Stock prices move for three reasons: new information, large orders, and the mechanics of how orders meet each other. The first gets the headlines. The second and third do most of the minute-to-minute work, and almost none of it is done by a human pressing a button. When you check stock market news today and see a stock up two percent by mid-morning, the odds are good that the move was shaped less by conviction and more by the plumbing.
This piece is a tour of that plumbing. It explains who market makers are, what the algorithms are actually doing, and why order flow — the stream of buy and sell instructions — matters more than opinion. One caveat up front: none of this tells you where a price will go next. It tells you why the number on your screen changed.
What does a market maker actually do?
A market maker is a firm that stands ready to buy and sell a stock at quoted prices, nearly all the time. It posts two numbers: a bid, the price it will pay, and an ask, the price it will accept. The gap between them, the spread, is its compensation for the service and for the risk of holding shares for a few seconds or minutes.
The mechanism is simple and the risk is real. If a market maker buys from every seller just before bad news lands, it owns shares worth less than it paid. That is the core liability of the business, and it explains most of its behavior. Spreads widen when uncertainty rises, because the cost of being caught on the wrong side goes up. In calm conditions the spread on a large, heavily traded stock can be a penny or two. In a panic it can widen sharply, which is one reason prices can gap between trades.
Market makers are not in the prediction business. They are in the inventory business. Their job is to be the other side of your trade, manage what they accumulate, and do it thousands of times a day without accumulating the wrong book.
What are the algorithms doing with the flow?
Most orders never touch a human. Algorithms slice a large parent order into small child orders, release them over time, and try to blend into normal trading so the market does not notice. The goal is to avoid paying for impact: if the whole market sees a giant buy order, the price rises before the order finishes, and the buyer pays more on every remaining piece.
Other algorithms do the opposite job. They detect the footprints of large orders — a persistent tilt toward buying at the ask, for example — and trade ahead of the expected continuation. This is the adversarial heart of modern markets: one side hides its intent, the other side hunts for it. Neither side is cheating in the legal sense. Both are responding to the same visible data.
The practical consequence for anyone reading a quote: the price you see is the residue of this contest. It reflects not just what people want, but how successfully they have hidden that they want it.
Where does the flow come from?
Order flow arrives from several distinct populations, and they behave differently. Retail orders tend to be small, uninformed about anything not yet public, and often routed to wholesalers who internalize them — matching them against their own inventory without sending them to the public exchange. Institutional orders are large, patient, and handled by the execution algorithms described above. Then there are fast traders, who hold positions for seconds and provide much of the visible liquidity.
These populations interact. A wave of retail buying that wholesalers internalize may never appear on the exchange tape, which is one reason the tape can understate demand. A large institutional order leaked into the visible market can trigger the detection algorithms, and the resulting momentum can look, from the outside, like a news-driven move when it was really an execution problem.
The word "actually" earns its keep here. As Dictionary.com's usage notes explain, the adverb signals that reality differs from expectation — and market headlines routinely describe the expectation rather than the actually-mechanics underneath. A stock "jumps on earnings" is often true. It is also often true that the jump's size was set by how much order flow had already quietly positioned before the release.
What happens when the machinery is stressed?
The plumbing has failure modes, and they are well documented in kind even when the details of any single episode stay debated. Liquidity is conditional: market makers quote tightly when conditions are normal and pull back when volatility spikes, precisely when traders most want to trade. Order book depth can thin out in seconds. Detection algorithms can amplify a move that started from one large execution.
Exchanges and regulators respond with mechanisms designed to slow things down rather than stop them outright — price bands around recent trades, pause rules for individual stocks, and market-wide halts at defined thresholds. For a walkthrough of those safeguards, see our explainer on how market circuit breakers actually work when panic hits. The design logic is consistent: give the information-processing side of the market time to catch up with the price side. Readers following this should also see How Do Market Circuit Breakers Actually Work When Panic Hits?.
The honest summary is that the machinery works well at absorbing ordinary flow and imperfectly at absorbing simultaneous, correlated flow. That limitation is structural, not a bug anyone has found a clean fix for.
What this means for how you read the tape
Our analysis, based on how these mechanisms fit together rather than on any single data release: treat short-term price moves as output, not as verdicts. A two percent move in a liquid stock on no news is usually flow mechanics — an imbalance between resting orders and arriving orders — not a change in what the company is worth. A large move on real news is information. The size of the move, though, still carries the fingerprint of positioning that happened before the news.
Three practical reading habits follow from the mechanism, not from any directive:
- Check whether a move came with volume and spread widening. Thin, gappy moves on wide spreads say more about liquidity than about demand.
- Distrust explanations that attribute every intraday wiggle to a cause. Some wiggles are just the machinery clearing imbalances.
- Separate signal from plumbing. Our guide on telling market noise from market signals covers that discipline in detail.
None of this is a forecast. Prices will keep being set by the interaction of information, inventory and flow, and the mix on any given day is not knowable in advance. What the mechanism gives you is a better set of questions to ask when the tape moves.
Where this fits in the bigger market picture
The plumbing connects to everything else covered on this site. Index rebalancing creates predictable, dated flow that execution desks plan around, as our piece on what happens when the big stock indexes rebalance explains. Sentiment measures built from options prices feed back into hedging flow, which is the subject of what the VIX is and why traders call it the fear gauge. And the day-to-day output of all this machinery is what we cover in Markets News, where each move gets tied to the release, auction or flow behind it. This connects to our earlier piece, What Actually Happens When the Big Stock Indexes Rebalance?.
The through-line is the same one this article started with: behind every quote is a counterparty, an inventory position, and an algorithm with a job. Read the plumbing, and the tape starts making more sense.
