Trading OS · web app + engine · anyx.trade
ANYX
A trading OS whose engine backtests 140 million candles in under 4 seconds.
Backtest engine
140,000,000
candles backtested in 3.9 seconds
You just watched it at real speed. Figure reported by the ANYX team.
Problem
Traders test ideas on tools that are slow, opaque or quietly wrong. A backtest that takes minutes kills curiosity, and one you cannot verify is worse than none.
What we built
The ANYX website and the product behind it: a strategy builder that reads plain language, a backtest engine we wrote from scratch that runs in the browser, and an AI strategist on top. Every number it shows can be reproduced.
Result
140 million candles in under 4 seconds. Fast enough that testing an idea feels like asking a question, and strict enough that impossible results are caught by an automated stress test, not by a trader.
The starting point
ANYX is meant to be the first tab a trader opens in the morning. That only works if answers arrive instantly and can be trusted, so speed and correctness were the product, not features of it.
What we built
A custom engine instead of an off-the-shelf library: strategies are described in plain language, compiled into a typed definition, and walked across years of real market data in a background worker so the interface never freezes. Strict rules stop the engine from ever using information a trader could not have had at that moment.
How we know it is right
Speed is easy to fake. Every engine change runs against an automated stress test that generates random strategies and fails the build if anything impossible comes out. Verified means nothing impossible was produced. It never means a strategy is profitable.
Where it stands
In early access at anyx.trade, used daily by the team that built it.