), or add the hosted URL https://canlicapital.com/mcp/canli as a connector. The page also covers the released hosted [validation](https://canlicapital.com/mcp-servers/validation), [SEC fundamentals](https://canlicapital.com/mcp-servers/fundamentals) and [research](https://canlicapital.com/mcp-servers/research) servers.+ - [Add canli-mcp to your client](https://canlicapital.com/mcp-servers/install): steps for Claude, Claude Code, Cursor, VS Code, ChatGPT, Codex, Gemini CLI, Windsurf, Cline and Zed.+ - [Privacy policy](https://canlicapital.com/privacy): what the site, the hosted servers and the package collect, send and keep.+ - [Every canli-mcp tool](https://canlicapital.com/mcp-servers/tools): all the tools by pack, with what each does and its arguments.+ - [How canli-mcp works and is tested](https://canlicapital.com/mcp-servers/how-it-works): what happens on a question, where the data comes from, test scores including first runs, and limits.+ - [Finance questions canli-mcp answers](https://canlicapital.com/mcp-servers/answers): one page per question, each with the call it makes and a recorded result:+   - [How do I get stock prices in Claude?](https://canlicapital.com/mcp-servers/answers/stock-prices-in-claude)+   - [How can an AI agent read a company's financial statements from SEC filings?](https://canlicapital.com/mcp-servers/answers/sec-financial-statements-for-ai)+   - [How do I backtest a trading strategy in Claude?](https://canlicapital.com/mcp-servers/answers/backtest-a-strategy-in-claude)+   - [Is there an MCP server for options chains?](https://canlicapital.com/mcp-servers/answers/options-chain-mcp-server)+   - [How do I get an implied volatility surface into an AI chat?](https://canlicapital.com/mcp-servers/answers/implied-volatility-surface-ai)+   - [How can an AI agent get the Treasury yield curve?](https://canlicapital.com/mcp-servers/answers/treasury-yield-curve-ai)+   - [How do I compare several stocks' returns and correlation in Claude?](https://canlicapital.com/mcp-servers/answers/compare-stocks-in-claude)+   - [How do I get today's market overview in an AI chat?](https://canlicapital.com/mcp-servers/answers/market-overview-in-claude)

  • server.json 1 field changed
    version
    0.14.0 0.15.0

  • llms.txt +10 −3

    https://canlicapital.com/llms.txt

    − > Canli Capital is an open research lab for quantitative finance that tests whether trading backtests are real or luck. Use its free validators in your browser, over an API or through an MCP client. Inspect SEC-traced company data and its own public paper-trading record, with failures and corrections kept visible. Built and maintained by Arhan Canli.− - [Home](https://canlicapital.com/): What the project is and where to go next.− - [Sitemap](https://canlicapital.com/sitemap.xml): all 341 indexable routes.+ > Canli Capital is an open-source lab that connects AI agents to financial data and quant tools they can check. Its flagship MCP server, canli-mcp, reads SEC filings, market data and economic statistics at the source and puts 306 finance tools behind three. The code is free and MIT-licensed. Built and maintained by Arhan Canli.+ - [Home](https://canlicapital.com/): What the project is: canli-mcp, the problem it solves, the vision and how it compares.+ - [Finance MCP servers, head to head](https://canlicapital.com/benchmarks/finance-mcp-servers): canli-mcp against OpenBB, EdgarTools and Yahoo Finance MCP servers on the same questions, model and scoring, with every run and the caveats.+ - [The record](https://canlicapital.com/record): The ALPHAC paper-trading record, published daily, with the trials that failed.+ - [Was the winning strategy skill or luck?](https://canlicapital.com/notes/best-of-399-strategies): I ran 399 textbook trading strategies on the same prices and asked whether the winner was skill. The answer depended almost entirely on which test I used, because the tests were answering different questions, and only one of them was the question I meant.+ - [The same SEC filing has two acceptance times](https://canlicapital.com/notes/edgar-acceptance-times): EDGAR's submissions API labels each filing's acceptance time as UTC. I checked 450 of them against the filing index pages: in 13 of 25 companies' files every time was late by one US Eastern offset, and the same filing read differently depending on which file it came from.+ - [My agent dropped numbers when it copied data into a tool call](https://canlicapital.com/notes/pasted-data-lost-numbers): Same model, same question, same tool. With 756 daily returns pasted into the prompt, the agent got 4 of 9 runs right; when the tool read the same series from a file, it got 8 of 9, with about a quarter of the tokens.+ - ["Silver price" returned US CPI](https://canlicapital.com/notes/silver-price-returned-cpi): A tool that turns plain words into a data series picked any series sharing half the words of the request. One generic word was enough, so "silver price" came back as consumer prices, with real numbers and no error.+ - [Your backtest beat a t-test. Would it beat a placebo?](https://canlicapital.com/notes/would-it-beat-a-placebo): In 300,000 simulated tests on markets with nothing to predict, a t-test on the best rule of a grid said "edge" between 10.9% and 78.9% of the time at a nominal 5%. A placebo test, which reruns the whole pipeline on the same returns in a random order, stayed between 4.6% and 5.5%.+ - [Sitemap](https://canlicapital.com/sitemap.xml): all 348 indexable routes.

  • llms.txt +6 −2

    https://canlicapital.com/llms.txt

    − > A public research and verification platform for systematic trading. You can inspect a live paper-traded portfolio, trace every decision to its evidence, and reproduce selected research yourself. Built and maintained by Arhan Canli.− - [Sitemap](https://canlicapital.com/sitemap.xml): all 336 indexable routes.+ > Canli Capital is an open research lab for quantitative finance that tests whether trading backtests are real or luck. Use its free validators in your browser, over an API or through an MCP client. Inspect SEC-traced company data and its own public paper-trading record, with failures and corrections kept visible. Built and maintained by Arhan Canli.+ - [Vision](https://canlicapital.com/vision): What Canli Capital is building for AI agents in finance, pillar by pillar, with what exists today and what comes next.+ - [MCP downloads](https://canlicapital.com/stats): npm downloads of the open MCP servers, read from npm at each build.+ - [FilingFacts benchmark](https://canlicapital.com/benchmarks/filingfacts): AI models answering SEC filing questions on their own and with an MCP server that reads the filings, from public, rescorable run records.+ - [Local paper execution MCP workflow](https://canlicapital.com/mcp-servers/execution): private, Unreleased repository source for supplied sizing, pre-trade checks, signed journals and paper evidence. Explicit local writes; no broker connection or order placement.+ - [Sitemap](https://canlicapital.com/sitemap.xml): all 341 indexable routes.

  • server.json 1 field changed
    version
    0.13.0 0.14.0

  • Resources +2
    • + strategy-spec
    • + openapi
  • Prompts +6
    • + backtest_and_validate
    • + stress_my_strategy
    • + production_check
    • + summarize_market_series
    • + check_signal_for_lookahead
    • + test_pipeline_on_placebos
  • Tools +7 −3
    • + validate_reality_check
    • + backtest_strategy
    • + summarize_series
    • + stress_test
    • + check_feasibility
    • + check_leakage
    • + placebo_test
    • − get_key
    • − get_receipt
    • − company_financial_history
  • server.json 1 field changed
    version
    0.12.0 0.13.0
  • server.json 1 field changed
    version
    0.11.0 0.12.0

  • server.json 1 field changed
    version
    0.10.1 0.11.0