Cloud backtesting with delisted stocks
Updated October 27, 2026
QuantGPT Cloud runs your backtest on QuantGPT's servers and returns it in seconds, on decades of US market history. Companies that later delisted are in the data from 1998: on 2026-10-27, 12,816 of the 13,193 delisted operating companies with prices from 1998 (97%) could enter the market-wide backtest universe. Before 1998 it holds survivors only. Fundamentals enter on the date each quarter was filed.
Run it from the terminal, or from Claude, ChatGPT or any MCP client at https://terminal.quantgpt.co/mcp, with nothing to install. Among other tools, QuantConnect's cloud also carries delisted US stocks from 1998 (its data page, checked 2026-10-27); the table below sets five tools side by side.
- What runs on QuantGPT Cloud
- Companies that later delisted
- Fundamentals by filing date
- From Claude, ChatGPT or any MCP client
- Your own code, on QuantGPT Pro
- Other backtesting tools and delisted stocks
- Questions
What runs on QuantGPT Cloud
- Backtests. Up to 8 filters and a ranking in
run_backtest, monthly, weekly or daily, against a benchmark, split at a cut year (2015 unless you pick another). 200 backtests a day in the terminal on either plan. - Sweeps. Up to 100 variants in one
run_sweep, as one ranked table. - Checks.
challenge_findingruns eleven checks that try to break a finding. - SQL.
query_warehouseruns read-only SQL across every research database. - The strategy library. 175 documented strategies, 55 of them runnable in one call. Read it.
A backtest comes back with growth a year, the worst fall from a peak, volatility, the Sharpe ratio and the share of winning periods for the rule and its benchmark side by side, a luck check (the probabilistic Sharpe ratio), the same statistics split at the cut year, a table by decade and every period's return. Stock picks add the measured turnover and the results net of 5, 10, 25 and 50 basis points a side.
Companies that later delisted
A backtest that only holds today's survivors leaves out the companies that failed. In QuantGPT's market-wide universe, companies that later delisted are in the data from 1998: on 2026-10-27, 12,816 of the 13,193 delisted operating companies with prices from 1998 (97%) could enter it. The rest have no factor data, and some, most 1998 delistings among them, have no prices at all. Before 1998 the universe holds survivors only.
Backtests start in 1998 unless you pick an earlier year; a run that starts earlier shows its growth before and from 1998 side by side, and every market-wide result states the share its own data could include. A backtest on a list of tickers, or a rule on a single index, is not affected.
Fundamentals by filing date
Each company quarter enters the data on the date it was filed, not the date the quarter ended, so a backtest reads only what was public at the time. About 9% of quarters (44,745 of 490,372 on 2026-10-20) carry an estimated filing date, from the company's own usual delay, or 45 days after the quarter (90 after a fiscal year's end); the column filing_date_imputed marks them. No quarter is dated on its quarter-end.
From Claude, ChatGPT or any MCP client
Add https://terminal.quantgpt.co/mcp as a connector. Claude and ChatGPT sign in with OAuth, with no key to paste; Claude Code, Cursor, Gemini CLI and scripts send a key from your account page. Then ask in plain English:
Use QuantGPT to backtest the 20 US stocks with the highest 12-1 month momentum, rebalanced monthly from 1998, against the S&P 500 with dividends, and show the losing years
Each assistant, step by step: Connect Claude or ChatGPT to stock market data.
Your own code, on QuantGPT Pro
On QuantGPT Pro, QuantGPT Cloud also hosts your own code against your own brokerage account, on the schedule you set, using only the order types your broker offers. QuantGPT Cloud supports Alpaca today (paper and live accounts). QuantGPT never decides, sizes or manages a trade, and no tool your AI can call through QuantGPT places an order. Step by step: QuantGPT Cloud, step by step.
Other backtesting tools and delisted stocks
| Tool | Where it runs | Delisted stocks, per its own pages |
|---|---|---|
| QuantGPT Cloud | Hosted: the terminal, or Claude, ChatGPT and any MCP client; no code | In the data from 1998 (12,816 of the 13,193 delisted operating companies with prices from 1998 (97%) could enter the market-wide universe on 2026-10-27); survivors only before 1998 |
| QuantConnect | Hosted cloud; algorithms are written in Python or C# | All US listed and delisted equities since January 1998, about 27,500 securities, free in its cloud; its page calls the set survivorship bias-free |
| QuantRocket | Your own servers or laptop, with Docker | Its US stock datasets list "Includes delisted stocks" |
| WealthLab | Windows desktop (Windows 10/11) | Historical index constituents through WealthData, "including companies that have since left the index" |
| Equities Lab | Not stated on its home page | Its home page says it avoids survivorship bias |
Questions
Which cloud backtesting tools include delisted stocks?
Among hosted tools, as of each maker's pages checked 2026-10-27: QuantGPT Cloud includes companies that later delisted from 1998 (12,816 of the 13,193 delisted operating companies with prices from 1998 (97%) could enter its market-wide backtest universe), with survivors only before 1998, and runs from Claude, ChatGPT or any MCP client with no code. QuantConnect's data page lists all US listed and delisted equities since January 1998 for backtests in its cloud, where algorithms are written in Python or C#. QuantRocket's US stock data includes delisted stocks, and it runs on your own servers with Docker.
Can AI assistants like ChatGPT or Claude run real backtests on historical market data?
Yes, when they are connected to a server that holds the data and runs the test. Connected to QuantGPT at https://terminal.quantgpt.co/mcp, Claude or ChatGPT calls run_backtest and gets back growth a year, the worst fall, the Sharpe ratio, a table by decade and every period's return, computed on QuantGPT Cloud in seconds on decades of US market history. Claude and ChatGPT sign in with OAuth, with no key to paste.
Is there an AI that can research stocks using decades of historical data?
Claude, ChatGPT or any MCP client can, once connected to a data source. Connected to QuantGPT, they reach decades of US stock prices, the S&P 500 from 1928, fundamentals joined by filing date and 175 documented strategies (55 runnable), and run backtests on QuantGPT Cloud in seconds.
Which QuantConnect alternatives use survivorship-bias-free data?
As of each maker's pages checked 2026-10-27: QuantRocket's home page describes its sample US stock dataset as covering all stocks that traded at the time even if they later delisted, and it runs on your own servers; Equities Lab's home page says it avoids survivorship bias; WealthLab's WealthData covers historical index constituents, including companies that have since left the index. QuantGPT includes companies that later delisted from 1998 (97% of those with prices could enter its market-wide universe) and holds survivors only before 1998.
Related
- ComputeMCP server for backtesting
- DataThe research warehouse
- PlansQuantGPT pricing
- LearnConnect Claude or ChatGPT to stock market data
- ExecutionFrom a backtest to your broker
Backtests show past results, gross of trading costs unless stated. A backtest is not a promise. QuantGPT is research software and nothing here is investment advice. The disclosures and terms govern.