Why we built Tickerbot

Almost every retail-facing stock alert tool was designed in the 1990s and never really moved on. The interfaces got prettier, the colors got darker, and the monthly subscription got more expensive — but the underlying assumptions about how people actually want to be told things about the market stayed frozen in time. We built Tickerbot because we wanted the alert tool we couldn't buy.

The state of stock alerts

If you set up a stock alert today, here's what you're picking from:

• Your brokerage's price-cross alerts ("notify me when AAPL crosses $200")
• Finviz's saved screeners ("filter by 67 dropdowns, manually re-check daily")
• Trade Ideas' desktop client ("install Windows software, learn the visual scan builder")
• TradingView's per-symbol alert builder ("write Pine Script for anything multi-condition")
• Discord bots that scrape the SEC for insider filings
• Your own custom Python scripts hitting Polygon's API on a cron

The first four are old. The last two require either being in someone else's community or being a developer. None of them do the obvious thing: let you describe what you want in your own words, and ping you when it happens.

Why this gap exists

The reason is fundamentally that turning a sentence into a stock query is hard. You need three things at once:

• A model that can translate the sentence into structured logic
• A live database that can answer the resulting query in milliseconds
• A scheduler that can run the query continuously and dedup the firings

Until very recently, the first one wasn't possible at consumer prices. Now it is — and the rest of the stack (live market databases, push notifications, mobile clients) has been consumer-grade for years. The pieces existed, just nobody had stitched them together for the actual use case.

What we wanted

The litmus test we kept coming back to: "Any S&P 500 stock with an analyst upgrade in the last week and EPS estimates revised up in the last 30 days that just broke the 50-day moving average on 2× volume." That sentence describes a real, useful, multi-factor setup. It is also impossible to build in any of the tools above without either (a) a paid Bloomberg terminal or (b) writing custom code.

We wanted to type that sentence into a phone, hit save, and walk away. So we built it.

The hard part wasn't the AI. It was getting the data continuous, accurate, and fast enough to evaluate against ~12,000 tickers every five minutes.

What we ended up with

Tickerbot today watches 12,000+ stocks plus 20 forex pairs, 100 cryptocurrencies, gold, oil, and 10 economic indicators. It updates every five minutes during US market hours. It precomputes 50+ named signal flags so most alerts don't need any AI translation at all — you can just say "small-cap breakout" and the flag is already there. For the more complex alerts, the AI translates the sentence to a query once, and the query runs forever.

It's a mobile app, because nobody wants to be tied to a desktop to find out what's happening in the market.