Orderbooks Are Fossilized Beliefs

There's a phrase that perfectly captures what a prediction market orderbook actually is: fossilized beliefs.

Every resting limit order on an orderbook represents a belief that someone held strongly enough to lock up real capital. "I believe the probability of this event is less than 35%, and I'm willing to risk money on it by offering to sell at 35 cents." That order sits on the book — frozen, waiting — until someone with the opposite belief comes along and takes it.

The orderbook is not just a price discovery mechanism. It's a geological record of conviction, layered by price level, shaped by time and information.

The Anatomy of a Belief Layer

Look at a typical prediction market orderbook:

ASKS (sellers — people who think probability is LOWER)
  40¢  ████████████  $12,000
  38¢  ████████      $8,000
  37¢  ████          $4,000
  36¢  ██            $2,000
  --- spread ---
  34¢  ███           $3,000
  32¢  ████████      $8,000
  30¢  ██████████    $10,000
  28¢  ████████████  $12,000
BIDS (buyers — people who think probability is HIGHER)

Reading this:

• The spread (34-36) is the zone of uncertainty. Nobody is confident enough to bid 35 or ask 35. This 2-cent gap is where the market genuinely doesn't know.
• The thick bid at 30 is a conviction level. Someone (or many someones) believe the true probability is above 30% and are willing to defend that level with $10,000.
• The thick ask at 40 is the opposing conviction. Someone believes the probability is below 40% and is willing to sell at that level.
• The thin zone at 36 is fragile. A small buy order would move the price to 38. There's no conviction defending 36.

What Depth Tells You That Price Doesn't

The market price — the midpoint or last trade — might be 35 cents. But the orderbook tells you far more:

Asymmetric depth = directional conviction. If there's $50,000 of bids below the current price but only $5,000 of asks above it, the market has more conviction on the buy side. The price is likely to drift up. Buyers are committed; sellers are not.

Walls = price magnets. A massive resting order at a round number (like $100,000 bid at 50 cents) acts as a price magnet. The market tends to gravitate toward large resting liquidity because that's where trades get filled.

Gaps = fragility. If there are no orders between 35 and 45 cents, the price can jump from 35 to 45 on a single trade. Gaps in the orderbook are where price discovery breaks down and volatility spikes.

Layered depth = consensus. When bids are evenly distributed (similar amounts at 34, 33, 32, 31, 30), the market has a broad consensus that the probability is somewhere in this range. No single price level dominates.

The Fossil Record

Orderbooks change constantly, but the pattern of how they change reveals information:

Belief accumulation. Over days, you see limit orders build up at certain price levels. Someone keeps adding to their bid at 30 cents. This is deepening conviction — they're not just guessing, they're building a position. They've done the work. They believe.

Belief abandonment. An order that's been resting at 45 cents for two weeks suddenly gets cancelled. The person changed their mind. What did they see? What information caused them to revise their belief?

Belief collision. Two large orders appear on opposite sides — a $20,000 bid at 32 and a $20,000 ask at 38. Two whales disagree strongly. When they collide (one takes the other's order), the winning side has been validated by the market.

These dynamics are invisible in the price. The price is the tip of the iceberg. The orderbook is everything below the waterline.

For Agents: Reading the Book

If your agent only reads the price, it's reading a headline. If it reads the orderbook, it's reading the full article.

Key metrics an agent should extract from an orderbook:

1. Spread — the gap between best bid and best ask. Narrow = confident market. Wide = uncertain.
2. Depth ratio — total bid dollars vs total ask dollars. Asymmetry = directional conviction.
3. Liquidity score — how much can you trade without moving the price? Important for sizing.
4. Wall detection — unusually large orders at specific price levels. These are conviction anchors.
5. Gap detection — price levels with zero orders. These are fragility points.

Most prediction market APIs expose orderbook data. Kalshi provides full depth. Polymarket provides depth via the CLOB.

Empty Orderbooks Are Information Too

Here's something most people miss: an empty orderbook is also a signal.

A contract with no bids and no asks means: nobody cares enough to risk money on this event. Either:

• The event is too obscure to attract attention
• The event is too far in the future to price
• The market was recently settled or created

An empty orderbook is not a 50/50 probability. It's undefined — the market has no opinion. This is different from a thin market at 50 cents (where the market's opinion is "coin flip") and different from a deep market at 50 cents (where the market's conviction is "genuinely uncertain").

The distinction matters. "No opinion" and "uncertain" are different cognitive states.

The Metaphor Holds

Geological fossils tell you what lived here, what conditions existed, and what events occurred — all frozen in stone. Orderbook fossils do the same for beliefs:

• What beliefs existed — the prices where orders were placed
• What conditions persisted — the depth and duration of those orders
• What events occurred — the trades that consumed those orders, and the price changes that followed

Read the orderbook. It's not just a trading tool. It's a window into what the world believes, how strongly it believes it, and where it's willing to put money behind those beliefs.

---

This is part of a series on prediction markets as cognitive tools. Read the full series at simplefunctions.dev/blog.