Advanced Stats, Explained: How xG, Corsi, Luck, QoC and GAR Help You Win at Fantasy Hockey

Box scores tell you what happened. Advanced stats tell you why, and whether it will keep happening. That second part is the whole game in fantasy hockey: you are not buying last month's points, you are buying next month's.

This guide walks through the six advanced stats on Kodo Hockey using real 2025-26 numbers for the 543 skaters who played at least 600 minutes at 5-on-5. Every chart moves. Play with them.

1. Expected goals (xG): how good were the chances?

Every unblocked shot gets a probability of going in. Kodo's model weighs 21 things about each one: distance and angle to the net, the type of shot, and the play that led to it (how long since the last event and how far the puck travelled, how fast it moved, whether it came across the ice, whether it was a rebound), plus the power play or penalty kill, an empty net at either end, and overtime. It was trained on 339,000 shots from the last three seasons. Add up those probabilities and you get expected goals: how many goals an average shooter would have scored on the same chances.

Here is every unblocked shot Anthony Mantha took last season. Switch between outcome and xG and watch where the big circles sit.

He scored 33 goals on 17.9 expected, the widest gap in the league among players with 100 or more shots. That gap is called finishing. Some of it is real talent. How much is the question section 3 answers.

For fantasy: a player generating lots of xG but not scoring is a buy. His chances are real and the goals usually follow. A player scoring far above his xG on few chances is the classic sell-high.

2. Corsi and xGF%: who drives play?

Corsi for % (CF%) is a team's share of all shot attempts while a player is on the ice at 5-on-5. xGF% is the same idea using expected goals instead of attempts, so a point shot from the blue line counts for much less than a chance in the slot. Above 50% means play tilted toward the other net with him out there.

Most regulars land between 45% and 55% CF. Players above the diagonal turn their attempts into better chances than average; players below it are piling up point shots.

For fantasy: xGF% predicts future goal share better than actual goals do. A top-six forward on a high-xGF% line gets more offensive-zone time, which means more shots, more points and a better plus/minus.

3. Luck and PDO: what will not last?

PDO is a player's on-ice shooting percentage plus his on-ice save percentage. It centres on 100 and barely repeats from one season to the next, so a number far from 100 is a forecast that results are about to move back toward it. Kodo measures the same thing in goals: on-ice luck is (goals for minus expected) plus (expected against minus goals against) while he was on the ice at 5-on-5.

Individual finishing works the same way, with one twist: part of it is real skill. We measured how much by pairing seasons for over a thousand shooters. Goals above expected repeat at only r = 0.26, so Kodo keeps shots ÷ (shots + 377) of the edge as skill and treats the rest as luck.

The first four in the list are last season's biggest over-finishers, the last four the biggest under-finishers. The Luck page runs this for every player, every day.

For fantasy: sell players whose points rest on luck they cannot keep, and buy the unlucky ones before their shooting percentage comes back.

4. QoC and QoT: who was he up against, and who was he with?

Context changes everything. Quality of competition (QoC) is the ice-time-weighted xGF% of the opponents a player faced at 5-on-5. Quality of teammates (QoT) is the same thing for his own linemates. A shutdown centre facing top lines every night will have worse numbers than a third-liner with easy minutes, and that does not make him worse.

QoC is a narrow stat: most regulars sit between 49.5% and 50.3%, because coaches can only shelter a player so much. QoT spreads far wider (46.3% to 54.4%), and that is where most of the context lives.

For fantasy: a player with strong numbers and a low QoT is doing it himself, so his production travels if he moves up the lineup. A player with a high QoT is riding his linemates: check the line combinations before you buy, because a line change can take it all away.

5. GAR-lite: one number for all of it

GAR-lite rolls finishing and play-driving into a single goals figure: finishing (goals minus xG) plus 0.3 × his on-ice expected-goal margin at 5-on-5. It is a quick all-in-one, not a full wins-above-replacement model. There is no replacement level and no credit for killing penalties.

Nathan MacKinnon led the league at +18.6. Pick a player to see how much came from finishing and how much from driving play. The two bottom entries show what a negative GAR-lite looks like.

For fantasy: GAR-lite is a tiebreaker, not a ranking. Two players with the same points are not the same player if one of them is carried by finishing luck.

Putting it together: a five-question check on any pickup

1. Is he getting chances? Individual xG and shots per game. 2. Is his team winning his minutes? xGF% above 50. 3. Is it luck? Finishing and on-ice luck far above zero on a small sample. 4. Who is he playing with? A high QoT means check the lines first. 5. Does it add up? GAR-lite as the tiebreaker.

Every one of these lives on the site: the Statbook for the full tables, Luck for buy-low and sell-high lists, Impact for GAR-lite, QoC and QoT, and every player page for his shots and on-ice numbers.