How a rating
is made

Every evening Tessera ranks about 2,000 US companies on ten research signals, averages the ten ranks and reads the rating off a fixed ladder.

Ten percentiles in, one rating out

  • 95th
  • 88th
  • 55th
  • 77th
  • 40th
  • 15th
  • 62nd
  • 70th
  • 81st
  • 58th
FairAverage 64th percentile · composite +0.28

Many ranks,
one rating

The universe is the 2,000 largest US common stocks on each date; funds, ADRs and non-US issuers are left out.

  1. Rank

    Each signal ranks every company on the same evening, from the 0th to the 100th percentile.

  2. Average

    The ten percentiles are averaged with equal weight, and at least three must be present. The average becomes a composite from −1 to +1.

  3. Rate

    Fixed bands of the composite give the rating: Exceptional, Strong, Fair, Weak or Avoid.

Ten research
signals

Quality minus junk

Profitable, growing, safe companies that pay their owners rank high.

Asness, Frazzini and Pedersen, 2019

Mispricing

Nine of the eleven documented anomalies, averaged into one rank.

Stambaugh and Yuan, 2017

Net payout

Dividends plus buybacks, less new shares, against market value.

Boudoukh, Michaely, Richardson and Roberts, 2007

Low volatility

Three years of weekly prices; the steadier, the higher.

Blitz and van Vliet, 2007

Profitability & value

Gross profit over assets, ranked with book value over price.

Novy-Marx, 2013

Earnings drift

After a report, prices tend to keep moving the way the surprise pointed.

Chan, Jegadeesh and Lakonishok, 1996

Intrinsic value

A discounted-cash-flow value against the price; the more upside, the higher.

Frankel and Lee, 1998

Asset turnover

Sales over total assets; companies that do more with what they own rank higher.

Tessera admission test, 2026

Piotroski F-score

Nine yes-or-no checks on profit, cash flow, leverage and efficiency.

Piotroski, 2000

Free-cash-flow yield

Free cash flow against market value; more cash per dollar of price ranks higher.

Lakonishok, Shleifer and Vishny, 1994

  • Exceptional+0.51 and up
  • Strong+0.39
  • Fair−0.24
  • Weak−0.41
  • Avoidbelow −0.41

The ladder

Each rating is a fixed band of the composite, so the same score always earns the same word.

Glass and gold mosaic tesserae sorted in rows by colour
Tesserae sorted by colour before they are set: the same work the ranking does each evening. Photograph: The Met, public domain.

Straight answers

Red, white and gold glass tesserae laid out in three rows
Glass tesserae, The Met, public domain.
Is a rating a recommendation to buy?

No. It is a relative rank of published research evidence on one evening. It cannot establish fair value, remove risk or predict a return, and it is not investment advice.

Why does a company have no rating?

Either fewer than three of the ten signals could be computed, or it sits outside the 2,000 largest US common stocks. Funds, ADRs and non-US issuers are not covered.

How often does it change?

It is recomputed after every US close, in the evening run that starts at 18:00 New York time. A change of rating is counted once it is confirmed.

Does the market regime change ratings?

No. The rating is read off the composite alone. Some portfolio policies hold more cash in a weak market; a company's rating never moves with the regime.

Are some signals weighted more?

No, all ten count equally. Weights are not tuned; a new signal joins only after it passes the same admission rule as the ten, fixed in advance.

Is the history point-in-time?

Yes. Each evening uses only figures available on that date, with statements dated by their filing, so a later restatement does not rewrite the past.

What does a composite of zero mean?

An average rank across the signals present, not a weak business. Fair covers a wide middle of the ladder for that reason.

Nine guides
worth reading

Try it on
a company

The 200 largest companies are free to look up, no account needed. The full page adds the ten signals, the financials and the history.

Open a company