How to Use This Book¶
This book is about an exchange that trades on its own. A hundred AI traders, a market maker and a hidden model of what every share is really worth keep 300 order books busy all day, every trading day, for as long as you leave it running. You watch it, teach with it, trade against it, and change how it behaves.
| Who it is for | Instructors who want a live market for a class; students who want a market to trade in; anyone curious how prices form |
| What you need | A computer with Docker or Podman for the container route, or Python 3.13 and pipx for the hand-built route; a terminal and a browser |
| What you do not need | Any finance background, any Python, or a copy of the source code |
| Time | Ten minutes to a running market (Part I); an afternoon to understand and tune it (Part II) |
What you will be able to do¶
After this book you can:
- start a complete simulated exchange — 300 shares, 100 AI traders, a market maker, news — with one command, and leave it running for weeks;
- explain how a price on the exchange can be wrong, what the true value behind it is, and how trading discovers it;
- read the news and rumours the market model publishes, and publish your own to move the market in front of a class;
- change how the traders behave — calmer, wilder, faster to react, more or less informed — and know which setting does what;
- run the calendar faster than real time, so a month of trading passes in a day;
- watch all of it with the browser applications and the command-line tools.
How the book is organized¶
| Part | Chapters | What you get |
|---|---|---|
| I. Start here | A Market That Runs Itself; Ten-Minute Start | What the simulation is, and a running one |
| II. How it works | The True Value; News and Rumours; How the AI Traders Decide; Tuning the Traders | The model behind the prices, the traders, and every setting that changes them |
| III. Run it | AI Traders; Running the AI-Trader Swarm; The Market Model; A Complete Set-up by Hand; The Simulation Container; Accelerated Time | The reference for each part, and two ways to run it all |
| IV. Watch and teach | Watching the Market; AI Traders & News; Running It for Weeks; Troubleshooting and FAQ | What to look at, classroom exercises, and keeping it healthy |
| Appendices | Simulation Glossary | The words this book uses |
Part I is enough to get a market running. Part II is the heart of the book: read it before you change anything. Part III is reference — dip into it when you need a flag or a file. Part IV is for the days and weeks after.
Chapters shared with other books
Five chapters of this book also appear in the book they were written for:
AI Traders and How the AI Traders Decide in the Participant Guide,
Running the AI-Trader Swarm and The Market Model in the Operator's
Guide, and AI Traders & News in the Training Guide. They are the same
text, kept in one place. Their links point into those books, and their
examples use the smaller s150-* examples, started by hand. What they
explain holds for the s300-sim set-up this book runs; their commands do
not all fit the container. In the container, skip the steps that deploy
an example (the container has deployed s300-sim, and the AI traders need
it), and remove the market maker as
Tuning the Traders
shows rather than with Ctrl+C: the container would restart it. Scenario
files are in the image under
/opt/edumatcher/.venv/lib/python3.13/site-packages/docs/examples/news/.
The set-up this book runs¶
Everything here uses one bundled example, s300-sim, and one process
profile, ai-sim:
| Symbols | 300 shares in 11 industry sectors |
| AI traders | AI001–AI100, eleven kinds (presets), each trading 30 symbols |
| Market maker | MM01, a bot quoting every symbol, nudged toward the true values |
| People | TRADER01–TRADER10 (desks for students), OPS01 (the instructor) |
| Market model | A hidden true value per share, with random news and rumours |
| Hours | 09:00–16:05, Monday to Friday, on the Swedish calendar and clock |
The container image edumatcher-ai-sim is exactly this set-up, ready to run.
Conventions¶
- A bare command in a
bashblock is typed in a normal terminal. Commands that run inside the simulation container say so, or start withdocker exec edumatcher-sim. From an installed simulation (cd ~/.edumatcher-sim),./edumatcher.sh shell <command>is the same thing and works with Podman too;./edumatcher.shalso wrapsstart,stop,statusandlogs. Where the book says./data-sim, an installed simulation's data directory is~/.edumatcher-sim/data. [OPS01|ADMIN]>is the prompt of the operator console (pm-admin);[TRADER01]>the prompt of a trader console (pm-alf-console).- Output shown in a plain block is what you should see. Prices, times and counts will differ on your machine: the market is random, and so is the moment you look at it.
- Every number this book quotes about the simulation — how far prices stray, how much disk a day takes, how much CPU it needs — was measured on the set-up it describes. Where a number depends on your machine, the book says on what it was measured.
!!! tip,!!! noteand!!! warningboxes hold advice you can skip on a first reading; Checkpoint paragraphs tell you how to know a step worked.
Where this book sits¶
This is one of nine EduMatcher books. If the exchange itself is new to you, read the Quick Start Guide first — an hour, and every chapter here will be easier. If markets are new to you, How an Exchange Works explains order books, auctions and market makers from scratch.