Simulated Markets and Trading Games
Trading simulations recreate the decision environment of a live market without the capital at risk. Firms use them for two distinct purposes: training new traders, and assessing candidates. The design is similar; what is being measured differs.
What a market making game replicates
You are shown an instrument with some underlying fair value, and you must quote a two-sided market. Counterparties trade against your quotes. Fair value drifts, sometimes jumps, and flow arrives unevenly.
The compressed loop is exactly the real one: quote, get filled, accumulate inventory, update your estimate, adjust, repeat. What is stripped away is the infrastructure, the instrument-specific knowledge and the timescale, leaving the reasoning.
What assessors watch
Not P&L, or at least not primarily. A short game is dominated by luck, and everyone running these exercises knows it.
Do you quote sensibly? A two-sided market centred near fair value with a defensible width. Absurdly wide quotes are a failure, not caution.
Do you manage inventory? Getting long and staying long, or worse adding to it, is the single most common failure. Assessors watch whether your position drifts or oscillates around zero.
Do you update on evidence? If you are lifted five times running, your price is wrong. Candidates who keep quoting the same market after repeated one-sided fills are demonstrating exactly the wrong instinct.
Do you explain yourself? Saying "I'm long so I'm skewing down" earns credit even when the amount is debatable. Silent good instincts are invisible.
Do you stay composed? Everyone has a bad run. What matters is whether the response is measured or panicked.
The game measures process, not outcome. A candidate who loses money with clear reasoning and controlled inventory outperforms one who wins by holding a lucky position.
The common mistakes
Quoting too wide from nerves. You never trade, so you demonstrate nothing. Assessors cannot evaluate decisions you did not make.
Ignoring inventory. Focusing entirely on each new quote while the position quietly grows.
Anchoring on the starting price. Fair value moves. Quoting around the opening level twenty trades later is not conviction, it is inattention.
Widening when you should skew. These fix different problems, as covered in quoting mechanics. Widening does nothing about an existing position.
Going quiet. Under pressure people stop narrating. The narration is a substantial part of what is being assessed.
Practising properly
Repetition builds the reflexes, but only with attention to the right things. After each session, ask: was my P&L driven by spread capture or by inventory drift? Did I update fair value when the evidence said to? Was my worst moment a bad decision or a bad outcome?
Our games simulate this loop with a P&L breakdown afterwards, so you can see whether your profit came from spread or from an accidental directional bet. The second kind does not repeat.
One thing the description above leaves out is that the counterparty is choosing too. The exercise is a game of private information, not an estimation test with trading attached, and the market making game as a game takes it apart on those terms: why a pass is evidence, why your width is a claim about your own uncertainty, and why the value conditional on a fill is worse than your unconditional estimate.