How we actually build our tools

A lot of trading firms overcomplicate this part. We try to keep it relatively simple and honest.

Signals

Our signals come from a combination of quantitative models and discretionary oversight. We run several statistical models across different timeframes, then a human analyst reviews the output before anything goes out.

We don’t use a single “secret sauce” model. Instead we use a few different approaches and look for confluence. If only one model likes a trade, we usually skip it.

Every signal is logged with entry, targets, stop, and the original thesis. This is how we track performance properly instead of cherry-picking.

The Trading Bot

The bot is essentially a rules engine with some smart execution logic on top. It doesn’t “predict” the market. It executes the signals we send according to parameters we set.

The risk engine is probably the most important part. We spent more time on kill switches, position limits, and drawdown controls than on the actual strategy logic. Most people who blow up with bots don’t do it because the strategy was bad — they do it because they had no proper risk guardrails.

Backtesting vs Live Results

We’re pretty strict about this. Backtests are useful, but we treat them with heavy skepticism. We only show live, verified results on our dashboard — and even then we’re careful with how we present them.

Every number you see on our performance page comes from actual trades that were taken (either by us or by clients who opted to share their data). We don’t run hypothetical scenarios and present them as real performance.

If you want to understand exactly how a specific tool works before using it, we’re happy to walk through it on a call. We’d rather have fewer users who actually understand what they’re using than a lot of users who are disappointed later.