How the Numbers Work
Scoring
Credit
Season Credit rates recorded production against players at the same position or specialist role in their league. Best Credit keeps the player's strongest eligible season, including a season in progress, and carries that earned Rating across transfers.
How a Stat Earns Credit
Each stat carries an event value: its average change in score or field position.
Sting multiplies that value by the recorded count and by the player's agency: his estimated share of the play.
Event values are population averages. They do not claim the exact value of one play.
Rounding
Each stat's event value × agency × recorded count rounds once to the nearest 0.05 Credit, with exact halves rounding away from zero.
Every raw Credit total is therefore a multiple of 0.05. A finer step would claim precision the box score cannot support.
The Season Window
Season Credit counts the games every conference team reached, so a longer postseason run does not raise the season Rating.
Those games still appear in Postseason Total and their weekly views. When incomplete evidence leaves no trustworthy shared window, Sting leaves the season uncapped and says so.
Offense
A completion earns Credit beyond its yards.
A true incompletion costs Credit. An interception is excluded from the incompletion count because it has its own turnover value; the same dropback is never debited as both.
Passing and receiving yards split evenly between the quarterback and receiver.
Both players record the same play.
Yardage earns steady Credit.
Touchdowns and completions earn more than routine gains.
Turnovers cost more than any single positive event earns.
Passing and rushing use the same value per yard.
Defense
A sack also appears as a tackle for loss and a tackle. Sting counts only the most specific event: sack first, then non-sack tackle for loss, then tackle. No play counts twice.
Tackles lose value farther from the line of scrimmage. Pass breakups gain value there because they prevent larger completions. The box score omits play location, so position stands in for depth.
The two scales run in opposite directions: tackles rate highest at the interior line, while breakups rate highest in the secondary.
A safety's open-field tackle can prevent a touchdown, but the play has already gained yards. The box score records the tackle and misses any breakdown that caused it, so Sting gives the tackle less Credit instead of rewarding only the visible finish.
Sacks and tackles for loss are flat by position: their spot is known (at or behind the line), so a corner's sack earns the same as a tackle's sack.
A safety usually comes from a sack in the end zone, so the same play earns both the sack and the safety. Sting leaves that small overlap in place. The safety adds the two points and the ball back, which the sack alone does not capture.
Turnovers and blocks earn Credit for the possession or field-position swing. A forced fumble earns about half a turnover because the defense recovers roughly half of them. An interception earns the full turnover. The source does not identify the type of blocked kick, so every block receives one blended value below a blocked punt.
A takeaway is worth the possession it wins plus the return that follows. The research value of an interception, 4.25 expected points, is an average that already includes a typical return, so adding the return to it would count the return twice. When the source reports the return, the interception earns 3.35 for the possession, and the return earns the same value per yard as offense plus the touchdown premium. A 40-yard interception return for a touchdown earns 8.85. When the source reports no return, the interception keeps 4.25. The two routes agree on the average interception, so a reported and unreported returns have the same average value within the Credit rounding step.
Turnovers are valued highest, sacks and tackles for loss next, generic tackles least (before the position gradient above adjusts them). Kick blocks and hurries follow the same order: the bigger the swing in field position or possession, the more Credit it earns.
Special Teams
Field goals measure execution against the difficulty of the attempt. When official structured play-by-play supplies distance and result, Sting uses both. A make earns the three points the average kicker was expected to leave unconverted; a miss loses the three points the average kicker was expected to convert.
The formula is 3 × (result − expected make probability), where result is 1 for a make and 0 for a miss. That makes a long make worth more than a short make and a short miss cost more than a long miss. It also excludes the coach's decision to try the kick and the field position the offense created. Those are not the kicker's execution. The formula also does not model what happens after the kick, such as the kickoff after a make or the opponent taking over at the spot after a miss, so a long miss is charged slightly less than its full expected-points cost.
Make probabilities come from a frozen current-era college field-goal curve using kick distance. Sting linearly interpolates between its published anchor distances and clamps attempts outside the 17-to-60-yard evidence range to the nearest anchor. As with every stat, each field-goal total is rounded to 0.05 once, after its kicks are added together; the table shows single kicks. It does not adjust for the named kicker, stadium, weather, altitude, or hash mark because the source and calibration do not support those distinctions consistently.
| Attempt | Average make estimate | Make Credit | Miss Credit |
|---|---|---|---|
| 25 yards | 90.8% | +0.25 | -2.75 |
| 35 yards | 79.1% | +0.65 | -2.35 |
| 45 yards | 63.8% | +1.10 | -1.90 |
| 50 yards | 54.2% | +1.35 | -1.65 |
| 55 yards | 45.5% | +1.65 | -1.35 |
| 60 yards | 36.6% | +1.90 | -1.10 |
Source quality changes what Sting can know, never the definition it pretends to know. Exact reconciled attempts are preferred. If only made-distance ranges survive, Sting represents the 20s, 30s, 40s, and 50-plus ranges at 25, 35, 45, and 52 yards. If only totals survive, it uses an explicit 11,035-for-14,881 corpus baseline (74.2%): +0.80 for a make and -2.20 for a miss. Those frozen counts come from the published team-seasons in the calibration sample. They remain fixed, so a new season does not change the value of an earlier kick. Sting never treats a kicker's longest make as the distance of every attempt. When official play-by-play covers only some of a kicker's games, the covered attempts keep their exact values and only the kicks it does not cover use the range or total fallback, so no kick is counted twice. An attempt list that claims more makes, misses, or blocks than the box score records is not used, and ranges that add up to more makes than the line records are ignored.
A blocked field goal remains a different event. The event is estimated at −3.10, but only 25% is assigned to the kicker because protection and the snap carry most of the responsibility. The resulting −0.775 estimate is published as −0.80 Credit under the common 0.05 rule.
Punting Credit comes from distance. Each yard beyond the average punt is worth the same 0.05 points as a yard of offense, and each yard short of it costs the same. The average is 40.4 yards, taken from every punt in the completed 2019 to 2025 seasons after touchbacks. Because the comparison is with the average punt, distance earns nothing for punting often, only for punting far.
A touchback gives back the 20 yards the ball comes out to, which costs -1.00. A punt that leaves the offense inside its own 20 adds +0.20 on top of its distance, because a yard near the goal line is worth more than a yard at midfield. The box score does not record where a punt was kicked from, so a short punt from midfield reads as a short punt; the inside-the-20 bonus is the only correction for that. Returns use the same value per yard as offense, plus a touchdown premium.
Ratings
Agency: Who Gets the Credit
Agency assigns a quarter, half, or all of a play to the player. Box scores cannot recover an exact share, so a number such as 0.42 would claim precision the source does not provide. The same three shares apply across every stat.
- Assisted tackle: 50%. The box score marks the tackle as shared.
- Fumble recovery: 25%. The recovery depends heavily on where the loose ball bounces; the forced fumble receives separate Credit.
- Sack taken by a quarterback: 50%. Research assigns roughly half of sack rate over expected to the quarterback.
- Blocked field goal or punt charged to the specialist: 25%. Protection and the snap share responsibility; the defender receives separate Credit for the block.
- Safety: 50%. Field position helped create the two points and change of possession.
Two of those shares do different jobs. A half divides one play between teammates: passing and receiving yards split evenly between the quarterback and the receiver. A quarter leaves most of the play uncredited because the source cannot say who earned it.
Every stat outside the list above receives 100% agency. The box score cannot measure support such as run blocking, pass-rush pressure, or return coverage consistently, so Sting does not estimate it.
The 0 to 10 Rating
Raw Credit cannot compare positions directly. A quarterback records production on most offensive plays; a cornerback records only a few events per game. The Rating converts each raw total into a percentile within its position pool.
A 5.0 marks the middle of the recorded seasons at a position. A 7.0 marks about the 70th percentile. Ties receive the same percentile. A kicker's 7.0 and a cornerback's 7.0 rank the same within their roles even when their raw Credit totals differ.
The Rating is a rank, not a film grade or a claim about starter quality.
Every tile's badge is colored on this same scale, so the number and its color always agree.
| Rating | What it means |
|---|---|
| 8.0 and up | 80th percentile or higher within the position pool. |
| 6.0 to 7.9 | 60th through 79th percentile. |
| 4.0 to 5.9 | 40th through 59th percentile; 5.0 marks the midpoint. |
| 2.0 to 3.9 | 20th through 39th percentile. |
| Under 2.0 | Below the 20th percentile. |
Where Sting Uses Credit
Team of the Week
Team of the Week applies Credit to one game. Each performance ranks against single games at the same position. A regular-season week qualifies when most of the conference played. Postseason and light bye weeks do not qualify. The highest-rated player at each slot makes the team.
The offensive line rates as one unit because linemen have no individual box-score stats. Team rushing measures run blocking; sacks allowed measure pass protection. Both use half agency and stay separate from the running back's Credit. The selected unit names that game's five recorded starters. A unit needs a complete, verified starting lineup to appear.
Four individual awards also appear each week: Offensive, Defensive, Special Teams, and Freshman Player of the Week. Each goes to the highest weekly Rating at that scope. The freshman pool counts redshirt freshmen alongside true freshmen, since both are in their first year of college play.
Special Teams Player of the Week keeps the four roles separate: kickers rank against kickers, punters against punters, and each return role against itself. The winner is the best role-relative Game Credit Rating. Raw Credit breaks a Rating tie; it does not replace the role-relative comparison.
Players of the Season rank by Season Résumé Score, the same season-only Credit and Recognition used for All-Conference selections. Both Season Credit and Season Recognition must contribute to win a Player of the Season award, so recognition-only OL cannot win. Special teams uses the selected specialist role. Weekly honors continue to rank by Game Credit Rating.
Beasts of the Week
Beasts of the Week crowns the highest-rated crested weekly performance in each conference band at QB, RB, WR, TE, DT, DE, LB, and DB. A crest must have been earned before the season began, so nothing that happens later changes who was eligible that week.
Each conference first supplies one nominee per tier from its Team of the Week build. The band winner has the highest Game Credit Rating; ties break on raw Game Credit, then the earlier crest year.
Ratings retain their conference comparison pools. A 9.1 from one conference can beat a 9.0 from another even when the conferences differ in strength; Beasts does not invent a second cross-conference rating.
Bestiary Inductions
Bestiary Inductions records when a player first qualifies for an Archetype through Credit or Recognition in a league. This can happen after a game or when season-ending honors arrive. Later results do not erase that admission.
A game-based induction belongs to that college-football week; an induction earned through season-ending honors appears at the close of the season. Both Ratings remain with the earning record.
A player who entered the season holding a crest that carries into his conference band is not a Bestiary Induction. Every other first qualification appears once; there is no ranking or winner cut.
Game Recaps
Game Standouts shows Each covered team's three highest single-game Credit Ratings, ordered as First, Second, and Third Star. A performance ranks against single games at the same position and in the same conference, not against season totals. A two-way player appears once under his highest-rated role.
First, Second, and Third Star are narrative ranks over that existing order; they do not introduce a new score. Each player note uses the verified box score and leads with the strongest production for that role. Missing evidence shortens the note. Sting gives a team its own ranked standouts only when a matching covered game ledger exists. A final score determines whether the page becomes a recap.
Game Standouts is the default view. Full Ratings shows the complete offense, defense, and special-teams formations under the same single-game Credit method.
Player Narratives
Award notes use fixed rules, not generated opinion. They may use only the winner's recorded box score, opponent, award category, Credit rating, and supported season history. Every clause traces to one of those facts. Missing evidence shortens the note.
Weekly notes lead with the strongest production for the award category. They may add one supported fact, such as a conference lead, milestone, repeat award, or Credit streak. Season notes pair the strongest category stats with earned honors and repeat-award context. The score stays in the card's pill.
The voice follows Sting Coverage's editorial rules: lead with the football point, use precise football language, and separate evidence from projection. A note never infers effort, character, injury, athletic testing, scheme assignment, or film traits from box scores.
Where Credit Appears
- The three stats on a player tile are their top Credit earners for the season.
- The Credit lineup orders players within each position by recorded production.
- Sting All-Conference is Sting's common-game selection; official conference honors remain a separate Recognition input.
Reading a Player Tile
- Top three stats: the player's biggest Credit earners this season.
- Jersey number & team colors: who they are at a glance.
- Badges: a diamond for each All-Conference selection, a star for each Sting All-Conference pick.
- The big number: Rating (0–10) or raw Credit in points; switch with the toggle atop any roster page.
Lineups & Selections
Team pages offer two lineups. Two Deep is the school's published depth chart. Credit keeps the same positions and orders players by recorded production. Switch between them in the Display menu.
The offense fills a Flex spot. After the fixed skill slots, the last one goes to the highest-rated remaining running back, receiver, or tight end, so a standout second tight end can make the lineup over a third receiver. The Flex player appears in his natural row (a Flex tight end shows as a second tight end), not a labeled slot.
The defense starts two defensive ends, one defensive tackle, three linebackers, and four defensive backs. The Flex spot goes to the remaining defensive tackle, linebacker, or defensive back with the highest Credit Rating. The player stays in his position group, making the final defense a 4–3, 3–4, or 3–3–5.
Sting All-Conference picks are built from the games every team reached (the common regular-season slate), so a deeper postseason run cannot inflate the totals behind a selection. These teams rank every position by Season Résumé Score. Official conference honors are published by the conference and shown separately.
Season Résumé Score
Season Résumé Score averages the selected season's applicable Credit and Recognition Ratings equally. Missing required individual evidence leaves it unavailable; Build Fit and earlier seasons never count.
Sting All-Conference ranks fixed positions, Flex spots, and specialist roles by this score. Offensive linemen use Season Recognition because individual Credit does not apply; official honors are not required. Specialists retain their own role evidence and comparison pools. A dark score pill opens the two ratings and the season's participation and honors.
The highest eligible Season Résumé Score wins each Player of the Season award, using the same season-only Credit and Recognition as All-Conference selections. Both ratings must contribute; a one-factor résumé cannot win.
Perfect Season
Perfect Season drafts a fixed offense and defense from randomly drawn team-seasons, with a bounded reroll allowance per side. The roster has 6 offense slots and 6 defense slots. Each side begins with 1 reroll.
Weekly Fantasy
Weekly Fantasy scores a five-player offense or defense by the average Game Credit Rating of its picks in one college-football week. A pick with no Rating that week counts as zero, and the average always divides by five.
Each week's lineups lock at midnight ET before the conference's first game of the week, the earliest time that is safe for every kickoff. A pick's value is the Game Credit Rating from his game that week; when his team plays twice, the later game counts. Scores post Sunday, and lineups with the same score rank by which was saved first.
Archetypes
Earning One
Each complete Archetype has a short description of its model-defined identity and a separate authored character line. Bestiary Inductions captions retain the description for every player, including when a large class needs compact entries or several posts.
An Archetype is a persistent earned status for a qualifying player. A season qualifies when either its Credit Rating or cumulative Recognition Rating first reaches 8.5, after a game or when season-ending honors arrive. Sting does not average the two. Credit reflects recorded production; Recognition reflects verified starts, backup appearances, and official honors. Offensive linemen can qualify through Recognition because they have no individual Credit. Specialists cannot earn an Archetype.
Build Fit selects the creature when roster measurements are available. Missing classification evidence uses the conservative below-threshold pole rather than taking away an earned Archetype. Observed position stats select the skill modifier; when a qualifying season has no skill-axis event, Sting likewise uses that position's below-threshold modifier. Offensive line has no skill axis.
Once earned, Archetype status persists through quiet or down seasons. A later run of games falling below the gate cannot revoke the earn. Every later qualifying season computes a fresh classification, and the newest eligible earn becomes the displayed crest.
Transfers follow the same conference-band rule as Recognition. An Archetype earned at the same level or a stronger level carries to the new school. After a move to a stronger level, the old Archetype remains part of the player's career history but he must earn a new crest at that level.
The Two Axes
Two axes name the crest. Build Fit selects one of two creatures for the position. A position-specific skill lean selects the modifier, producing a name such as Sky Mage or Stone Mammoth. Offensive linemen have no skill axis, so Golem and Yeti are the only one-word Archetypes.
The complete name is an identity, not two independent labels pasted together. Each of the 34 Archetypes therefore has its own character line, written to express how that creature and modifier play together. Those lines bring the bestiary to life; the thresholds below remain the literal explanation of how a player earns the identity.
Both axis thresholds below were remeasured against every player-season that cleared the Credit-or-Recognition gate in a product-complete conference season from 2019 through 2025. Each player first clears the same gate within conference-local calibration Rating pools; only then are the qualifying measurements combined across conferences. Each axis targets a roughly 50–50 split at each position. Sting selects the boundary that minimizes the difference in group counts: half-point Build Fit steps, one-percentage-point share steps, and quarter-yard gain-rate steps. Tied measurements stay together, and missing measurements do not enter calibration. This does not require four equally common complete names. A skill lean describes a player's evidence relative to other eligible players at his position. These thresholds remain fixed: future earners do not move the line under an existing player.
For quarterbacks, skill lean compares passing involvement with carrying the ball. NCAA sacks count as rushing attempts in the box score, so Sting moves the observed sacks back to passing involvement: (pass attempts + sacks taken) / (pass attempts + rush attempts). Only his own passes and carries enter that comparison. Sky begins at 83%; Earth means more frequent non-sack carries for a quarterback. Kneels, scrambles and designed runs cannot be separated here.
Defensive backs now lean toward playing the ball (interceptions, their reported returns and pass breakups) or stopping the carrier (tackles, non-sack tackles for loss and sacks). Pick begins when ball-play Credit is at least 73% of those two families; Lock leans toward carrier stops. Forced fumbles, recoveries, fumble returns and hurries still count toward Credit but do not choose this DB lean. For tackles, ends and linebackers, sacks, hurries, breakups and interceptions remain pass-game work. Ordinary tackles can also occur on passes, so their run-side grouping is an approximation.
Lightning and Thunder describe higher and lower net yards per carry. Arrow and Axe describe higher and lower yards per catch, with separate receiver and tight-end lines. These rates cannot establish speed, power, route depth, blocking or short-yardage skill. A larger sample strengthens the evidence; it does not itself choose the modifier. Missing evidence keeps the documented fallback pole and never revokes an earned crest.
| Position | Build Fit threshold | Skill lean | Skill threshold |
|---|---|---|---|
| QB | 9.00 | Share of quarterback involvement through passing | 83% |
| RB | 8.50 | Yards per carry | 5.25 yd/carry |
| WR | 9.00 | Yards per catch | 13.50 yd/catch |
| TE | 9.50 | Yards per catch | 11.25 yd/catch |
| OL | 9.50 | No skill axis | - |
| DT | 8.50 | Share of Credit from pass disruption vs tackles and run stops | 48% |
| DE | 9.00 | Share of Credit from pass disruption vs tackles and run stops | 64% |
| LB | 9.00 | Share of Credit from pass game vs run game | 47% |
| DB | 9.50 | Share of Credit from playing the ball vs stopping the carrier | 73% |
Archetype catalog
Nine positions, each with its own pair of creatures and, offensive line aside, a pair of modifiers shared with the positions that play the same way. Every combination below is a real, earnable Archetype: 34 names in all.
| Position | Names |
|---|---|
| QB | Sky MageEarth MageSky WizardEarth Wizard |
| RB | Lightning RamThunder RamLightning BullThunder Bull |
| WR | Arrow ViperAxe ViperArrow CobraAxe Cobra |
| TE | Arrow WyvernAxe WyvernArrow DragonAxe Dragon |
| OL | GolemYeti |
| DT | Siege RhinoStone RhinoSiege MammothStone Mammoth |
| DE | Siege PantherStone PantherSiege LionStone Lion |
| LB | Spear HoundHammer HoundSpear WolfHammer Wolf |
| DB | Pick HawkLock HawkPick EagleLock Eagle |
The sample player on the tile above wears a crest: Siege Panther, earned in 2024.
Each league publishes its bestiary as a hall of fame: every player who earned an Archetype in that league, grouped by position, newest inductions first. A crest carried in from another league stays on the jersey but does not open a seat until the player earns one in that league. Evolving to a new creature extends a member's chain without re-inducting him; the seat keeps its original induction year.
Build Fit
What Build Fit Measures
Build Fit asks two questions: does the player's listed height fit the position's preferred frame, and does his listed weight fit that frame? Sting scores both on the same straight line, then counts frame twice. Higher is a closer fit to the preferred build.
Build Fit is an absolute 1.0 to 10.0 model score, never a percentile. The same measurements and position produce the same score on a conference board, the national board, and a game preview. The ordered color bands are consistent with Sting's other scores: higher Build Fit is better Build Fit.
Build Fit does not measure strength, speed, movement, skill, or film quality. Credit records production. Recognition records verified starts, backup appearances, and official honors. A player can fit the preferred build and still play poorly, or fall outside it and play well.
Combine research supports the same limit. Measurements can help identify professional prospects, but they do not predict long-term success by themselves.
Build Fit counts Frame Fit twice and Mass Fit once. An 8.4 means an 8.4 out of 10 fit with the model, not the 84th percentile and not an 84 percent chance of success.
Reading Build Fit with Credit
Read Build Fit beside Credit: it describes listed build, Credit records production. The four combinations mean:
| Build Fit | Credit | Reading |
|---|---|---|
| High | High | Looks the part and produced. The projection and the box score agree. |
| High | Low | A prototype body whose production has not matched it yet. |
| Low | High | A productive outlier who beat a non-prototype build. |
| Low | Low | Limited production from a less conventional build. |
Frame Fit and Mass Fit
Frame Fit scores listed height against the position's preferred range. Mass Fit scores listed weight after accounting for that frame. Frame counts twice what mass counts: a program can add or trim fifteen pounds, but it cannot add an inch.
Frame
Each position has a preferred height band. A listed height inside the band receives full Frame Fit. Outside it, the score falls with distance from the nearest edge. One inch short and one inch tall receive the same penalty.
Every inch outside the band costs a full point, on both sides alike, so a build three inches off the prototype reads three points below one that is on it.
The bands are the position-specific part: each position's band covers the couple of inches its role usually asks for, and a build just outside it still reads as near-prototype. These are prototype preferences, not cutoffs. A great player below the band is an outlier the score is meant to place under the prototype, not evidence that the prototype is wrong.
Every position on the field carries its own version of that same band: a receiver's runs a few inches taller than a running back's, a lineman's taller still, each one a football judgment about the role rather than a fixed rule.
Mass
Mass depends on frame because the same weight carries differently at different heights. Sting restates listed weight at the midpoint of the position's preferred height band: three pounds for each inch, capped at four inches in either direction. The player's listed weight never changes.
That translation is Scoring Mass, not a second listed weight. A taller player reads a little lighter and a shorter player a little heavier before Mass Fit is scored.
Every position uses the same three-pound adjustment. Only the preferred height and mass bands change by position.
Mass uses the same straight line. A scoring mass inside the preferred range receives 10.0. Outside it, each step costs half a point, and a step is about 2.5 percent of that position's prototype mass: roughly six pounds at linebacker, eight on the line. Any part of a step counts as a whole one, the same way any part of an inch would. There are no position-specific tail curves.
The model sees only listed height and weight. Arm length, wingspan, hand size, speed, body composition, leverage, bend, and play strength remain outside the data.
Completed seasons start with that season's listed height and weight. When a historical roster row is missing a measurement entirely, Sting recovers it from another confirmed season for the same player. Schools also sometimes lower a listed height after remeasurement. Because players do not shrink, Sting corrects the earlier, taller listing down; genuine growth remains intact, and a present seasonal weight is never rewritten. If a measurement remains missing, malformed, or implausible, Build Fit is unavailable, but missing classification evidence never erases an Archetype already earned through Credit or Recognition. Sting uses one generous integrity envelope for every position: 4'6" to 7'6" and 100 to 500 pounds. Those limits catch broken feed values; they are not position ideals.
Prototype vs Outlier
Great NFL players do not automatically define ideal size. Some fit the preferred build. Others overcome a size concern through traits and production that this model does not measure.
Aaron Donald is the cleanest warning. He is an all-time interior defender despite being short for the prototype, not evidence that 6'1" is the ideal defensive tackle height. Outliers can soften the penalty where the sport has proved they can win, but they should not move the peak of the model to the outlier.
| Position | Prototype examples | Outlier examples | Lesson |
|---|---|---|---|
| QB | Josh Allen / Justin Herbert type | Kyler Murray / Bryce Young type | Prototype frame helps the prior; accuracy, processing, and creation still decide the player. |
| RB | Compact, dense 5'11"-6'1" backs | Derrick Henry as the tall exception | Running back has the clearest tall penalty, but rare traits can beat it. |
| WR | X receiver length plus slot/Z alternatives | DeVonta Smith / Tyreek Hill style exceptions | The WR bucket is broad, so Build Fit should avoid one-archetype absolutism. |
| TE | Travis Kelce / George Kittle / Rob Gronkowski frame | Undersized move-TE bodies | TE needs more frame and functional mass than WR. |
| OL | Trent Williams / Penei Sewell type | Shorter interior exceptions | Frame clears the projection gate; mass-readiness matters too. |
| DT | Chris Jones / Dexter Lawrence type | Aaron Donald / Calijah Kancey type | Do not move the prototype band to the outlier. Great leverage can beat a size concern. |
| EDGE | Myles Garrett / Nick Bosa type | Shorter bend-and-quickness rushers | Length is the prior; bend and explosion can beat it. |
| LB | Fred Warner / Roquan Smith type | Safety-sized hybrids | The role is changing, but mass still matters against blocks. |
| DB | Sauce Gardner / Patrick Surtain II type | Small speed-and-technique corners | The current DB bucket mixes corner, safety, and nickel bodies. |
What the Buckets Flatten
Build Fit grades nine position buckets, and several fold roles a scout would keep apart. The model is honest about where it is coarse.
- Defensive line respects an explicit role: DE/EDGE uses the edge model and DT/NT uses the interior model. For a generic DL listing, the shared line-role method first compares height-adjusted listed weight with the frozen average end and tackle builds. Build Fit then scores only that resolved role; it does not choose whichever prototype would award the higher score. Weight never switches an explicitly labeled end or tackle into the other model.
- Defensive back mixes corner, safety, and nickel, which want different builds.
- Offensive line mixes tackle, guard, and center.
- Linebacker mixes off-ball, edge-adjacent, and hybrid bodies.
- Receiver mixes outside, slot, and return types, so one band cannot fit them all.
Special teams has no Build Fit. That covers kickers, punters, and long snappers, and it covers a receiver or defensive back shown in a returner role. His listed build was scored against the position he is listed at, not the role he is ranked in, so the board shows no Build Fit for that row and leaves it out of his Prospect Score and his Gem Score.
Bands by Position
Every position has its own prototype bands, scored on the same straight line, the same three-pound frame adjustment, and the same 2-to-1 frame-mass weighting. Special-teams specialists are not graded, so they have no bands here.
Quarterback frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Running back frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Wide receiver frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Tight end frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Offensive line frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Interior line (DT) frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Edge (DE) frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Linebacker frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Defensive back frame-led
A build inside both preferred bands scores highest. The same straight line handles measurements above and below them.
Recognition
What Recognition Measures
Recognition scores official all-conference honors earned through the selected season. Higher honor tiers earn more points, and a unanimous first-team selection earns a bonus. An honor counts toward the position it was won at: a return honor is return recognition, not receiving recognition.
Every verified start adds a smaller fixed amount, and every game a player was listed as a backup adds a smaller amount again. The two never stack, so one game counts once, as a start or as a backup appearance. Starts and backup appearances separate experienced players who have no official honors, while the larger honor values keep selected players ahead. At a twelve-game season, four years of backup appearances stay under one season as a starter; a longer eligible career can close that gap.
Starts and backup appearances come from official lineup records. If a lineup conflicts with itself, Sting does not infer the answer from playing time or season continuity. A correction needs another source or direct verification from the player or team.
A transfer keeps the history earned at every school. Starts and official honors continue to count when they came from the current conference band or a stronger one. Lower-band evidence stays visible after a move up but does not raise Recognition. Sting awards remain visible and add no points.
How Transfer Careers Are Matched
Sting connects seasons at the same school through the source profile and matching roster details. A transfer match requires the same normalized name, one resolved previous school, and either the same source identity or two stable roster facts: hometown, high school, or home state. A name alone is never enough.
Each confirmed move extends the same career, so a player can carry history through three or more schools. Matching changes the career identity, not the facts underneath it: every season, start, honor, award, and stat lead remains labeled with the school and conference where it happened.
When a previous-school label is missing or ambiguous, or more than one earlier player fits, Sting leaves the records separate. A reviewed correction can connect them later when stronger evidence is available. An incomplete career is better than attaching another player's history.
How Recognition Becomes a Rating
Recognition appears as raw points and as a 0 to 10 percentile within position and conference. Verified evidence below every peer receives a 0.0 Rating. Without verified starts, backup appearances, or official honors, Recognition is unavailable. Production alone supplies Credit evidence and does not supply Recognition evidence.
Verified starts and backup appearances supply Recognition evidence to players without official honors.
The comparison pool includes contributors with verified Recognition evidence at that position. Production alone does not enter it. Specialist-role pools use verified starts, backup appearances, and official honors attributed to that role. Build Fit does not use that pool because it is not a percentile.
Season Recognition
Season Recognition scores only the selected season's verified starts, backup appearances, and official All-Conference honors within the position or specialist role. Its comparison pool contains season-only evidence from the target season's named completed source seasons. Starts add 0.05 points and backup listings add 0.01, using the same weights as cumulative Recognition. Earlier seasons never contribute.
Career Units
One Unit at a Time
The player card's career table answers one unit at a time: offense or defense at the position where the player was listed or observed, plus each special-teams role on its own. Place kicker, punter, punt returner, and kick returner never share one rating.
Every unit follows the same scoring path. Credit is the unit's production against its pool. Recognition on offense or defense is the honors earned at that unit, plus 0.05 per verified start and 0.01 per backup listing made there. Special-teams roles use those same weights for their attributed starts and backup listings, plus honors won at that role. Both are scored against the same conference pool the season Recognition uses, so splitting a player's evidence by unit changes what counts, never how it is counted. A listed player and a player observed at the position use the same formula.
Because the pool is shared, a player whose whole body of work sits on one unit reads the same there as he does in his season rating. A player who splits his work rates each unit on that unit's share alone.
Build Fit scores the player's listed body against the unit position's prototype. Special-teams roles carry no Build Fit because no useful returner, kicker, or punter body prototype exists. Unit ratings use the program's own recorded evidence. The season ratings above them keep their existing definitions.
Top Prospects
The Résumé Score
Résumé Score equally averages a player's available Credit and Recognition: what the player has established, without Build Fit. Credit uses the player's best season through the selected year, and Recognition includes every qualifying start and honor earned through that year.
Team Rankings, Team Outlook, and preview lineups rank players by the Résumé Score, so a listed build never moves a player up a depth chart. On the Top Prospects board, switch from Prospect Score to Résumé Score to rank the same players without Build Fit.
The Prospect Score
Prospect Score equally averages a player's available Credit, Build Fit, and Recognition. Any player who appeared in a game that season can qualify.
Each factor uses its own window. Credit uses the player's best season through the selected year. Build Fit uses that season's repaired and height-anchored listed build. Recognition includes every qualifying start and honor earned through that year.
Sting rates a player at every position he produced at, including a special-teams return role, and the board lists him under whichever one scores highest; a tie keeps his roster position. Build Fit always follows his roster position, since his body does not change with the role.
Offensive linemen have no individual Credit because Sting rates the line as a unit. Their Prospect Score averages Build Fit and Recognition.
Kickers rank against kickers and punters against punters. Build Fit stays out because listed build does not project specialist performance. Their Prospect Score averages Credit and Recognition. The default All Positions view hides specialists; use the Kicker or Punter filter to rank them.
Switching to the Résumé Score removes Build Fit from the average. The score and rankings update immediately.
Credit and Recognition percentiles use their named target-year normalization sets, with the historical-conference rule for carried Credit. Build Fit stays absolute. The selected year changes the eligible players and evidence, but never changes what a listed height and weight score.
The National Board
The national board keeps Credit and Recognition in their conference context. One mixed pool would erase differences in production and honors across leagues.
Build Fit uses the same absolute model on conference and national boards because listed build does not change meaning by league. It is carried across without a second normalization step.
Group of Six, Flagship Six, and conference filters change the candidates, not their Build Fit.
Team Rankings
Team Rankings use each team's top four offensive and top four defensive players by the Résumé Score of that unit's Credit and Recognition. Return, kicking, and punting evidence stays out of both sides. A two-way player can qualify separately on each side. The team score is the average of those Résumé Scores. A side with fewer than four qualified players averages the players available.
Offensive linemen qualify through Recognition. Preseason rankings use the new season's rosters and carry forward what each player established in earlier seasons.
Game Previews
How a Preview Is Built
Each watch list includes a team's top three rated offensive players, top three defensive players, and available specialists. Hot Last Time Out adds the best previous-game Credit performer on each side who is not already listed. Opening week never borrows a game from the previous season.
Matchups to watch pair natural opponents: receiver against defensive back, back against linebacker, and quarterback against the pass rush. They do not predict a staff's coverage or assignment.
Build Fit at the line averages the same player-level Build Fit used on prospect boards across each projected starting front. It does not measure strength, movement, or line play.
Every panel names its basis year. Before the new season produces data, previews use the most recent completed season. Cross-conference games keep each side's own-league percentiles. An opponent outside Sting's conferences receives no rating and reads "not covered."
Game Preview Matchups
Matchups to Watch
A Contest pairs two players with season Credit Ratings of 8.5 or higher and ranks pairs by their combined rating. A Mismatch pairs a player rated 8.5 or higher with an opposing unit that trails him by at least 2.0. The unit average uses projected starters and needs at least 2 rated players. The unit, not one opponent, appears as the weaker side. Every number remains a 0 to 10 percentile within the player's own league.
Natural pairings include receiver against defensive back, tight end or back against linebacker, and quarterback against the pass rush. Linemen have no individual Credit, so their comparison appears under Build Fit at the line. The Contest ranking covers both directions of the game. Mismatches rank by the gap between a player and the opposing unit. The preview skips a Mismatch if its player already appeared.
What "Starter" Means Here
Group averages use the projected starters shown in the formation. Individual watch-list players can come from the full rated roster, so a productive backup can still appear. If the lineup cannot be resolved from the available source files, a group average falls back to all rated players in that unit instead of presenting an incomplete subset as the starting group.
Build Fit at the Line
Build Fit at the line averages each listed starter's Build Fit; it does not score the unit's average body as if it were one player. The height and weight rows summarize the same starters, while Frame Fit and Mass Fit summarize their component scores. A gap of 1.0 or more reads Mismatch; a smaller gap reads Contest. The same all-rated fallback applies when the lineup cannot be resolved. Missing Build Fit data leaves the row unlabeled.
Expected to Play
The most recent usable game sheet in the fixture's football season supplies Listed Starters and everyone who played. A usable sheet has 18 to 24 starters. Only games before the fixture count. A bye changes nothing. A newer sheet without usable starters is skipped for an earlier usable one.
When no game has a usable sheet, the latest game's recorded participants supply the matchup pool, Listed Starters is unavailable, and the line comparison uses rated linemen. Before a team has played, its preseason roster supplies the pool. This is not an injury report.
Top Credit asks the same Season, Total lineup owner as the team's page. Both flexible slots and its preseason Résumé Score ordering follow that owner. Individual offensive linemen carry no Credit pill.
Archetype Matchups
Each preview lists up to five Contests, ranked by combined Rating. A Contest is one player from each side who face each other: an Archetype pair on its skill lean, or a position pair such as a receiver against a defensive back. Both players need a Rating of 8.5 or better on the side's own basis (Best Credit before its first game, Partial Season Credit after), and a pair that misses the bar never uses up a player. Each player appears in one Contest, and only players who played last time are eligible once a team has a game. A badge shows the earning season’s Build Fit and skill lean when those values were recorded.
Pass rushers can meet quarterbacks, pass-leaning linebackers can meet tight ends or quarterbacks, and ball-playing defensive backs can meet receivers or tight ends. Run-leaning defenders can meet running backs or Earth quarterbacks. Position pairs enter the same ranking. Ties go to the higher individual Rating, then to the offensive and defensive surnames.
On a headline badge, each meter places the value recorded when the player earned the crest against the position’s frozen threshold. A carried badge names the earning season when it differs from the season on screen. The web preview can show Build Fit and skill lean. The social card shows only skill lean. A badge with no recorded values keeps its crest and shows no meters.
The Team Outlook
Team Outlook removes graduating players and ranks position groups by returning production and recognition. The Team Outlook position-group template assigns the two flexible formation places to a third receiver and a second defensive tackle so every team is measured against the same group capacities; it does not reproduce a particular projected lineup's Flex choices. Each returning player is valued by his Résumé Score, so listed build stays out. Group strength sums the selected players' Résumé Scores across the template's slots, with an empty slot counting zero. The prospects panel shows the team's top five Prospect Scores.
The list measures proven production. Promising but untested underclassmen count as a need until they produce. That describes the record, not their talent.
Each need links to potential transfer targets at that position. The board requires a known, non-graduating class and limits candidates by competition level. These players are scouting candidates, not confirmed portal entrants. Roster data does not establish portal status or availability.
Sources & Limits
Sources
- Advanced Football Analytics: Expected Points
- nflfastR: Exploring Expected Points Added
- cfbfastR: College Football Expected Points Model
- cfbfastR: Field-goal probability and derived models
- Puntalytics: Punt Metrics
- Football Study Hall: Putting a Point Value on Special Teams
- PFF: How Important Are Sacks for a Defense
- PEP: Tackle Value in Football (arXiv)
- NFL Scouting Combine: standardized prospect measurement and testing context.
- Szekely et al.: NFL Career Success as Predicted by NFL Scouting Combine: combine data can help identify NFL matriculation, but long-term success remains much harder to predict.
- NFL.com player profiles used as illustrative measurement anchors, not proof of the bands: Justin Herbert, Myles Garrett, Chris Jones, Sauce Gardner, and Aaron Donald.
From Source to Screen
Sting's historical target covers 2019 through 2025. Sting collects official school and conference records: season and weekly box scores, rosters, schedules, depth charts, all-conference teams, and major awards. Missing source data remains missing. Sting does not invent a season or an unpublished field.
Processing connects the same player across sources and adds roster details and honors only when the identity match is strong enough. Weak matches remain unresolved for review. Official fallback sources may fill a gap but never replace a known value to make coverage look complete. Stable bio fields can carry across a confirmed player match, but completed-season height and weight remain specific to the season that listed them. A labeled preseason projection can use the latest prior roster as its measurement basis.
Every refresh checks teams, years, roster completeness, honors, and identity matches. A clean refresh can still carry missing source data: it means Sting processed the evidence available, not that every source published every field.
Which Records Own the Answer
Official school and conference records supply the evidence. Each season uses the conference the team belonged to then, so a later conference move does not relabel a player's earlier honors. Confirmed player identities connect the records; similar names alone do not establish a career.
Each team-season has one body of evidence. Strongly matched fragments of a player's season are combined before scoring, so overlapping records never count as separate games. Starts and reserve appearances come from game-by-game participation decisions, and every published count must match them. For a confirmed split season, the headline Credit and unit Credit breakdowns use the same combined production, with each unit rated in its own scope. Unresolved identities and missing game evidence remain visible limitations.
A live roster-only row does not enter a Rating pool or prospect board. Its verified player identity can still preserve the stable link to earlier contributing seasons whose source rows had no slug. Identity continuity does not turn the roster-only season into production evidence.
Each conference and target year has a named normalization set: the comparison population used to turn a score into a Rating. That set records its source years. The season's Credit window records whether the performance is complete or covers a particular number of games. These records, together with the historical affiliation, own the comparison; a destination team's current league does not.
Each season's Credit Rating is computed once. Career views, prospect boards, rankings, previews, narratives, and social cards all show that same Rating, so a player's number never differs between pages. A corrected source record updates every page that uses it.
A Credit Rating exists only when the record contains production in the position or special-teams role being scored. Roster membership, appearances, starts, and production on another side do not create Credit evidence. A production record can still be eligible when its games-played counter is missing, and a valid low-end Rating can be 0.0. Missing Credit remains absent rather than becoming a zero Rating.
A prospect board lists a player at a special-teams role instead of the roster position when the role's evidence is stronger. A role or position with Credit always outranks one without it, so a returner with real return production is never listed at a position where the player produced nothing.
Credit After a Transfer or Realignment
A season belongs to the conference the team played in that year. Its Credit Rating uses that conference's target-year normalization set and the season's own Credit window. A later transfer, or a team's move to another conference, carries the finished Rating forward. It does not rescore the old production against the new conference's players.
Best Credit is the highest eligible Complete Season Credit through the selected year. A tie uses the more recent season. The displayed year and contributing-stat explanation come from that same winning season. Every candidate is measured against completed seasons. A season still in progress competes on the Complete Season Credit it would hold if it ended today, so a strong opening stretch cannot outrank a full season until the production adds up. The season's own Rating elsewhere on the site is its Partial Season Credit, which compares its games so far with the same number of opening games and can read higher.
A historical conference outside Sting's supported coverage, a withheld incomplete conference population, or an unscorable season cannot supply normalized Credit. Sting does not borrow another conference's pool to fill the gap. Another eligible season can still supply Best Credit; if none can, it remains absent. Missing Credit is not a zero Rating.
Recognition answers a different question: how much eligible participation and honor evidence the player has accumulated by the selected season. It uses the destination's Recognition comparison population and the transfer rules described above. Carrying an old Credit Rating does not mean carrying an old Recognition percentile unchanged.
The Normalization Basis: Which Seasons Count
A normalization basis names the exact source seasons used to judge one conference's target year. It is not simply all the data currently loaded, and it is not selected by the date you open the page. Only eligible completed source seasons enter the basis; an unfinished source season does not become calibration evidence.
The data's as-of date helps decide whether a season is complete before any years are selected. During the season window, that decision uses the conference schedule's regular-season endpoint; missing or inadequate schedule evidence keeps it incomplete. After the season window has ended, the date rule treats it as closed. Separately, conference-year product coverage must be eligible. Being chronologically complete does not certify that every box score or roster field is present.
- Prior complete seasons: when at least 3 eligible seasons precede the target year, use all those predecessors, not just the most recent three.
- A live target season: use only completed predecessors, even when fewer than 3 exist. Never include the live target or a later season. With no eligible predecessor, a normalization basis is unavailable.
- Early completed history: when a closed target has fewer than 3 eligible predecessors, use the conference's first 3 eligible completed seasons as inaugural calibration. This deliberately retrospective basis can include the target and later seasons. It is not a claim about what could have been known during that historical season.
- Limited completed history: if fewer than 3 eligible completed seasons exist in total, a closed target uses the available completed history at first publication. With none, no basis is available. Sting does not substitute a different league to manufacture a population.
For example, if a conference's first eligible completed seasons are 2019, 2020, and 2021, those years can calibrate its early closed seasons. A target of 2022 uses 2019–2021 as predecessors; a target of 2023 also includes eligible completed 2022. Missing or ineligible years are not filled in merely because they fall inside that calendar range.
Once published, a basis keeps its exact source years. Newly available history does not widen it, and a source year that stops qualifying is never silently replaced. A limited-history basis likewise does not grow when coverage expands.
One Saved Set, Four Comparison Pools
Each target year's normalization set builds four kinds of comparison from the same saved source years: Complete Season Credit, same-length Partial Season Credit windows called Games(G), single-game Game Credit, and Recognition. These share source years, not identical observations: each pool measures its own kind of performance and applies its own eligibility rules.
The live-season comparison uses the same number of team games from completed source seasons. Game Credit instead compares individual games. Recognition uses contributors with verified Recognition evidence from the selected completed seasons. The set records population counts and sufficiency separately for each pool and position tier, including each available Games(G) window.
A real contributor has recorded production, a verified start or backup appearance, or an official all-conference honor in that season. A roster listing or games-played total alone is not enough. Each factor keeps its own evidence rule: production-only players do not enter Recognition pools, while backup-only players can enter them without receiving Credit. Recognition points are bounded through each source season. Specialist-role pools use verified starts, backup appearances, and official honors attributed to that role.
Fewer than 20 observations is insufficient: that tier's normalized Rating is withheld. From 20 to 49 observations, the pool is thin but can publish; 50 or more is adequate under this sample-size rule. These labels describe population size, not a guarantee that the underlying records are complete or that the model predicts future success.
Source corrections can change a set's pool values while its source years stay fixed, so a fixed membership never freezes an error in place. Awards are scored against the same Season Credit, Games(G), and Game Credit pools the published page explains.
Player History Cutoffs Are Separate
A selected season bounds the player's own evidence: later seasons cannot contribute their production, honors, or a future-earned crest to that row. This is separate from the explicitly retrospective calibration allowed for early closed seasons. A career view can use the available career history; a preseason projection uses prior evidence and is labeled as a projection.
Weekly honor badges use earlier-season honors. A weekly crest comes from the recorded crest held going into that season; without that prior record, it is omitted rather than guessed from the end-of-season crest. Corrected historical records can revise results, so an evidence cutoff is not a promise to reproduce a past page exactly as it originally appeared.
The Data
Credit comes from official school records. The common layer is season and weekly box-score totals: completions, yards, touchdowns, tackles, and other recorded events. Some structured feeds also publish individual plays. Sting currently uses that richer evidence for reconciled field-goal distance and result, and uses the documented range or aggregate fallback for any kick it does not cover. A field not present in the accepted source remains invisible to the model.
Season stats and Ratings cover each team's first games in common, the number every team in the conference reached that year; postseason games are shown separately and rated by week. If incomplete source evidence makes that shared window untrustworthy, Sting leaves the season uncapped and identifies the limitation. Within a capped season, each longest-play field is the largest value in the selected weekly game lines, so a later postseason maximum does not enter the shared window.
Heights, weights, and hometowns come from each school's published roster and are listed as the school reports them. Those figures can run generous, get revised from one season to the next, and a few programs do not publish them at all, so some players carry no Build Fit.
The flag beside a player's name is read from that hometown. A hometown that names a US state or territory carries that flag, and a hometown that names another country carries its own where Sting holds one. Rosters spell the same place many ways, so the reading is normalized first: "U.K.", "England", and "GBR" are one country, and a Hawaii hometown written with an okina, or by island rather than by state, still reads as Hawaii. A country Sting holds no flag for carries a generic international mark instead. Where a city shares a country or state name, the final listed region decides: "Spain, Ohio" is Ohio, while "Washington, England" is England.
A player whose hometown places nobody carries no flag at all. That includes a blank hometown and a city listed without a state or country. A missing mark means the roster never said, and it is never a claim that the player is from outside the United States.
Official all-conference teams and major awards come from conference releases. Recognition uses only honors that can be matched to a rostered player with sufficient identity evidence. An unmatched listing stays unmatched; Sting never attaches the honor to the nearest name.
Credit and Recognition Ratings are 0 to 10 percentiles within their defined position and conference comparison populations. Credit belongs to the scored season's historical conference; cumulative Recognition uses the target conference's eligible evidence and named comparison set. The source years are explicit, rather than whichever seasons happen to be loaded. Build Fit is different: it is an absolute 1.0 to 10.0 model score and is directly comparable everywhere. A high Build Fit means a closer match to the preferred listed build, not stronger production or a better player.
While a season is in progress, its Partial Season Credit compares a player with others measured over the same number of team games. Two games into a season, a receiver is ranked against other two-game starts instead of full seasons. Finished seasons share the full-season pool, where only Complete Season Credit applies.
Game previews compare those normalized Ratings even when the teams have played different numbers of games. When one team has started and the other has not, the first uses its actual in-season Rating and the second uses its career-best preseason Rating. The evidence bases remain visible as provenance; they do not block qualified Contest or Mismatch stories. A direction that earns neither stays empty instead of filling the card with a best-available fallback.
Archetypes compare each run of games with finished-season pools. A player earns one after a game or when verified season honors bring Credit or cumulative Recognition to the 8.5 gate. That first qualification is permanent; a later dip below the gate does not revoke it. The crest's type can change in a later season as the player's production and role change; keeping an earned qualification does not freeze the first type forever.
Offensive linemen carry no individual Credit, because the line is graded as a unit, so their score leans on Build Fit and Recognition. Special teams is rated by role, so kickers and punters are ranked among their own.
What This Can't Do
Outside the specifically documented play-aware field-goal path, Credit uses fixed event values. A third-and-long incompletion and a first-and-ten incompletion therefore look the same. Each costs the same Credit regardless of context.
Sting cannot see tips, drops, or blown pass-protection assignments. A wide receiver who drops five catchable balls earns the same reception Credit as one who catches everything he touches. A quarterback who escapes pressure and converts third downs gets no premium for it.
What Production Leaves Out
The best player on a unit can be the one the box score never mentions. A lockdown cornerback gets thrown away from, so he makes no plays on the ball and the model reads zero. A defensive tackle who draws a double team every snap frees the teammate beside him, and that teammate banks the Credit. Sting sees what happened, never what a player deterred or set up.
Production is shared, and the totals cannot say who created it. A great receiver turns a loose throw into a long gain and lifts his quarterback's line; a sharp quarterback makes ordinary receivers look open. Passing and receiving yards split evenly because that is the honest default; the split cannot point to the player who actually earned the gain.
Take Credit for what it is: a measure of recorded production, and only that. It belongs next to what you saw on the field, not in place of it.
Archetype eligibility reflects that limit. A completed run of games or verified season honors can clear the 8.5 gate through Credit or cumulative Recognition. Either is enough; Sting records when the player first qualifies. Verified starts, backup appearances, and official honors can therefore supply evidence that recorded production may leave out.
Both Ratings are read at the position a player is listed at. An honor won returning kicks counts toward his return Rating, not toward the running back or receiver he lines up as, so a return specialist earns a return specialist's record. That is why a player can carry a high Recognition on his return work and still hold no Archetype at his listed position.