Method
Around here, βscaryβ cuts both ways. A stock can be scary because the opportunity is big enough to be unnerving, or because the risk is. The Scarymeter puts both on one line from β100 to +100. The sign and the colour tell you which kind of scary. The distance from zero tells you how much.
Scary good
+45 to +100
The opportunity is large enough to make you nervous. Trend, value, and the story mostly agree.
Tempting
+15 to +45
More going for it than against it, with something still holding it back.
Not scary
β15 to +15
Balanced, or nothing much happening. Fine is a valid answer.
Spooky
β45 to β15
The risks have the upper hand, though the case isn't closed.
Scary bad
β100 to β45
Risk is winning: broken trend, stretched valuation, bad news, or all three.
Six factors are scored from β100 to +100 on live market data and combined by weight. When the AI desk has filed a brief for the stock, its conviction score makes up 35% of the final number and the six factors the other 65%. Each stock page shows every factor and what drove it.
| Trend | 20% | Where price sits against its 50-day and 200-day averages, and whether the 50 is above the 200. |
|---|---|---|
| Street view | 20% | Distance to the mean analyst price target, plus the consensus rating. |
| Momentum | 15% | One-month and three-month returns, trimmed when RSI says the move is stretched. |
| Valuation | 15% | Forward P/E, judged against revenue growth rather than in isolation. |
| Fundamentals | 15% | Revenue growth, operating margin, free cash flow, and net debt against EBITDA. |
| Risk | 15% | Realized volatility, distance from the 52-week high, beta, and short interest. This one mostly subtracts. |
Prices refresh in your browser every 15 seconds. Technical indicators are recomputed from daily bars every few minutes, and fundamentals a few times a day. All of it comes from Yahoo Finance and can lag the exchange.
The AI desk runs once a day with no one in the loop. For each stock, an AI model searches the web for what changed, then writes the read, the bull and bear cases, the catalysts, and bear, base, and bull price scenarios for one month, six months, one year, and three years, with its reasoning. A separate pass writes Scary News. News links are checked against what the search returned before anything is published.
When the desk canβt file, the site says so. Pages fall back to a numbers-only read and a statistical scenario cone, labelled as such, rather than showing analysis that was never written.
It is not advice. The score is a fixed formula plus a language modelβs opinion; it knows nothing about your situation, and both parts can be wrong. Scenario targets are a way of laying out assumptions, not predictions to trade on. Use this to decide what to look into, then look into it.
Charts are drawn with TradingView Lightweight Chartsβ’ (Apache 2.0). Market data: Yahoo Finance.