Where our board disagrees with the room

What each column holds

ADP is the average pick in real drafts. Sources counts the independent forecasters whose projections stand behind our line, and reads Ours when no outside forecaster covers the player and our own model is filling the gap. Our rank and Room rank are ranks at his primary position, printed as the position and a number, so C7 means the seventh center. The final column is Value in a category format, the sum of his category scores with punts left out, or Points in a points format, his projected fantasy points for the season. On a narrow screen the board drops the team, the sources and both rank columns (and the value column while the platform spread is showing) so the gap stays in view; a wider screen shows everything. The board opens on its first 40 rows and reveals 40 more at a time.

When a gap is coloured, and when it is not

A gap turns green at +3 or more and red at -3 or lower. Anything closer is printed in grey, because two rankings of the same position will always disagree by a place or two, and that wobble carries no information. A gap on an Ours row stays grey however large it is. Our own model builds its lines from last season’s box scores, so it knows least about exactly the players most likely to surprise: rookies, and anyone whose role is about to change. When a player has no draft position at all, the gap cell shows a dash rather than a guess.

The three tiles

Before you touch a control, the tiles above the board answer the first question most people have. The first shows the player our board rates furthest above his room rank, the second the player the room is taking furthest ahead of our rank, and the third, when two platforms’ drafts are both on the page, the player those two platforms disagree about most. Only players with an average pick inside the first ten rounds of a twelve-team draft, the first 120 picks, can appear, and a grey gap never earns a tile. Narrow the board to one position and the tiles narrow with it. Tap any tile to jump to that player’s row.

Punting: our side moves, the room stays still

Punting means giving up one or more categories on purpose so the rest of your roster can win the others. The Punt row lists every category in the format you chose. Tap one and our board is priced again over the whole player pool without it, so every rank on our side can shift. The room’s side cannot shift, because nobody runs a draft market for a single strategy: an average pick blends every kind of drafter. That is exactly why the gap is the number to watch in a punt build. Drop free-throw percentage and the centers who struggle at the line climb our board while their draft price stays where it was; drop assists and the guards whose value came mostly from passing slide down, which is your cue to let someone else draft them. You can punt more than one category, but at least one must stay in play, and the copy button puts your punts in the link it makes.

A punt is a commitment, not a discount code. In a head-to-head league you will lose that category in most weeks, so the rest of the roster has to be strong enough to take a majority of what is left. In roto the cost is a season near the bottom of one standings column, which is fixed and easy to see. The best moment to commit is after your first few picks have already leaned one way. If the gaps keep pointing at players those picks do not fit, they are describing a different team from the one you are building.

Nine categories, eight, points or roto

Nine-category leagues count points, rebounds, assists, steals, blocks, three-pointers made, field-goal percentage, free-throw percentage and turnovers, with turnovers scored so that fewer is better. Eight-category leagues use the same list without turnovers, which helps the high-usage ball handlers whose giveaways no longer count against them. Roto scores the same nine, judged across a season of standings instead of weekly matchups. Both percentages are weighed by volume: a player’s shooting counts for more the more shots he takes, because a big man taking a dozen shots a night at a high percentage moves your team’s number far more than a reserve who takes three.

Points leagues hide the Punt row and offer two scoring tables instead. Makes and misses gives one point per point scored and one per three-pointer, two per made field goal and minus one per attempt, one per made free throw and minus one per attempt, one per rebound, two per assist, four per steal and per block, and minus two per turnover, so an inefficient shooter pays for his misses. Six stats scores only points (one each), rebounds (1.2), assists (1.5), steals and blocks (three each) and turnovers (minus one), with no shooting line, so a volume scorer loses nothing for the shots he misses. Choose whichever table is closer to your league’s settings.

Two platforms and the spread between them

The ADP column blends the average picks from two platforms’ drafts. When both list a player the blend is the midpoint of the two figures; when only one lists him, that figure is used on its own and nothing is invented for the other. The Platform spread column prints both picks, earliest first, with the distance between them underneath, and a dash where only one platform lists the player. A spread of a full round or more, twelve picks on this board, is printed in heavier type. Splits like these happen because each platform shows its users its own default rankings and drafters lean on whatever list sits in front of them, so a player can go far earlier on one site than the other. If your league drafts where he goes late, you can afford to wait on him; if it drafts where he goes early, the blended figure will tell you to wait too long. The Biggest platform split sort brings the widest of these disagreements to the top, and the line under the page title counts how many players each platform prices and how many both price, so you can see when a dash reflects a shallow market rather than an obscure player.

Reading one platform’s room on its own

When the data allows it, a My league drafts on control appears above the board. All platforms keeps the blend. Yahoo switches the board to that platform’s drafts alone: ADP becomes its average pick, Room rank becomes the player’s rank at his position inside its drafts, and the gap, the tiles and the sorts are all worked out again against it. The spread column, its sort and its tile disappear in this view, since they compare two rooms and you have picked one. That platform prices fewer players than the blend covers, so deeper rows show a dash and drop to the bottom of the draft order, and a note above the board says how many of the players in view it prices. It is the only platform offered on its own because its terms allow its draft data to be shown by itself, with the credit line printed under the board, and your choice is saved in the link like every other setting.

Four sorts, and when to use each

Draft order follows ADP and is the view to keep open during a live draft, since it shows who is likely to go before your next turn. Biggest disagreement lists the largest positive gaps first and finishes with the most negative ones, so value targets sit at the top and the players drafters are reaching for sit at the bottom. Our value ignores the room and orders the board by our number alone. Biggest platform split is offered only while the spread is on the page. The Position chips narrow any of these: PG, SG, SF, PF and C, plus Guards and Forwards for the combined seats. Eligibility decides who appears, so a player listed at both guard spots shows under each of them, while his two ranks are always counted at the first position listed for him.

Games played and the late rounds

Our numbers are season projections, so they already account for how many games the forecasters expect a player to suit up for: a player projected for a partial season brings fewer total rebounds, steals and points, and his value shows it. That matters in basketball, where planned rest and minor injuries routinely cost players games. The draft side says nothing about why a player goes where he does, and deep in the draft it says less in general. A late average pick can be built from many drafts or from only a handful, and nothing on the row tells you which. Give a large gap past the tenth round a second look before you spend a pick on it, and remember that the board ends at 400 players.

Where our numbers come from

For each statistic our line takes the median of the independent forecasters who publish one, counting each forecaster once however many sites carry its numbers. Our own model fills in only for players no outside forecaster covers, and it never joins the median. The line is then scored in your format, punts removed, and ranked at the player’s main position, while the draft side is built separately from average picks in real drafts. Neither number is nudged toward the other, which is what makes the difference between them worth reading. The consensus table shows the projections behind each line, and the methodology sets out the arithmetic.

Where to go next

The rankings show the same values without the draft market beside them. The mock draft room lets you rehearse a punt build against opponents who draft from this market, the auction values turn the same board into dollars, and the live companion takes players off the board as they are picked in your real draft.