"How the hell do you read these things?"

Game Score and Impact Cards as explained by a Sabres fan who likes analytics (2026)

TLDR: Blue is offence, orange is defence. Left of 0 is bad, right of 0 is good.

Dom Luszczyszyn's Game Score is an advanced all-in-one advanced stat that rates a skater's efficiency per game, combining offensive and defensive impact. Game Score is calculated using box score counting stats (stats that count from 0 each game) and advanced stats (AKA analytics; equations that crunch box stats to show more specific values).

Cam Palmer's HockeyStatCards graphs full team Game Scores on an Impact Card after each game played. The website also calculates Net Rating for every NHL player, which is the average Game Score per season, per skater.

Production considers individual productivity using box stats. Play driving considers contribution to team performance using 5v5 metrics. Special teams consider the effectiveness of special teams using power play goals, shorthanded goals, and penalty kill rate. At the bottom of every single Impact Card is a colour-coded legend that describes each metric.

If you're familiar with hockey stats, you can stop reading here, because the legend explains everything you need to know; if not, the next explanations contextualise what these metrics mean. Feel free to skim: highlighted parts are the most important, and bolded parts are answers to questions that get asked a lot about Game Score.

Games are linked to NHL.com game summaries. Screenshots are linked to game Impact Card.

Note: HockeyStatCards was just revamped, and Palmer switched to use Hockey Stats (previously named Advanced Hockey Stats, run by Patrick Bacon and JFresh). This will also include goalie scores based on saves and save quality, so I'll update this guide sometime soon with information on goalie score.

Production

The dark blue and the dark orange are the "production" metrics, which rates a skater's individual offence and defence. How many goals did they produce? How many goals did they stop the opposing team from producing? Offensive production includes goals, primary and secondary assists, shots, and o-zone faceoff wins. Defensive production includes blocked shots and d-zone faceoff wins. It's easiest to think about this as one guy taking an isolated action, and how much that isolated action impacts the game.

Major goal scorers, finesse skaters, and breakaway/rush skaters rate the best in offensive production. Low individual production means low shooting. Offensive production can't be negative because it's calculated by box stats that don't go into the negatives (i.e. shot attempts can't be less than 0).

Faceoff specialists, physical dmen, and agitators rate the best in defensive production, though keep in mind that defensive production is generally the lowest score for all players across the league, because defensive box stats are low tallying. Negative defensive production means defensive plays that led to a goal scored against, like playing in front of the crease and deflecting a shot from the opposing team into their own net.

Penalties are rated kind of weird. Penalties drawn are offensively and defensively positive: a skater individually produces an offensive chance (power play), but it also forces the opposing team out of the d-zone, which itself is a form of individual defence. Penalties taken that are killed are positive individual defence, and penalties taken leading to SHG are also positive individual offence. Penalties taken that lead to opposing PPG are only negative individual defence.

Examples

On Jan 29 against the Kings, Tuch scored 3 goals: 1 deflected, 1 tipped in, 1 on an empty net. Quinn assisted on all 3, and McLeod assisted on 2. They rated high for production and low for play driving because these individual plays contributed directly to a goal, but they weren't necessarily high-danger chances that their line intentionally set up with rebounds or more than 2 passes. (Benson assisted on 1, but had better play driving.)

On Feb 2 against the Panthers, Power at 4 blocked shots and Samuelsson had 7. Their individual defensive play prevented the Panthers from scoring. Kesselring had no blocks and no drawn penalties (and no faceoff wins lol).

On Mar 14 against the Leafs, Doan drew 2 penalties and Malenstyn drew 1. Their offensive production reflects their drawn penalties and their defensive production also gets a boost. Doan's is higher because Quinn scored on one of his drawn power play chances.

On Nov 15 against the Red Wings, Krebs took 1 penalty. His individual offence rating is positive because McLeod scored a shorty, but his individual defence rating is negative because he fucked up earlier in the game on a Patrick Kane goal. Dunne's individual defence rating is negative because he took a penalty that led to a DeBrincat PPG.

On Jan 29 against the Kings, Kesselring took 1 penalty (and 1 misconduct oops). His individual defence rating is positive because his penalty was killed and the Sabres scored pretty quickly after. Metsa's individual defence rating is a little higher because he blocked 3 shots.

Play Driving

The medium blue and the medium orange are the play driving metrics, which rates a skater's offence and defence in relation to team play. How well do they drive the team into plays? How well do they stop the opposing team from driving plays?

Play driving includes actual on-ice 5v5 goals and on-ice 5v5 expected (X) goals. Offensive play driving is calculated by 5v5 goals for (GF) above league average and on-ice 5v5 xGF. Defensive play driving is calculated by 5v5 goals against (GA) above league average and onice 5v5 xGA. All X stats are analytics, so they have formulas. It's not exactly as cut and dry as production.

Playmakers, heavy shooters who generate rebounds, and give-and-go/passing skaters rate the best in offensive play driving. Two-way skaters, mobile skaters, and puck movers rate the best in defensive play driving. Negative offensive play driving comes from not converting plays into goals, like giveaways or fumbling passes in the o-zone. Negative defensive play driving comes from getting scored on.

Obviously, this metric is important because hockey is a team sport, but it's particularly valuable for rating depth teams with limited top 20 goal scoring talent like the current Sabres, or shutdown defence teams, like the Panthers cup teams or the current Hurricanes. Teams with high-scoring top lines and grinder bottom lines rate worse in play driving, like the current Oilers or the Core 4 era Leafs.

This season, the Sabres have been outshot in over half their games and rank in the bottom half for zone possession time, yet consistently win and consistently rate high play driving. That's because their high xGF indicates their skill in creating quality chances (largely off the rush), and low xGA indicates their ability to prevent their opponents from generating high-danger chances. Quantity low, quality high. No Nikita Kucherovs here, but they still get the job done.

Tangent on expected goals (xG):

Different stat sites have different models for calculating the xG analytics, and they can vary pretty widely. (HockeyStatCards currently uses Hockey Stats.) All models will generally factor in shot data (distance, angle, velocity, shot type, etc) and event history (game situation and microstats like passes/rebounds leading to the shot, zone entries, etc) to compare with multi-season aggregate data. How likely will (Shot A) from (Location B) at (Velocity C) after (Set Up D) by (Player E) score, based on the history of ABCDE plays?

A 90+ mph net-front one-timer after a minute of o-zone control has a higher xGF than a weak side backhand amidst a scramble. Even if they're both shots on goal, one-timers score more than backhands, skaters score more net-front than on their weak side, etc. Ovechkin's xGF is higher when he gets a lot of chances in his office. If a team has a lot of 90+ mph one-timer opportunities per game, they'll have high xGF; if they limit their opposing team to only weak side shots, they'll have low xGA.

Examples

You can tell a line/pair played well with each other if they have similar play driving scores. On Jan 24 against the Islanders, the Zucker-McLeod-Quinn line had clear control with each other when positioning as a line and passing to each other in the o-zone, so their offensive play driving matches up. Production reflects Zucker's 2G, McLeod's 2A, and Quinn's 1A.

On Jan 20 against the Predators, the kid line (and Krebs, on a different line) all have significant offensive scores, with Benson standing out for the highest play driving and the lowest production. It speaks to his passing and handling, with Helenius and Ostlund as more competent finishers that got the puck in the net.

The Sabres dmen generate most of the offensive chances from the flank with the forward line rotating in the slot to shoot, so Benson's play driving indicates his ability to set up beside the defence. Play making in the Sabres forward group isn't as evenly dispersed, so games feel sloppy when Benson's out because he bears the load of creating lanes and moving against the boards to give the forwards space.

Power is a favourite punching bag for a certain kind of Sabres fan because he was drafted as a goal scoring offensive dman, but he's transitioned his play since. While he doesn't score much, he's taken on a puck-controlling role that boosts the front and the back. On Dec 27 against the Bruins:

and Mar 10 against the Sharks, his play driving rated way better than his production.

As an agile lefty skater playing right side, exits/entries are Power's bread and butter because he can outskate opposing forwards, shield the puck with his stick from opposing defence, and play it up the boards for Byram on his strong side to pass or shoot. Power stops the opposing team from driving into the d-zone, and drives the Sabres into the o-zone.

His play driving scores are fairly high and kind of the same for most games, even losses, because he's gotten extremely good at this but he doesn't do a whole lot else right now. Dahlin, Byram, and Samuelsson score more, and Dahlin's play driving is just as dynamic, but Power is more consistent than Samuelsson defensively, and more consistent than Byram on both ends.

Special Teams

The light blue and the light orange are the special teams metrics, which are the most self-explanatory and probably the least applicable to the Sabres. Offensive special teams are calculated with PPG and defensive special teams are calculated with PK% and SHG. They're weighted by league average.

Examples

On Jan 14 against the Flyers, Byram and Samuelsson gave up a PPG to Zegras, but Dahlin scored 2 PPGs!

On Mar 5 against the Penguins, Doan assisted Norris on a PPG. McLeod, Tuch, and Ostlund didn't score on their power plays, but McLeod and Tuch each scored individual SHGs.

Usage

Luszczyszyn's original Game Score and Net Rating formulas don't factor in for the quality of teammates (QoT) or the quality of competition (QoC), but he has his own model for calculating QoT and QoC through points and shared time on ice (TOI), which HockeyStatCards shows through grey bars in front or at the back of a skater's Game Score . Skaters who played poorly but the opposing team played worse, or they were elevated by their lines/pairs, have negatively rated usage. Skaters who carried their lines/pairs, or were played against the best players on the opposing team, have positively rated usage.

Examples

On Dec 9 against the Oilers, the Byram-Timmins pair didn't contribute to scoring at all, but they spent the majority of the game marking Connor McDavid's line.

On Mar 14 against the Leafs, Stanley and Schenn were pretty bad together but were bailed out by the forward lines and goaltending, as well as the Leafs being unable to convert.

Disclaimers

It goes without saying that these aren't perfect metrics, they're just more standardised than the nebulous, person-to-person "eye test". Take Impact Cards with a grain of salt. Luszczyszyn's tweaked his Game Score model over time, but it's not perfect and he refreshes the models every couple of seasons or so.

The calculations follow a general Game Score = (Value * Stat X) ± (Value * Stat Y) ± (Value * Stat Z) ± (...), but even then, his assigned point values are somewhat subjectively chosen. He weighs each stat by its probability in creating a goal scoring chance or taking away a goal scoring chance from the opposing team. Shots on goal have more offensive value than missed shots because SOG are statistically more likely to convert to goals. Blocks have more defensive value than d-zone faceoff wins because blocks are statistically more likely to stop a goal scoring chance.

Because of this, GA becomes extremely punishing, since there's fewer high-counting defensive variables to "cushion" the negatives, versus higher-counting offensive variables. MacKinnon leads in average shots per game this season at 4.48, and McCabe leads in average blocks at 2.44; one of them has more room for error than the other.

It also inevitably rates defence-minded skaters worse, since defensive plays aren't easily countable. How much do blocks really mean? Is McCabe a dark horse Norris candidate? How do you count Slavin getting in the way of passing lanes in the d-zone, or Fox's forecheck positioning? Microstats only go so far. Heat maps, zone entry location, and disruptions can't be compressed into a single number without a lot of funky math.

This is the most apparent if you get into Net Rating. This is the HockeyStatCards Net Rating leaderboard for all skaters this current season.

The #1 defence rating for skaters doesn't come close to the #10 offence rating. Of the 6 dmen in the top 10 Net Rating, none of them have higher defence ratings than offence ratings. (Bouchard has -2.9!) The top 10 dmen by Net Rating are a fair mix of offensive and 2-way dmen, with an absence of defensive dmen.

Note: Forward net ratings follow the same trend with a lack of 2-way forwards in top 10, but a lot of elite 2-way forwards are having bad seasons or are out for injury currently.

I'm not saying Lindell (+2.5 +13.1 +15.6) is better than Makar (+19.8 +8.4 +28.2) so these numbers are all garbage. What I'm getting at is that goal scoring is always prioritised by Game Score and Net Rating, so Metsa is a lot less likely to have a stand-out night on Impact Cards even though he's a nightly difference maker.

Further Reading

Here are the articles Luszczyszyn has written about Game Score if you want to know more:

• Creating Game Score, 2016
• Factoring in for QoT and QoC, 2017
• Using Game Score to project team and player ratings, 2017
• Moving from Corsi to xGF and other updates, 2019
• Delineating offensive vs defensive scores, 2023

And here's some of the stuff that got me into analytics that might be interesting if you like Impact Cards as much as I do:

• Reddit thread on advanced hockey stats
• District 5's videos on home ice advantage and shooting hand dominance
• JFresh's Substack article on microstats