How the Pick Score works

• The big idea (3 steps)
• Score colors
• Value: rank & ADP
• The two master dials
• 1. Position limits (hard gates)
• 2. Stacking
• 3. Playoff stack
• 4. Draft capital
• 5. Bye spread
• 6. Bench pick penalty
• 7. Same-team RB fade
• 8. QB scarcity boost
• 9. Stack-partner ding
• 10. Need urgency
• How it all combines
• Worked example
• Tuning & defaults

These are hard rules, not nudges. A player hits 0 when:

• Position maxed — you already have the max at that position.
• Must-fill — you have just enough picks left to still hit your minimums, so this pick must go to a position you still need; everything else is zeroed.
• Turned off — you toggled that position off in the "Best Pick Score" header in the rail. Live drafting only: post-draft grades (the Draft Grades board and the board grader) always score every position, so a hidden position never zeroes a finished board.

Underdog's 18-round format seeds two of these differently: RB min 4 and TE max 3 (the WR minimum of six holds on both sites — fewer than six WRs has historically hurt both cash and win rates). All eight numbers are editable in the Scoring weights panel. A safety guard prevents the whole board from going to 0 — if every player would be gated, the soft scores come back so you always have a recommendation.

Boost for a player on the same NFL team as someone you've drafted. QB + pass-catcher (a QB with a WR or TE) is the real correlation and carries almost all the weight; every other pairing adds only a little. The numbers below are the per-pairing strength (read as: a candidate at the row position who joins an existing teammate at the column position).

Roughly, at the default slider, a clean QB + WR (or QB + TE) stack runs about +10% in the early rounds and climbs to the +15% cap by the mid-rounds (it's damped early so it doesn't over-inflate your studs). WR+WR with no QB is a smaller correlation (about +2–4%), WR+TE smaller still (~+1–2%), and an RB stacks only with a QB (a token ~1–2%). A second teammate at the same position counts at most twice, and the total stack boost is capped at +15% at the 50% slider — the cap scales with the slider (about +30% at 100%).

QB→QB, TE→TE, and RB→RB are zero (no self-stack), and an RB no longer stacks with a WR or TE.

Over-stacking damper: the bonus runs at full strength while your roster has up to 4 players stacked with your QBs, then decays gently — barely felt at 5–6 (≈96%/84%), steeper by 8 (≈36%), gone at 9 — and from the 10th stacked player it flips into a penalty that grows with the pile. Best-ball cohort data shows 4–6 total stacked players is the sweet spot and 10+ actively hurts advance rates. The popover names the pile size when the damper is acting.

Site note: on DraftKings the whole Stack factor runs at 90% of its Underdog strength (full-PPR scoring correlates less with the QB than half-PPR) — the slider reads the same 50% on both sites; the base effect is simply smaller on DK.

Boost when the player's real-life Week 15, 16, or 17 game is against a team you already roster — you and this player are in the same game (a "bring-back"). The size depends on both positions: the candidate's, and the position of your player on the opponent. It's a matrix like stacking — read as a candidate at the row position lined up against your rostered player at the column position:

You get one boost per shared week — the single highest pairing that week (not a sum) — added across the up-to-3 weeks. The marquee is a QB ↔ your opposing pass-catcher (0.15); anything involving an RB — and QB-vs-QB — gives nothing.

Week 17 leads at 1.75× by default (championship week), with Week 16 at 1.25× and Week 15 at 1× — the earlier playoff weeks matter more than a Wk17-only view suggests (improving any playoff week helps your title odds about equally, and earlier weeks bank prizes), so the defaults spread the weight while still favoring the final. Each week's strength has its own slider (Wk 15 / 16 / 17). Playoff stacks are deliberately gentler than same-team stacks: at the 50% slider, a QB bring-back is about +3.75% in Week 15 and +6.6% in Week 17; recurring across weeks tops out at the +10% cap at the 50% slider (it scales — about +20% at 100%).

Visual cue: a playoff week badge gets an outlined ring (black on the board, white in the rail) when the bring-back includes a QB you roster on that opponent — the strongest version of this stack.

Nudges you toward a healthy share of your total draft value per position. Each pick's "value" comes from the draft-capital curve (early picks are worth far more than late ones), and your spend so far is compared to these target bands:

Two things shape how big this nudge is:

• When each side fires. The penalty starts the instant a pick's projected share would cross the top of its band — no grace zone, but it ramps up from zero, so barely crossing barely registers. The boost for being under a band is exactly zero for as long as passing still leaves the floor reachable at all: the model asks what the position could reach if this pick went elsewhere, and any cushion above the floor — even a small one — means no boost. It only ramps in once skipping would make the floor unreachable (this pick is the last exit). Being at 30% with a 42% floor and picks in hand is a plan, not an emergency. Because the capital curve is so front-loaded (your first five picks are roughly 64% of your budget), a WR-less start hits that last-exit point around round 5 — and a balanced roster never sees the boost at all.
• One gentle curve on both sides. The size ramps quadratically from zero at the band edge: 1 point past the top (or a 1-point prospective miss) is ≈±0.4%, 3 points ≈ ±3.6%, 5 points ≈ ±10%, reaching the ±35% cap around 9–10 points, at the default slider. Once a floor is genuinely out of reach the boost keeps pushing and then fades on its own as you pour picks in and the gap shrinks.

So a small ±1–3% here usually means you're close to the band. The bands are editable, and the "Draft Capital by Position" graph in the rail shows your current spend vs each band — a hatched red bar there means that floor can no longer be reached with the picks you have left.

A penalty for piling too many of the same position onto the same bye week. Counting this player, crossing the first threshold is a moderate hit and the next is a steeper one:

The moderate penalty is waived if you already have another player at that position on a different bye week — i.e. you've shown you're spreading that position's byes, so a little clustering is fine. The severe penalty always applies, diversity or not.

Free agents and players with no bye listed are never penalized (no guessing on missing data).

A penalty for spending an early pick on a player who couldn't crack your starting lineup — 1 QB / 2 RB / 3 WR / 1 TE / 1 FLEX (flex = RB/WR/TE) — even with everyone you drafted before him healthy and starting. Your earlier picks get slot priority, so the candidate only "starts" if a primary slot for his position is open or the flex is still free.

A second bench player at the same position is dinged harder and two rounds longer — the model really doesn't want you taking a 3rd QB before round 10:

Examples: a 2nd QB in round 4 is a bench pick (only 1 QB starts, and QBs don't flex) — −10%. A 3rd QB in round 8 is a second bench QB — −10% even that late; from round 10 it's free. A 3rd RB is fine while the flex is free, but a 4th RB — or a 3rd RB after a 4th WR already claimed the flex — is a bench pick. WR gets a pass in round 9 because WR-heavy builds are normal builds. All numbers at the default 50%; scales with its slider.

If you already roster an RB from an NFL team, another RB from that same team takes a −50% hit through round 13 — same-backfield RBs split one backfield's touches. The penalty then eases off (≈−43% in R14 down to ≈−14% in R16) and is gone from round 17, where a cheap handcuff is a fine dart. It's a strong ding rather than a hard block: the best-ball handcuff data shows same-team RB pairs don't hurt advance rates (and dual-role backfields have even helped), so an exceptional value can still surface instead of reading as undraftable.

This is an on/off rule ("Fade same-team RBs" in the Scoring weights panel), not a slider — and it applies in post-draft grades too when it's on.

Uses the same math as the board's Next Pick Avail % column — the chance each player is still there at your next pick if you pass on him now. When you're short at QB and waiting looks likely to leave you stranded, every remaining top-200-ADP QB is boosted so the best of them becomes the board's top choice:

• 1 QB rostered: fires when no top-200 QB is ≥66% to make it back to your next pick.
• 0 QBs rostered: fires unless one QB is ≥66% to survive to your next pick and a different QB is ≥66% to survive to your pick after that (you need two).

Shows as "QB scarcity" in the score breakdown. Automatic — no slider. When your slot isn't known yet (or there's no next pick), it stays off rather than guess.

If your roster has no QB stack (no QB with a same-team WR/TE) and none of your current QBs can realistically still be stacked — no same-team WR/TE left in the top 200 of ADP who's ≥66% to make it back to your next pick — then candidate QBs who also have no path to a stack are dinged a flat −10%. It steers you toward QBs a stack can still be built on.

A QB counts as stackable if a partner is still on the board or is already on your roster — a QB whose WR/TE you already hold completes the stack the moment you draft him, so he gets the Stacking boost, never this ding.

Shows as "No stack path" in the score breakdown. Automatic — no slider; uses the same pass-on-him availability model as the Avail % column.

A small automatic boost for a position you still need to reach its minimum when you're running low on picks. It ramps up as the squeeze gets tighter and hands off to the hard "must-fill" gate at the very last moment. There's no slider — it just protects you from missing a minimum.

Ryan Flournoy WR · DAL · Rank 173 · ADP 168 — you roster a QB, a WR, and a TE on DAL (a same-team stack), and you also hold a QB on LAR, whom DAL plays in Week 15 (a bring-back).

The QB drives almost all of the stack; the extra WR and TE add only a few percent. Because his adjusted value tops the available pool at that moment, he renormalizes to 100 — the popover's "100 = the best available pick right now."