Live Betting Basics: How In-Play Markets Move
Pre-Match and In-Play Are Not the Same Problem
Pre-match analysis is a research problem. The information set is close to fixed — squads, form, schedule, venue, the price the market has settled on. You can sit with it for an hour and the market will still be roughly where you left it when you look up.
In-play is not that. Live markets are a problem of continuous revision under time pressure. The real probability of every outcome in the match is being rewritten by events as they happen: a goal, a red card, a substitution that changes the shape of a side, twenty minutes of one-way pressure that has not produced anything yet. The price moves with those events, and it usually finishes moving before you have finished thinking.
That has one important consequence. In-play, the question is almost never which team is better. It is whether the price has adjusted correctly to what just happened, and to what has been happening for the last ten minutes. Those are different questions from the pre-match ones, and they need different machinery to answer.
Why Live Markets Move So Fast
Four forces, all running at the same time.
Discrete events reprice everything at once. A goal in the seventieth minute does not nudge a market, it rewrites it. Every handicap line, every goals line, every side's price is recalculated in the same instant, because one event changed the score, the remaining time and both teams' incentives simultaneously.
Time is itself a variable. In a pre-match market nothing decays. In-play, every minute that passes deletes chances that could have happened. A goals line that looked generous at kickoff looks completely different on the hour with the same score on the board, purely because there is less match left. This is the most underrated driver of live prices, and it is the one beginners account for least.
Markets suspend and reopen. Around any significant event — a penalty award, a red card, a goal under review — live markets stop taking action and then reopen at a new price. The position you thought you were about to take frequently does not exist by the time the market comes back.
Your picture is behind the pitch. Whatever you are watching reaches you after the event actually happened, and how far behind depends entirely on your feed. Structurally, you are reacting to a past state of the match while the market reacts to the present one.
Put those together and live pricing stops looking chaotic. It is a fast, orderly process — just faster than a person watching a stream can participate in on equal terms.
Three Structural Problems With Watching It Yourself
These are not discipline failures. They are properties of being a human being with one screen.
You are always a step late
By the time you have seen an event, understood it, formed a view and reached for the market, the market has already repriced. This is not about reaction speed you could train. The delay is built into the chain — feed, perception, judgement, action — and each link costs seconds that the market does not spend.
Your attention has a hard ceiling
On a busy evening there are dozens of matches in play at once, each with several markets moving independently. A person can genuinely watch one match, and can perhaps glance at a second. Everything else is invisible. That is not a small limitation: the whole value of a live approach comes from comparing many simultaneous situations and acting on the small number that are mispriced. Watching one match means you are choosing from one match.
You decide while you are emotionally involved
The moment that most tempts a live decision is almost always the moment right after something has gone against you. Conceding an equaliser, watching a chance go begging, seeing a position that was comfortable ten minutes ago turn awkward — those are exactly the moments when people reach to fix things. A model has no position it is trying to rescue. It reprices and moves on, which is the single largest behavioural difference between a system and a person.
How the Agent Handles In-Play
The OddsFlow network was built around those three constraints rather than in spite of them.
It reads the clock as data. Every in-play card carries the elapsed minute at which the signal fired. That is not a label for readers. Remaining time is one of the model's inputs, because the value of any live line depends on how much football is left to play.
It ingests prices continuously. Odds across a large number of sources are pulled in real time and compared as they move, on a millisecond scale rather than a "check it every few minutes" scale. The Market Divergence Scanner is the component that watches for one source moving before the rest, or the same line being priced differently in different places.
It watches the match, not the highlights. The Momentum Shift Detector works on the underlying event stream — shots, territory, pressing intensity — to find the inflection points that should cause a live reprice, including the ones that have not yet produced a goal.
Most of the work is deciding not to publish. Roughly one candidate in twenty-two survives the full filter. On a busy night the model examines a great many live situations and stays silent on nearly all of them. Silence is the system working; a live channel with a signal every ten minutes is telling you it has no filter.
What comes out is a structured card with a fixture, a line, a side, a published price and a minute stamp — every field of which is broken down in How to Read an OddsFlow Signal Card: Every Field Explained. If the line notation on those cards is unfamiliar, the quarter-ball scale is covered step by step in Asian Handicap Explained: Lines, Quarter Balls and Price.
In-Play Discipline
Five habits that make live signals readable rather than stressful.
- 1.Read the minute first. Before anything else on the card, check when it fired and how long ago that was. An in-play card ages in minutes, sometimes in seconds. A pre-match card does not.
- 2.Check the price is still there. The number on the card is the published price, and it is the number the record is settled against. If the market has moved past it, the situation the card described no longer exists — that is a normal outcome, not a failure.
- 3.Never chase the last event. The event you just watched is the single most heavily priced piece of information in the market. By definition you are late to it.
- 4.Accept quiet nights. A selective filter produces long gaps by construction. Treating a quiet evening as a malfunction is how people talk themselves into taking the candidates the filter rejected.
- 5.Settle everything afterwards, winners and losers. Judging live signals one at a time teaches nothing, because variance dominates any short run. What a record means, and how wins, losses, half settlements and refunds are counted, is set out in How We Count Our Record: The Formula and the Timestamps.
The Short Version
In-play is not pre-match with the clock running. It is a different problem: continuous repricing, driven by events and by time itself, moving faster than a person with one screen can follow. The honest way to approach it is to stop trying to win the speed contest and start reading structured, timestamped output that can be checked afterwards against a real final score.