Research · 01
Market Microstructure
A price is not a number the market has an opinion about. It is the outcome of a queue, a set of rules, and someone deciding they would rather be done than wait. Microstructure is the study of that mechanism, and of what the record it leaves behind can honestly be asked.
The mechanism
How intention becomes price
Almost every modern exchange is a continuous double auction: buyers and sellers post bids and offers simultaneously and continuously, and a trade happens the moment one side is willing to cross to the other. Nearly everything that follows — the spread, the queue, the cost of immediacy — is a consequence of that one arrangement.
The order book
Limit order books, queue position and priority; what the spread actually compensates; how a market order's cost divides between the quoted spread and the impact it causes; why the same trade costs differently at different times of day.
Information and flow
How the arrival of informed and uninformed order flow shows up in the observable record, what can and cannot be inferred from a tape without seeing the book, and the difference between a price that moved and a price that was moved.
Liquidity as a variable
Depth, resilience and the speed at which a book refills; why liquidity is least available exactly when it is most wanted, and what that does to any strategy whose backtest quietly assumed it was there.
Cost, honestly
Spread, impact, slippage, fees, taxes and the fills that would not have happened. The six costs that decide whether an idea survives contact with a real book, treated as first-class objects rather than a footnote.
Price formation
What the record can be asked
The observable record of a session is a great deal thinner than it looks. Much of the work is establishing what a given dataset genuinely supports before asking it anything.
What the data actually contains
The gap between a full book and a public tape; what is reconstructable and what is inference dressed as observation; and the venue and session mechanics — boundaries, auctions at the open and close, contract rolls — that quietly corrupt a study when treated as ordinary time.
Regularity, or its appearance
Most apparent structure in market data is what randomness looks like when you go looking. The question worth asking is never whether a pattern exists but whether it survives a null model fixed in advance and a count of the variants tried.
Intraday behaviour
How spreads, depth and volatility vary through a session, and how much of what looks like a signal is the shape of the trading day reasserting itself.
Execution as the test
A result that survives frictionless arithmetic and nothing else has not survived. Execution realism is where most microstructure findings are decided, and it belongs in the design rather than in a caveat at the end.
Boundary
This page names the subject and the questions. It does not name the techniques currently in use, their parameters, or the instruments and horizons under study — see the publication policy. What gets published from this area arrives as a finished study or a monograph, with its method and its null attached.