Match Story

The Match Story tab turns a map into a readable narrative: a chronological account of how the map was won or lost, fight by fight. It exists to answer the one question every coach asks after a scrim — "why did we lose?" (or "why did we win?") — without making you scan raw stats and reconstruct the flow yourself. Match Story walks you through what happened, when the map swung, and why each swing happened. It sits on the Match Story tab of a scrim's map page.

At the heart of Match Story is a win-probability model. As the map progresses, the model estimates each team's chance of winning from the current game state. Match Story uses those estimates to find the moments that mattered and to assign credit and blame in plain terms.

Win Probability and WPA

Two ideas show up throughout the tab, so it helps to define them up front in coaching terms.

• Win Probability (WP) is the model's estimate, at a given moment, of one team's chance of winning — the map on Escort, Hybrid, and Flashpoint, or the current round on Control. It is read from your team's (Team 1's) perspective, so 70% means your team is favored and 30% means you are behind. WP moves as the state of the map changes: kills, the objective, and ult banks all push it up or down.
• WPA (Win Probability Added) is how much a player moved that number over the course of the map. A positive WPA means their actions helped their team's chances; a negative WPA means they hurt them. WPA is a way of asking "who actually swung this map?" rather than just "who had the most damage?"

WPA is also rolled up at the scrim level. The scrim overview has a Win Probability Added section that sums each player's WPA across every map in the scrim, so you can see who carried the day across the whole block. See Scrims for where that lives.

A note on what WPA is — and is not. WPA distributes each fight's swing across the players who took part in it, by convention: final blows and assists earn positive credit, deaths carry negative weight (the first death in a fight weighted heaviest, since it usually starts the collapse), and ults spent earn credit when the fight is won and carry a cost when it is lost. It is an attribution, not a causal verdict — it tells you who was involved when WP moved, not who is to blame in the film-room sense. Because of that, Match Story always exposes the per-fight breakdown behind a player's number, so you can audit where a WPA came from rather than taking the total on faith.

How the Map Unfolded

The tab opens with a short narrative strip that calls out the beats of the map in order, followed by a win probability timeline chart. The chart plots WP across the whole map with markers on each fight, so you can see at a glance where the map was close, where one team pulled ahead, and which fights flipped the trajectory. Hovering a fight in the ledger below focuses it on the chart.

The Fight Ledger

The fight ledger is the spine of Match Story: a fight-by-fight table where every engagement gets a row. Every column is sortable, so you can re-order the map by whatever you are studying. By default it sorts by the biggest swings first.

The columns are:

• Fight: the fight number, in order.
• Time: when the fight happened, in map time.
• Zone: the map location the fight took place in, when that is known.
• Result: whether Team 1 won, lost, or the fight was even.
• Swing: how much win probability the fight moved, shown as a signed percentage with a bar that extends toward the team that gained. This is the single most useful column for finding the fights that decided the map.
• Ults: how many ultimates each team spent in the fight, shown as a Team 1 to Team 2 count.
• Context: objective captures from the fight plus carryover effects such as ult economy and stagger that were paid into the fight before it even started.

Because the swing column is driver-aware, hovering a row breaks the swing down into how much of it came from the objective, from kills, and from ults. That breakdown is what powers the takeaways below.

Carryover Between Fights

Fights are not independent. When you lose a fight by burning three ults for nothing, or you get staggered and a teammate is still respawning, you do not start the next fight from even — you start it already behind. Match Story estimates that inherited disadvantage as a carryover cost: the slice of a fight's starting win probability that was handed to you by the previous one, split into an ult economy part (you entered down on ult banks) and a stagger part (you entered down a body). It surfaces as a badge in the Context column only when the carried cost is large enough to matter, so you can see how a single thrown fight snowballed into the next one or two. The same mechanism flags the good version too — a clean win that lets you steamroll the follow-up.

Driver-Aware Coaching Takeaways

Match Story does not just tell you a fight swung; it attributes the swing to its drivers. For each fight, the model isolates the contribution of three things — the objective, kills, and ults — and the percentages always add up to the fight's actual swing. That makes the difference between "we lost the fight" and "we lost the fight and it converted straight into a capture."

The takeaways strip turns those drivers into recurring, fixable patterns across the whole map rather than one-off observations. Depending on what the map shows, it can surface things like:

• Where your losses came from — for example, most of your lost win probability across your losing fights came through the objective, the kill exchange, or your ults.
• Ult discipline — losing fights where you spent more ults than the enemy, which fund their next push.
• Staggers — fights you entered down a player because deaths bled across the reset.
• Ult deficit — fights you took while behind on ult economy.
• First deaths — a player who repeatedly died first in your losing fights.

Each takeaway is tied to the fights it came from, so you can jump straight to the evidence instead of taking the summary on faith.

Missed Opportunities

The Missed Opportunities section highlights the fights that should have gone your way and did not: moments where your team went into a fight as the favorite and then lost a large chunk of win probability. A fight qualifies when you entered it at least slightly ahead (favored to win) and then bled a meaningful amount of WP out of it — small wobbles from a lead do not count, only drops big enough to be worth a film review. The list is ranked by how much WP you lost, biggest first, and caps at the few most costly fights with the full count shown. These are the point-flip moments where the map turned against you despite a lead.

Each entry shows:

• Why it slipped — short tags for the primary driver of the loss (objective, kills, or ults), an early first death if one happened, and any stagger or ult-deficit you carried in.
• A per-fight ultimate breakdown — which ults were spent at that decisive turn and what each one returned, so you can see ults that got value, ults that were wasted, and ults whose owner died mid-cast. A summary line shows how many of the spent ults actually converted.
• The swing itself — the WP you held going in, the WP you were left with, and the size of the drop.

This section only appears when the model actually finds these moments; if the map had none, it is left out.

How to Use It

Start with the timeline and the ledger sorted by swing to find the two or three fights that decided the map. Open those rows to see what drove the swing, then read the takeaways to check whether the same driver shows up across multiple fights — that is the pattern worth a VOD review. Missed Opportunities is the fastest way to find leads you threw away, and the WPA table tells you who to talk to. From there, jump into the Replay Viewer or Events to see exactly what happened in a decisive fight.

About the Win-Probability Model

The model is retrained every week on league-wide scrim data and only ships when it stays calibrated and beats the previous week's version, so the numbers are guarded against a bad update. A few things are worth knowing so you read them well:

• What it scores. On Control it estimates your chance of winning the current round; on Escort, Hybrid, and Flashpoint it estimates your chance of winning the map. Those four modes are supported. Push and Clash are not — the logs carry no usable progress signal for Push, and Clash is a retired mode — so neither gets a Match Story.
• It scores states, not decisions. The model can see that the ult banks changed and the fight was lost; it cannot know an ult was "wasted." When Match Story flags a wasted ult or a missed opportunity, it is reading the outcome — the judgment about whether the decision itself was wrong is still yours in the film room.
• Some inputs are approximate. Respawn timing is modeled as a fixed delay, so stagger and carryover costs are estimates rather than exact figures. The model also does not use player positions or hero identity yet — both were tested and did not improve its accuracy; positional data is the next avenue being explored.