How it decides

Everything the tool knows, where it came from, and exactly what it does with it. If a recommendation ever looks wrong, this page should explain why it said what it said.

What the tool is

A second screen for a draft that goes in rounds: snake, linear, or third-round reversal. You keep the draft running wherever it runs (ESPN, Yahoo, Sleeper, a whiteboard) and mark picks here as they happen. When your turn comes, the top of the page names the player it would take, with the reasoning next to the name. It works for one team: yours. It does not know who the other teams have, only who is gone.

It does not do auction drafts. Auction value depends on every team’s remaining budget and who has been nominated, and that state lives inside the platform. A draft that goes in rounds needs only a list and a pen.

The data

The data is built into a file before each deploy and refreshed daily during draft week. The build date is in the footer of the board. Nothing is fetched while you draft, which is why the page is fast and why an injury reported an hour ago may not be here yet.

Each scoring format has its own ADP and its own projection. Switching the format switches both. ADP tells you when the market takes a player. Projections tell you what the player is expected to score. The tool uses each for what it is good at.

Scoring formats

PPR gives a point per reception, half PPR gives half a point, standard gives none. The difference moves pass-catching running backs, receivers, and tight ends up or down relative to rushing backs and quarterbacks. Because the ADP and the projections are both format-specific, the tool does not approximate one format from another. If a player’s projection is missing for a format, it is estimated from the projections of neighbouring players at the same position by ADP, and marked with a tilde on the board.

Value: points above replacement

The Value column is the player’s projected points minus the projected points of the replacement player at that position: the last starter in your league. In a 12-team league with two running back slots and one flex, the tool expects about 30 running backs to start (24 in RB slots, plus half of the 12 flex slots), so the replacement running back is the 30th best by projection. For quarterbacks it is the 12th. This is why a quarterback projected for 360 points can be worth less than a running back projected for 280: the 12th quarterback is still very good, and the 30th running back is not.

The flex share is fixed: half of flex slots go to running backs, four tenths to receivers, one tenth to tight ends. A superflex slot, if your league has one, counts mostly toward quarterbacks: eight tenths of superflex slots go to quarterbacks, one tenth each to running backs and receivers. That is why a two-quarterback league makes quarterbacks scarce, which is the point of a two-quarterback league. Replacement level is computed once from the full pool and does not move during the draft. That keeps the Value column stable, so a player’s number means the same thing in round two and round nine.

The recommendation

The player named at the top of the board is the one with the highest score. The score has three parts.

1. The wait gap. For each position, the tool finds the best player still on the board who is likely to survive until your next pick, meaning their ADP is later than your next pick number plus a small margin of two. A player’s wait gap is their projection minus that survivor’s projection: what you lose at this position by waiting one turn. This is the core of the method, and it is why the tool needs your draft slot. Two players with equal wait gaps are equally urgent even if one is a star and one is a tight end, because taking the star now and the tight end later leaves you with the same total as the other way round. If your slot is not set, the wait gap falls back to the Value column.

2. A nudge toward value, and a floor. Fifteen percent of the player’s Value is added, so that ties between equally urgent players break toward the better player overall. A small floor is added too: three points times the player’s projection divided by the replacement level at their position. It exists for the picks where the wait gap is zero for everyone, which happens at the turn in a snake draft, when your next pick is one away and every player will still be there. Without it, every candidate scored zero and the recommendation was whoever happened to sort first. With it, equally urgent players sort by how good they are relative to their position.

3. The roster-need multiplier. The sum is multiplied by a factor that depends on where the player would go on your roster.

The multiplier is what stops the tool from recommending a fifth running back for your bench when you have no tight end, and what keeps kickers and defenses out of the middle rounds. Those two positions are interchangeable, the gap between the best one and the twelfth is small, and a replacement is usually sitting on waivers in week three, so bench depth at the skill positions is worth more than locking them in early. The tool does not recommend them at all until your remaining picks are nothing but your open kicker and defense slots, and never recommends a second one for the bench. They stay on the board the whole time if you want one early. An earlier version gave them full weight as soon as the other starters were filled, which produced a kicker in round eight. A reader caught it during a live draft. A weight instead of a zero was tried first, and lost: late in a draft the bench candidates are all below replacement, and any positive weight on a defense eventually beats them. Tight ends in a flex slot get half weight for a related reason: tight end projections fall off steeply at the top, so the wait gap oversells a second tight end when a receiver would start every week. It is a blunt instrument on purpose. A draft is not the time for a subtle one.

Players on injured reserve, on the physically-unable-to-perform list, suspended, or otherwise inactive for the season are never recommended, though they stay on the board with a red tag in case you want them anyway. Week-to-week designations are different, because a draft is for the season: questionable costs a player five percent of their score, doubtful or out for the coming game costs fifteen, and all three carry a yellow tag naming the status.

Tiers and position runs

Within each position, players are sorted by projection, and a new tier begins wherever the drop to the next player is more than six percent of the previous player’s points, or eight points, whichever is larger. The Position runs panel counts how many players are left in the top remaining tier and the size of the drop after them. When that count reaches two, the note turns yellow. That is the moment a run at the position is about to cost you.

The clock

Three draft orders are supported. A snake draft reverses direction every round. A linear draft runs the same direction every round, so the first slot picks first every time. Third-round reversal is a snake draft where round three runs the same direction as round two, and the alternation continues from there, which softens the advantage of the first slot. Auction drafts are not supported. Given the order, the number of teams, and your slot, the tool knows every pick number that is yours, and shows the next one and how many picks away it is. The overall pick number is simply how many picks have been marked plus one, so the clock is only as accurate as your marking. If you fall behind, mark the missing picks and it catches up. Undo takes back the last one.

What it does not know

• Who the other teams have drafted. It knows who is gone, not where they went, so it cannot predict that the team picking before you needs a quarterback.
• Your league’s exact scoring beyond the reception rule. Bonuses, six-point passing touchdowns, and return yardage are not modeled.
• Anything that happened after the data was built. See the footer for the date.
• Keepers and dynasty rules. Mark kept players as taken before you begin.
• Auction drafts.

Privacy

The draft is saved in your browser’s local storage so a refresh does not lose it. Nothing about your picks is sent anywhere. If analytics are enabled, the page reports three anonymous events: that a draft started, that a pick was recorded, and that a draft finished, with the scoring format and league size attached. No names, no rosters.

Credits

ADP from Fantasy Football Calculator and Sleeper. Projections by RotoWire, published through Sleeper. Injury reports from Sleeper. The method is the value-based drafting approach Joe Bryant described in the 1990s, with the wait gap replacing a fixed baseline. Made by Colin Sherrill. The source is on GitHub, MIT licensed.