AI Prediction Markets
0/15 complete

Module 01 · Prediction Markets 101

Reading Order Books Without a Finance Background

An order book is a snapshot of displayed buy and sell interest at price levels. It is not a promise that every displayed order will remain available or that a paper decision could execute at the midpoint. Learn to describe liquidity before interpreting movement.

// concept

Read One Side at a Time

The best bid is the highest displayed buy price; the best ask is the lowest displayed sell price. Their difference is the spread:

// prompt — copy me2 lines
spread = best ask - best bid
midpoint = (best ask + best bid) / 2

These calculations require valid non-empty sides. A midpoint is not an executable price. Buying generally interacts with asks; selling interacts with bids. Size at the top level can be small, so a larger hypothetical order may cross several levels.

// concept

Measure Depth and Hypothetical Slippage

For education, simulate walking the book without submitting anything. Given a paper size, consume visible price levels until filled, then compute volume-weighted average price (VWAP):

// prompt — copy me2 lines
paper VWAP = sum(level price × paper fill size) / total paper filled size
paper slippage = paper VWAP - chosen reference price

This remains an estimate. Orders can be added/cancelled, latency matters, fees may apply, and the book snapshot may not represent availability when a real request reaches a venue.

// concept

Interpret Movement

A price change can reflect new information, one participant, thin liquidity, canceled orders, mechanical repricing, or noise. Compare trade history, depth, spread, timestamps, related markets, and primary-source events. Do not let an LLM invent a narrative from two snapshots.

Use consistent sampling. A one-second stream and a daily snapshot answer different questions. Record server/local time, dropped messages, reconnections, sequence/order where documented, and schema changes.

// worked_example

Worked Example

Book A has bid 0.49, ask 0.51, and meaningful size across nearby levels. Book B has bid 0.35, ask 0.65, with tiny top size. Both have midpoint 0.50, but B provides far less precise/executable information. A chart of midpoints alone hides that distinction.

A paper ledger records the snapshot, hypothetical size, average fill, spread, and unfilled remainder. It never calls the midpoint a completed trade.

// failure_cases

Failure Cases

7 cases to diagnose

  • Calling midpoint the price you can obtain.

  • Ignoring top-level size and deeper levels.

  • Comparing snapshots taken at different times.

  • Treating canceled displayed liquidity as guaranteed.

  • Inferring news from price movement without sources.

  • Forgetting fees or contract tick/precision rules.

  • Turning a read-only exercise into order placement.

// pakistan_angle

Pakistan Angle

Slow or unstable connectivity can make a snapshot stale. Measure retrieval time and do not present an old order book as current. Use UTC internally and add PKT display only with explicit conversion.

Paper examples can show PKR equivalent for learning only if the exchange-rate source/date is labeled. Do not translate hypothetical price movement into an income claim or encourage deposits.

// hands_on

Hands-On Exercise

Capture ten read-only order-book snapshots for one market at a declared interval. Calculate spread, midpoint, top-level size, depth within two price bands, and paper VWAP for two hypothetical sizes. Note staleness, missing sides, and what cannot be inferred. Place no order.

// completion_rubric

Completion Rubric

3 grading bands

  • Complete

    bid/ask/midpoint/depth/paper VWAP are distinguished and every snapshot has timestamp and limitations.

  • Needs revision

    arithmetic is correct but book changes, missing levels, or hypothetical-fill caveats are weak.

  • Not complete

    midpoint is called executable or the exercise submits an order.

// sources

Sources