Why Breakout Scanner uses a text box instead of dropdowns

The first version of Breakout Scanner used on a traditional dropdown-based interface. Pick a moving average, type in a threshold, add another filter, repeat. It worked, but there was a consistent pattern in user testing: people would stare at the options, scroll through the dropdown menus, and struggle to map the strategy in their head to the controls on screen.

That observation led to the decision to replace the filter interface with a plain-English text input powered by AI.

The problem with dropdowns

Traditional screener interfaces force you to think in the tool's language. A trader has an idea: “I want stocks that are about to clear their all-time high on strong volume, trending above the 200-day.” That's one sentence. But to express it in a typical screener, you need to find the “52-week high proximity” filter (if it even exists), set it to some percentage, find the relative volume filter, pick a threshold, add the SMA 200 filter, and specify “price above.” Four or five clicks and dropdown selections to say one thing.

Dropdowns work fine for simple filters. But the more specific the idea gets, the more you end up fighting the interface to express it. And if the screener doesn't have a filter for the exact condition you want, you're stuck.

What the AI actually does

When you type a strategy description into Breakout Scanner, the AI reads it and maps it to a combination of filters: which moving averages to use, what the thresholds should be, which price-relative-to-level conditions to apply, and what volume ratio to require. It performs the same mapping a trader would do manually in a dropdown interface, just faster and from a natural description instead of a form.

The AI works from the same set of indicators available in the platform: SMA (5, 20, 50, 200), EMA, relative volume, 52-week high/low proximity, moving average crossovers, price channels, and ceiling detection logic. What it does well is determine which combination matches the description and set reasonable defaults for thresholds when exact numbers aren't specified.

Where it's better, where it's not

The AI approach is clearly faster for complex screens. Three or four conditions combined in a specific way can be expressed in one sentence rather than clicking through multiple dropdown menus. It's also better for rapid exploration: five different strategy descriptions can be typed and backtested in a few minutes.

Where it's less ideal: very precise numeric thresholds. If you need the SMA(20) to be exactly 3.2% above the SMA(50), the AI will typically interpret that correctly, but it's worth verifying. Breakout Scanner shows the full generated configuration so every parameter can be reviewed and adjusted before running anything.

Both approaches coexist

The manual filter editing wasn't removed. After the AI generates a screener, every individual parameter can be opened and adjusted by hand. Some users let the AI handle the initial setup and then fine-tune one threshold manually. Others type everything and trust the output. Both workflows are supported.

The goal was not to replace the trader's judgment. It was to remove the friction between having an idea and testing it. The less time spent configuring an interface, the more time available for evaluating whether the strategy itself is sound.

Iteration speed changes behaviour

One notable effect of the text-based approach: users iterate significantly more. With dropdown interfaces, building a screen takes enough effort that people tend to stick with their first attempt. With a text box, the cost of trying a new configuration is effectively zero. Users routinely test eight or ten variations in a single session.

This higher iteration rate tends to produce better screens, because strategies get refined through rapid experimentation rather than settled on after a single attempt.