Hockey Matchup Planner

How a category total is built

Every skater in our model carries a per-game rate in each counting category: goals, assists, points, power-play points, shots, hits, blocks, penalty minutes and plus-minus. A roster's projected total in a category is the sum, over every active player, of his rate times the games his club still has in the week, counted from today when the week is the current one and from Monday when it is a later one. A player parked on IR, IR-plus or the bench contributes nothing, and a player our injury feed lists as Out contributes nothing either, because an injured skater in an active seat is the most common way a week is lost quietly. Goalies are different. A tandem shares the crease, and the club's schedule tells you how many games the club plays, not how many your goalie starts, so each goalie's counting stats are scaled by his expected share of starts from the goalie feed. Goals-against average and save percentage are never averaged across two goalies: the goals against are summed, the starts are summed, and the division happens once, which is how the categories are actually scored on the platforms.

Reading the grid

Each row is one category. Your projected total sits on the left, your opponent's on the right, and the last column says which way the race leans. A race reads as leading when your figure is clearly ahead, trailing when it is clearly behind, and a coin flip when the two totals are within about six percent of each other for a counting stat, or within a few thousandths for a rate. Goals-against average and penalty minutes are judged the right way round, so a lower figure leads. The headline above the grid counts the races: leading seven, trailing two, one too close to call. That count is what you are managing all week. A race you lead by a wide margin needs nothing from you; a race you trail by a wide margin is not worth chasing; the coin flips are where a single add or a single healthy scratch changes the result, and the page lists those separately as the swing categories.

Games played is the whole game

The per-night view shows the seven nights of the fantasy week with the league-wide game count under each date and the number of your active players who have a game that night. Two things to look for. First, the light nights, when six or fewer games are played: on those nights most rosters have empty seats, and any player who dresses is worth more than a better player who does not. Second, the nights when several of your players sit at once, which is when a bench skater on a club that does play would have earned you real category totals. If your opponent's roster is built on clubs with four-game weeks and yours is built on clubs with three, the grid will already show it, and the fix is in the schedule, not the projections: the schedule planner ranks the clubs by games in any week you name.

The what-if row

Pick one unrostered player and, optionally, one of your own to drop, and the grid re-scores with the change beside the original. The candidates are the players nobody in your league owns, with games left in the week, ranked by what they are projected to do per game in the categories your league counts, so a hits-and-blocks defenceman rises in a league that counts hits and blocks and sinks in one that does not. The point is not to find the best player available. It is to find the add that flips a coin-flip race without costing a race you already lead, which is a narrower question and one a projection can actually answer. When the new count is no better than the old one, the add is not worth an acquisition, however good the player looks on the waiver wire.

Where the numbers come from, and where they are weak

The rates are from our own projection model, the same one behind the category board and the streamers page, built from game logs and an ageing curve; before the season opens they rest on last season alone, so a rookie, a player who changed clubs, or a goalie whose tandem changed is projected from less than the rest. The schedule and the game counts are our own feeds, refreshed daily. The roster, the seats and the category list come from your league through Front Office at its last sync, so if you moved a player this afternoon, re-sync there before you trust the seats here. The page does not read your league's weekly schedule yet, so you pick the opponent you face; it does not know tonight's line combinations, power-play units or a healthy scratch until the club announces one; and a player it could not match to our directory is listed by the name your league gave and counted in nothing, which is said on the roster rather than hidden. Without a linked league the page shows an example: two lineups dealt from the top of our projections, labelled as such, so the tool is visible before you connect anything.

The week, the opponent and any what-if add or drop are written into the address, so the link you copy opens the same plan for anyone in the group chat. A linked league is read only when you are signed in; nothing about it is shared through the link.