About us
We build verifiable quant tools, not stock tips
Why we built this
Most investment advice cannot be checked. "This one is going up" sounds confident, but it comes with no conditions and no time frame, so it can be justified afterwards whichever way the price moves. The value of a quantitative approach is not that it is more accurate — it is that it turns an idea into a falsifiable rule: when to enter, when to exit, what happens when you are wrong, all written down, then run against real historical data.
NineQuantAI exists to make that as easy as possible. You do not need to write code, pull data or build a backtest framework. Enter a symbol and a set of parameters, and within seconds you can see what those rules would have done in real history — including the stretches where they lost money.
We do not forecast markets, recommend securities, or provide investment advice of any kind. We provide tools and data. The conclusions are yours to draw.
Four things we believe
Results must be reproducible
The same symbol, the same parameters and the same window should give anyone the same result. Every number in a backtest must be traceable back to its inputs.
Data must declare its origin
Which convention the prices come from, how far coverage extends, whether they are adjusted, whether delisted names are included — these are not footnotes. They decide whether a conclusion holds at all, so we put them in the API response and on screen instead of burying them in a document.
State the costs up front
Every strategy has conditions under which it fails. Trend following wins only 35–45% of its trades and underperforms simply holding during a strong advance. We write that in the articles and on the interface, rather than letting you discover it after losing money.
Tools should not be expensive
Guests can run a backtest without signing up, registration comes with a free quota, and Pro is ¥99 a month. Testing your own trading ideas should not require an expensive ticket at the door.
What we do and do not do
We do
- Strategy backtesting on US equities and ETFs: chandelier-exit trend following and dual moving average, fully tunable
- A market data API with one consistent response envelope: daily bars, quotes and batch queries, open to guests
- AI assistant integration, so tools like Claude Code and Cursor can run and compare backtests for you
- Label the data convention, coverage window and timeliness honestly in every single response
We do not
- Recommend securities, publish buy or sell signals, or give investment advice in any form
- Connect to brokers — the platform has no order-placing capability and never touches your money
- Promise returns, or show performance figures without stating the convention and window behind them
- Market with equity curves, or hide the periods in which a strategy stopped working
Facts worth knowing before you start
Rather than advertising "five years of history," here are the real boundaries. Every one of them affects how you should read a backtest.
- Launch market
- US equities and US-listed ETFs
Hong Kong and mainland China markets have a place in the architecture, but are not offered today and we will not pretend otherwise.
- Consolidated daily bars start
- 2024-07-01
An objective boundary on the data-source side. We cannot produce backtests earlier than this, and we will not splice in another convention to fake it.
- Data convention
- Consolidated
Single-venue and consolidated data may never be mixed within one backtest: measured closing-price divergence reaches 7.1%, and single-venue volume is only 23%–34% of consolidated volume.
- Delisted symbols
- Kept in the symbol library
Excluding delisted names creates survivorship bias and systematically flatters backtest results.
- Distributed timeliness
- End-of-day data, T+1 or older
Redistributing real-time and intraday data requires additional exchange licensing that is not yet in place, so we do not offer it.
About this site itself
Articles in the Research section carry no sponsorship and no affiliate links. The measurements they cite — such as the ratio of single-venue to consolidated volume — come from our own day-by-day comparison, with the method and the conclusion both stated in the article so you can check them. If you find an error in our arithmetic, write to us and we will fix it.
Run a backtest first, then decide whether to trust us
No sign-up required for guests. Enter a symbol and parameters, get results in seconds — including the stretches where the strategy lost money.