How Market Grades work

A Market Grade is a letter, from A+ to F, for how well one prediction market trades. It tells you whether you can rely on the market’s price and trade in it at a fair cost. The A range is meant for markets where $50,000 can trade within 5¢ of the mid, a $1,000 trade costs 1% or less, and 30 or more traders take part. F is for markets with no money near the mid, or no trades behind the quotes. These describe what each band is for (known limitation).

​

Why it matters

Two markets can both show a 60% chance and be very different things. One has a deep order book and hundreds of traders. The other has a few quotes and a handful of traders. The price looks the same, but only one is worth relying on. A grade tells you which kind of market you’re looking at.

​

What a grade is not

A grade is about the market, never the event. It doesn’t tell you whether an outcome is likely, or whether the price is too high or too low. It grades one market, never a whole venue. It is not a forecast, an audit, or investment advice. A grade is not a recommendation to buy, sell, or hold any position.

​

How to read a grade

​

The letter

The methodology has the full definition of each band.

​

Liquidity and integrity

Each grade combines two scores, each also shown as its own letter:

• Liquidity: how much money can trade near the mid price, and what a $1,000 trade costs, fees included. It carries 60% of the weight.
• Integrity: how many different traders are active, and how often the order book is broken (one-sided, crossed, or empty). It carries 40%.

​

Caps

Some grades show a cap: a limit on the highest letter the market can get. Quotes alone can’t earn a top grade, so a market with little trading behind its quotes is capped at B+ or C+. And a market that is very weak on liquidity or integrity can’t hide it behind the other, so its letter is capped too. Any cap is shown with the grade.

​

Not Rated (NR)

NR means we don’t give the market a grade right now, and the page says why. NR is not a grade below F. The usual reasons:

• The market is new. A market needs 30 days of observation before it gets a grade. Until then, the page shows its progress, such as “day 17/30.”
• The market ends too soon to build up a record and hold a meaningful grade.
• The market is inactive. No two-sided order book for more than 80% of the last 30 days, and no trades.
• Our data had a gap. If we couldn’t observe the market well enough, we say so instead of guessing.

If a graded market is paused, its last grade stays on display next to the reason.

​

Near the end of a market’s life

Order books thin out as a market nears its end date. So from 14 days before the end date, the letter is frozen and marked Near resolution. A thinning book at the very end doesn’t drag the grade down, and the freeze can never raise it.

​

How often grades change

Grades are computed once a day, from the last 30 days of data. A letter doesn’t flip on one bad day. A downgrade needs the score to move clearly past the boundary two days in a row, and an upgrade needs a bigger move. Before a change, the market is usually placed on Watch (Negative or Positive; shown as “Under review”). A sharp decline can bring a faster downgrade. Every change is recorded with its reason and stays in the market’s history.

​

Dates and versions

Each grade shows the date it was computed for, the 30-day window it covers, and the version of the rules it was computed under. Read a grade as of its date.

​

Independent by design

• Graded markets never pay. No market or venue asks for, pays for, or previews its grade. Belief Systems holds no positions in graded markets.
• One rulebook for every market. The rules are published, and the scoring thresholds are fixed numbers. Markets aren’t graded on a curve, so an A means the same thing every month and on every venue.
• Markets can’t time our checks. We sample order books at random times that are never published. A market can’t look deep only when it expects to be checked.
• The record is never rewritten. Corrections are added as new revisions, and the original stays as published.
• Rule changes are announced first. Every change is posted in the Governance Log at least five calendar days before it takes effect, and never changes past grades.

​

Beta

Grades are live under the published methodology. Its validation runs in public, and each result is published whatever it shows. A grade describes a market as measured. It makes no claim about how the market will trade next until the predictive study reports. The first review after launch is on January 19, 2027. The full schedule is in the validation roadmap.

​

Next