What makes a 9.8: how the Buzzr Score works

Every live game in Buzzr ends with a single number between 1.0 and 10.0. It's the first thing a fan sees , on the feed, on the game card, on a friend's post-game text. It is not a prediction. It is not a win probability. It is not a power rating. It is an entertainment score: the answer to the question "was this game worth watching?"

This post explains, in as much detail as is useful, how we get to that number.

The north star: entertainment, not outcome

Almost every number attached to a sports game today , spreads, over/unders, power ratings, ELO, moneylines, ESPN's Matchup Quality , is ultimately about predicting or evaluating who won and how much. Those are valuable numbers. They are not the number most fans want when they wake up on a Sunday and ask "is last night's game worth catching the replay of?"

That question has a different shape. It cares about swings, not margins. It cares about context, not raw production. A 40-point NBA blowout with two stars sitting is technically a game; it is not one you'd recommend to a friend. A 118-117 triple-overtime chaos opera with three lead changes in the final 30 seconds is rewatchable a decade later.

The Buzzr Score is our attempt to build the number for that second kind of question.

The three dimensions

Every Buzzr Score decomposes into three sub-scores, each also on a 1–10 scale: Chaos, Energy, Drama. You can see all three on any game card in the app, and the overall score is a weighted blend of them.

Chaos (unpredictability)

Chaos measures how hard the game was to forecast as it unfolded. Our inputs include:

• Win-probability variance. We track modeled win probability minute-by-minute (for clock-based sports) or possession-by-possession (for paced sports like basketball and hockey). High variance , the number swinging from 80/20 to 30/70 to 60/40 , produces high chaos.
• Lead changes and ties. More lead changes = higher chaos, with a nonlinear weight on changes in the final 25% of regulation.
• Score-margin oscillation. Measures how much the margin "whips" , a game that bounces between +8 and -6 scores higher than one that drifts linearly from 0 to -12.

A game that was a comfortable blowout from tip to final whistle , pick your most one-sided playoff loss , lands in the 3–5 range on chaos. A best-of-seven Game 7 that goes to double-OT with the lead changing eight times lands in the 9–10 range.

Energy (the room)

Energy is a proxy for how much the game mattered to the people watching it , not how many people watched, but how hot the watching was. Inputs:

• Rating velocity. How fast are fans opening Buzzr and rating this game in real time? Games with heavy mid-game rating volume score higher than games that generate only end-of-night ratings.
• Reaction rate. 🔥, 💯, and ❄️ reactions logged during the game. A big-play burst of 🔥 in the 4th quarter spikes energy.
• Take volume. The density of community posts, replies, and quick takes dropped during the game window.
• Party activity. Number of active Watch Parties running on the game, plus per-party engagement.

Energy is the most network-dependent dimension. A historic regular-season game in October still scores lower on energy than a quarterfinal elimination , correctly, because the room wasn't as hot. That's a feature, not a bug: we're measuring what fans actually felt.

Drama (the moments)

Drama is about the shape of the decisive minutes. It overweights the last 25% of regulation, overtime, and explicit "win or go home" context:

• Late-game tension. A one-possession game in the final two minutes scores high; a 20-point runaway does not.
• Overtime depth. Each extra period compounds. Double-OT games almost always clear 9.0 drama.
• Clutch plays. Game-tying or go-ahead plays in the final 30 seconds of any period trigger drama weight.
• Stakes multiplier. Win-or-go-home context (playoff elimination, championship, relegation, qualification cutoff) multiplies the raw drama number , stakes are the scaffolding that makes late-game moments feel heavier.

A stakes multiplier is why an OT thriller in an NBA regular season game in January lands around an 8.8 drama, while a structurally similar game in Round 1 of the playoffs lands at 9.3.

How the three combine

The overall Buzzr Score is a weighted blend:

BuzzrScore = 0.30 × Chaos + 0.30 × Energy + 0.40 × Drama

Drama is weighted most heavily because, empirically, it's the strongest predictor of whether a fan would tell a friend to watch the replay. Chaos and Energy are equally weighted because they measure different things (in-game structure vs. out-of-game reaction) and we want both to matter.

Two additional adjustments apply:

1. The late-game boost. A score near 8.5 that includes a true buzzer-beater moment gets a small bump, capped at +0.3. The idea is to recognize that some games become great only in the final 30 seconds, and the dimensional sub-scores can slightly underweight that.
2. The fan-rating pull. If the community consensus rating (the average user rating submitted during and after the game) strongly disagrees with the algorithmic score , say, more than 1.2 points apart , we pull the score 20% toward the fan number. This is a small but meaningful correction: fans see things the model doesn't (a referee moment, a star's 1-leg return, a stadium atmosphere). The algorithm stays anchored, but the crowd gets to vote.

Band names and what they mean

Every game ends in a band. These are the labels you see next to the number:

• 10.0 , Perfect Game. Once or twice a season, league-wide. The algorithm and the crowd agree: this one goes in the book.
• 9.5–9.9 , All-time classic. Drama was maxed, chaos was high, and the room was white-hot.
• 9.0–9.4 , Absolute banger. The game you rewatch.
• 8.5–8.9 , Must-watch. Worth the two hours. You'll remember two moments.
• 7.5–8.4 , Worth your time. Solid night of sport. Not a classic, but a real game.
• 6.0–7.4 , Decent game. Some moments. Not a priority replay.
• 4.0–5.9 , Skippable. You didn't miss much.
• 1.0–3.9 , Skip it. Blowout, lifeless, or both.

Bands are meant to do the heavy lifting of communication. The number is exact; the band is what you remember.

What we deliberately don't count

A methodology isn't just what you include. It's also what you refuse to count. The Buzzr Score explicitly does not factor in:

• The spread or any betting line. Whether a game "beat the number" is not our problem. Ever.
• Individual player fantasy output. Scoring a lot of fantasy points doesn't mean the game was entertaining.
• Market size or TV rating. A Bills-Dolphins Thursday night draws a different audience than a Yankees-Dodgers World Series game. Both can be 9.5s. Both can be 4.0s.
• Pre-game hype. If the game is being scored after the fact, what people expected going in is irrelevant. Only what happened counts.

Where the score breaks (and what we do about it)

The score is honest about its own failure modes. Two examples:

• Low-scoring games in paced sports. A 1-0 EPL match with elite defense and one late winner can feel transcendent to a soccer fan and read as a low-drama, low-chaos event to a model that was trained mostly on higher-scoring formats. Sport-specific weighting corrects for this, but we continue to tune it.
• Historic milestones. When a player breaks a decades-old record in an otherwise ordinary game, the game's dimensional score will underrate what fans experienced. We don't auto-inject a "milestone bump." Instead we trust the fan-rating pull to catch it. It usually does.

Both are areas we keep investing in.

The simplest explanation

If you've made it this far and you want the one-line version, here it is: the Buzzr Score is the number a friend with perfect judgment would give you when you ask, "was the game last night worth my time?" Everything else is how we try to get as close as possible to that friend.

---

Further reading: Esports isn't eating traditional sports · Betting fatigue · How the group chat ate the sports bar · Building Buzzr · Shipping Buzzr Bets · Changelog