Method

How the stock score works

StocksScores gives every US stock one number from 0 to 100. It is plain arithmetic, not a model: 122 scorers each turn one measure into 0–100, and weighted averages fold them into a fundamental and a technical total, then into the composite. This page is the whole method. The counts and weights on it are read from the engine that scored last night’s stocks, so it cannot drift from what actually runs.

  • 80%

    fundamentals, ranked within sector

  • 20%

    technicals, across 6 timeframes

  • 0%

    news sentiment: shown, not counted

  • 104

    scorers that count, of 122

01 · Data

What goes in

Every night the engine collects, for each of about 4,118 US-listed companies: price and volume history on several chart timeframes, quotes, financial statements, analyst estimates and price targets (from Yahoo Finance); company news, social sentiment, insider sentiment, congressional trades, SEC filings and sector metrics (from Finnhub); and news and earnings-call transcripts (from Alpha Vantage).

2,352 of those companies are worth $1B or more and have both a composite and a fundamental score; they have public, indexed pages under stocks. Smaller companies are scored the same way and shown in the app.

02 · Scorers

Each scorer returns 0 to 100

A scorer reads one measure and maps it to 0–100, where higher is always more bullish. That holds even when the good reading is a low one: a cheap P/E or low debt scores high. Every scorer and what it does is listed in the scorer glossary.

When the data a scorer needs is missing, it returns nothing. It is never filled in as 0, which would read as maximally bearish, or as 50, which would pretend to know. A missing scorer simply drops out, and the average is taken over the scorers that have data.

03 · Fundamentals

The fundamental pillar: 80% of the score

39 fundamental scorers count, each with a fixed weight. By share of the pillar’s weight they group like this:

Shown for context but weighted 0: Fair Value, Revenue Quality (Annual), Sector Percentile, Institutional Ownership, Fund Ownership and Insider Transactions.

Ranked within the sector

Fundamental ranges differ by industry: a bank’s margins, a utility’s debt and a biotech’s earnings are not comparable. Scored against one market-wide yardstick, whole sectors drift up or down. On the November 11, 2026 batch, Energy averaged 65 on fundamentals and Healthcare 49, and Financials made up 21% of the large companies but 43% of the top 100.

So the fundamental total is restated within its sector. The engine finds where a company’s raw total ranks among its own sector’s, then gives it the score that sits at the same rank across the whole market. A bank at the 90th percentile of banks gets the score of a 90th-percentile company. Every sector ends up with the same spread, the order inside a sector never changes, and in our data the restated total ranked future returns better than the raw one. Sectors with fewer than 30 companies keep the raw total.

04 · Technicals

The technical pillar: 20% of the score

65 technical scorers run on each chart timeframe, and the timeframe totals are blended with these weights:

TimeframeWeightShare
30-minute0.55%
Hourly110%
Daily329%
Weekly219%
Monthly219%
Quarterly219%

On the daily chart the scorers group like this:

Technicals carry 20% because that is what the evidence supports. In our returns test, the most technically bullish stocks went on to trail the least bullish, so the pillar describes the chart rather than steering the score.

05 · News

News sentiment: shown, not counted

12 news scorers read article sentiment over windows from three hours to a week, social sentiment, earnings calls and SEC filings. Every stock page shows them. None of them moves the composite: in testing, sentiment shifted the score around without improving how it ranked stocks, so it is excluded in the engine itself, and no preset can put it back.

06 · Composite

One number, and how to read it

The composite is the weighted average of the two pillars: 4 parts fundamental to 1 part technical. A stock with a fundamental total of 70 and a technical total of 50 scores (70 × 4 + 50 × 1) ÷ 5 = 66. If one pillar has no data, the composite is the other pillar alone.

ScoreReads as
80–100Very bullish
66–79Bullish
50–65Neutral
40–49Weak
0–39Bearish

The labels describe what the measured signals say. They are not trade instructions. The highest composites today are on the top-scored stocks list, and each stock page breaks its number down to the scorers behind it.

07 · Testing

Does it work? We publish the answer

Every nightly score is stored with the price at that moment, so the method can be checked against what stocks did next, with no hindsight. Our first test, over four months of published scores, found the earlier composite had no edge. Fundamentals ranked within their sector were the one consistent signal; technicals pointed the wrong way, and news added nothing. The score was rebuilt around that on November 12, 2026. The same test will be rerun on the new score once it has four months of its own history, and published whichever way it comes out.

For background on scores in general, read what a stock score is and how to read one. The raw data for all of it is in the free stock API.

08 · Limits

What the score cannot do

  • It does not predict prices. It summarises measured signals on one date.
  • It is only as good as its data. Vendor errors, late filings and restatements flow straight through.
  • It knows nothing that isn’t in the numbers: management, products, lawsuits or the next earnings call.
  • It moves every night as prices and data change; a single day’s reading is noisy.
  • It is information, not investment advice.

Questions

What is a good stock score?

On StocksScores, 66–79 reads Bullish and 80–100 Very bullish; 50–65 is Neutral, 40–49 Weak and below 40 Bearish. A score is relative: it says how a stock's measured fundamentals and technicals compare with every other stock scored the same way that night. It is not a forecast and not a buy rating.

How often are stock scores updated?

Every night. The batch starts at 21:30 UTC, after the US close, and scores the whole universe in about four hours, weekends included. Each stock page shows the date it was scored.

Is a high stock score a buy signal?

No. In our own test of four months of published scores, the earlier composite had no edge over the following four weeks; fundamentals ranked within their sector were the one consistent signal, which is why the score is now built mostly from them. Treat the score as a summary to start research from, not as advice.

Why does a stock have no score, or a missing pillar?

A scorer whose data is missing returns nothing and drops out of the average; it is never counted as zero. If a whole pillar has no data the composite is built from what remains, and stock pages say how many scorers were left out. Companies under $1B are scored in the app but their public pages are not indexed.

Can I get the scores through an API?

Yes. The nightly scores and their history are served by a public JSON API with no key, and an MCP server lets AI assistants read them. See the free stock API guide.