How the Signal Score works - and what it can't do
One honest page: what goes into the score, what the number actually means, what it does notmean, and why we don't show a “beat the market” track record.
What the Signal Score is
Every night after the close, AlphaRanks reads a wide set of measurable signals on ~1,500 US stocks (the S&P 1500 plus high-demand ADRs) and top ETFs - price and volume technicals, company fundamentals from SEC filings, insider-trading (Form 4) filings, upcoming earnings, and news sentiment - and distills them into a single Signal Score from 1 to 10.
The Signal Score is a relative, descriptive snapshot. It tells you how strongly a stock's measurable signals line up right now, ranked against every other stock in the universe that night: a 10 means the signals are among the most favorable tonight, a 1 among the least. Because it's a rank, a stock's score can move simply because the rest of the market shifted - not only because the stock did. It is a description of the present, not a prediction of future returns.
Each stock is scored on four observable factors - momentum, fundamentals, insider activity, and price stability (how calm or volatile it has been). Momentum and insider activity are ranked across the whole universe; fundamentals and stability are ranked within a stock's own sector, because both valuation/quality and volatility are sector-driven and ranking them universe-wide would simply favor whichever sectors are cheapest or calmest. The four subscores are then averaged into a single 1-10 score - so you can always reproduce the Overall from the four numbers shown. There is no black-box model and no return forecast - every input is something you could look up yourself.
What the score is NOT
We'll be direct, because most of this industry isn't: the Signal Score is not a prediction that a stock will beat the market. It does not forecast returns, and a high score is not a buy signal.
We know this because we tested it properly. We ran a point-in-time, survivorship-free backtest - replaying the model quarter by quarter on the stocks that were actually in the index at each date, including the ones that later failed. The result: the score has no reliable ability to predict which stocks beat the market. The stocks it rates highest did not, on average, outperform the ones it rates lowest.
So we won't show you a performance chart that implies otherwise. A rising “our picks vs. the S&P” curve is easy to produce with hindsight and survivorship bias - and it would be misleading. We'd rather tell you the truth and lose the sales pitch.
What AlphaRanks is genuinely good for
The value isn't a crystal ball - it's clarity. AlphaRanks reads everything so you don't have to, and lays it out in plain English:
- Fundamentals straight from SEC filings, point-in-time correct.
- Insider activity - who's buying and selling, from Form 4 filings.
- Risk - how much a stock tends to swing, so you're not blindsided.
- Earnings - what's coming up, and when.
- Momentum and news - where the stock and its story stand right now.
Use the score as a quick, honest read on where a stock's signals stand - one input into your own thinking, never a recommendation.