# PlainQuant

> Stock backtests with the honest number, for Hong Kong and US stocks (daily prices, long only). Type a trading
> rule in plain English or Chinese; PlainQuant tests it with delisted US stocks kept in, every trade charged commission,
> fees and the spread, and the result discounted for the number of settings tried. It also shows what the US stock
> market did in the 2 weeks, 4 weeks and 3 months after past events (Fed decisions, CPI, jobs reports, elections,
> sharp falls). Facts about the past only: no advice, no forecasts. English and Traditional Chinese (`?lang=zh-Hant`).

## Use it

- Test a rule: open https://plainquant.com/ and type the rule, e.g. "US stocks above their 200-day average, 10% stop". An
  account (email link, no password) runs it.
- Check stocks: https://plainquant.com/check?q=NVDA,AAPL,700 shows each stock's trend and business scores and how often stocks
  like it fell sharply afterwards. Symbols are US tickers, HK codes or company names, comma-separated.
- After an event: https://plainquant.com/events lists the events; each one has its own page (below).
- Example backtests: https://plainquant.com/demo/US

## MCP server and JSON API (free, read-only, no account or key)

- MCP (streamable HTTP): https://plainquant.com/mcp, with tools list_events, event_study(event_id), check_stocks(symbols).
  Claude Code: `claude mcp add --transport http plainquant https://plainquant.com/mcp`. Codex: in `~/.codex/config.toml`,
  `[mcp_servers.plainquant]` with `url = "https://plainquant.com/mcp"`.
- JSON: https://plainquant.com/api/v1/events, https://plainquant.com/api/v1/events/{event_id}, https://plainquant.com/api/v1/check?symbols=NVDA,700.
  Figures are fractions (0.037 = +3.7%). Limit: 60 requests a minute per IP.
- [Developers](https://plainquant.com/developers)

## Events (US stock market, equal-weighted, not the S&P 500)

- [Fed (FOMC) rate decision](https://plainquant.com/events/us-fomc-decision): 70 times from 2018-05-02 to 2026-10-30. 3 months later the US market's median move was +3.7%, higher 76% of the time; any 3-month window: +3.8%, higher 72%.
- [US inflation (CPI) report](https://plainquant.com/events/us-cpi-release): 101 times from 2018-05-02 to 2026-10-30. 3 months later the US market's median move was +3.9%, higher 73% of the time; any 3-month window: +3.8%, higher 72%.
- [US jobs report (payrolls)](https://plainquant.com/events/us-jobs-report): 101 times from 2018-05-02 to 2026-10-30. 3 months later the US market's median move was +4.4%, higher 69% of the time; any 3-month window: +3.8%, higher 72%.
- [US midterm election](https://plainquant.com/events/us-midterm-election): 2 times from 2018-05-02 to 2026-10-30, too few to summarise. The US market 3 months later: 2022-11-08 +10.3%; 2018-11-06 -1.1%.
- [US presidential election](https://plainquant.com/events/us-presidential-election): 2 times from 2018-05-02 to 2026-10-30, too few to summarise. The US market 3 months later: 2024-11-05 +1.3%; 2020-11-03 +17.9%.
- [Quarterly options expiry (quad witching)](https://plainquant.com/events/us-quad-witching): 34 times from 2018-05-02 to 2026-10-30. 3 months later the US market's median move was +3.3%, higher 79% of the time; any 3-month window: +3.8%, higher 72%.
- [Market down 10% from its high](https://plainquant.com/events/us-market-drop-10): 4 times from 2018-05-02 to 2026-10-30, too few to summarise. The US market 3 months later: 2025-03-13 +9.3%; 2022-01-21 -1.3%; 2020-02-27 -1.2%; 2018-10-24 +2.5%.
- [Market down 20% from its high](https://plainquant.com/events/us-market-drop-20): 2 times from 2018-05-02 to 2026-10-30, too few to summarise. The US market 3 months later: 2022-06-13 +5.7%; 2020-03-09 +19.3%.
- [Market falls 3% in a day](https://plainquant.com/events/us-market-down-day-3): 16 times from 2018-05-02 to 2026-10-30. 3 months later the US market's median move was +2.9%, higher 75% of the time; any 3-month window: +3.8%, higher 72%.

With fewer than 5 past occurrences a page lists each one instead of a summary.

## How the numbers work

- [Method](https://plainquant.com/method): the data, the costs, how the score is discounted for luck, and whether the stock
  scores have predicted anything.
- [Plans](https://plainquant.com/waitlist)

Past results don't predict future returns. Educational tool, not investment advice.