Tony Pollard turned 28 offensive touches into 38.00 FanDuel points in Week 7. Jahmyr Gibbs had 34 touches and finished with 17.80. Both topped 140 yards from scrimmage. One scored three touchdowns; the other scored none and lost a fumble.
That contrast captures the central question in this FSN DFS Week 7 report card: are we evaluating opportunity, production or the final fantasy score? They are connected, but they are not interchangeable. A substantial workload can accompany a major projection miss, while a handful of scoring plays can transform an otherwise modest stat line.
This retrospective uses the original archived forecasts, not numbers recalculated after the games. It evaluates 269 matched DraftKings records and 266 FanDuel records. All belong to the older archive, whose exact equivalence with every projection displayed by the public builder was not recorded. These results are not a grade for today’s model.
Week 7 DFS accuracy: the full matched sample
| Measure | DraftKings | FanDuel |
|---|---|---|
| Matched records evaluated | 269 | 266 |
| Mean absolute error | 5.14 | 4.79 |
| Bias: actual minus forecast | -0.29 | -0.20 |
| Root mean square error | 7.0 | 6.6 |
| Coverage of archived offensive forecasts | 70.60% | 69.82% |
Errors are fantasy points. Mean absolute error describes the average size of a miss without allowing overestimates and underestimates to cancel. Bias preserves their direction; a negative number means forecasts ran high on average. Root mean square error gives large misses greater influence.
The small overall biases conceal a useful positional split. Running backs outscored their forecasts by about 1.08 DraftKings points and 1.04 FanDuel points per matched record. Receivers finished about 1.00 and 0.75 points below expectation. Those opposing tendencies partly canceled in the overall average.
That is why being close in aggregate is not enough. A lineup needs useful player-level estimates, not an accounting balance between unrelated misses. Nor does FanDuel’s lower absolute error establish a superior forecast: scoring scales and the included samples differ.
Running backs: Pollard’s touchdowns and Gibbs’ heavy workload
The Titans’ official game log credits Pollard with 25 carries for 129 yards and three touchdowns, plus three receptions for 26 yards. His 39.50 DraftKings points exceeded the archived 8.32-point forecast by 31.18. On FanDuel, 38.00 against 7.89 produced a 30.11-point underestimate. Both were the largest positive errors in their matched samples.
Touchdowns explain much of the result, but not all of the miss. As an arithmetic decomposition, removing only Pollard’s 18 touchdown points would still leave 21.50 DraftKings points and 20.00 FanDuel points. His yardage, receptions and rushing bonus alone substantially exceeded the saved forecasts. That does not reconstruct a different game; it shows why dismissing the whole gap as touchdown variance would be inadequate.
Gibbs presents the opposite problem. His official Week 7 log records 29 carries for 116 yards, five catches for 27 yards, no touchdowns and one lost fumble. The FanDuel forecast was 34.12; his 17.80-point result missed it by 16.32.
A lack of touches is not a credible description of that outcome. Gibbs accumulated 143 scrimmage yards on 34 touches, but his fantasy total did not approach the forecast. The archive does not preserve a usable breakdown of the assumptions behind that mean, so we cannot assign the error specifically to expected touchdowns, efficiency or workload. We can identify the question: was the final expectation too aggressive even for a heavily involved back?
Kenneth Walker III had a different kind of disappointing result. The Chiefs’ game log shows 14 carries for 36 yards and four catches for 21 yards in Seattle. His 9.70 DraftKings points fell 19.59 below the 29.29 forecast; 7.70 FanDuel points fell 20.43 below 28.13.
Walker received 18 touches but generated only 57 scrimmage yards and no touchdowns. Kansas City lost 20-3. This was a poor offensive result, not merely an absent yardage bonus. Together, these three backs show why a report should retain the score error while distinguishing what actually happened on the field.
Quarterbacks: Nix’s rushing contribution changed the comparison
Bo Nix’s official line against Arizona included 222 passing yards and one passing touchdown, without an interception. Those passing statistics contributed 12.88 points. His seven carries for 73 yards and a touchdown contributed another 13.30.
The combined 26.18-point result beat a 9.81-point forecast on each platform by 16.37. More than half of his points came on the ground. Evaluating this as a passing-yardage miss would overlook the part of his performance that mattered most to the final score.
That does not prove the old model ignored quarterback rushing. The archived total cannot establish its internal reasoning. It does justify checking whether rushing opportunity and explosive runs were represented adequately when these forecasts were produced.
A closer quarterback forecast came from Kirk Cousins. The Raiders’ box score records 295 passing yards, two touchdowns, no interceptions and six rushing yards. His 20.40 FanDuel points were only 0.99 above the saved 19.41. That is a close point forecast, not proof that the model correctly predicted each component.
Receivers: Concepcion’s breakout did not need a touchdown
KC Concepcion supplied an important counterexample to a touchdown-only explanation of Week 7. The official game statistics show eight receptions for 129 yards and three rushes for 23 yards, with no touchdowns.
His 26.20 DraftKings points beat the 6.66 forecast by 19.54. His 22.20 FanDuel points beat 4.98 by 17.22. The saved ceilings, labeled as 90th-percentile outcomes, were 10.64 and 9.37. This result exceeded both without a six-point touchdown contribution.
Receptions, yardage and the receiving bonus created the gap. Receiving opportunity and yardage distributions therefore deserve investigation alongside scoring opportunities. Raising every receiver projection would not follow from this example: the entire receiver sample was overprojected on average, even while Concepcion was substantially underprojected.
The useful question is which players were assigned the wrong expectations, not whether the whole position should receive a blanket increase. A single standout cannot answer that, but the opposite signs of the individual and positional errors make the distinction concrete.
Tight ends: a few scoring plays carried substantial weight
Mason Taylor caught five passes for 62 yards and two touchdowns, adding a two-point conversion, according to the Jets’ official Week 7 notes. His 25.20 DraftKings points exceeded 5.42 projected by 19.78. On FanDuel, 22.70 against 3.80 produced an 18.90-point gap.
His two touchdowns and conversion supplied 14 points. Those scoring events mattered enormously, but the remaining receptions and yards also exceeded the original forecasts. As with Pollard, the point breakdown argues against explaining away the entire miss with one label.
Davis Allen’s performance was even more concentrated in touchdowns. His official game log lists four catches for 20 yards and two scores. The touchdowns supplied 12 of his 18.00 DraftKings points and 16.00 FanDuel points. The forecasts were 1.59 and 1.22, respectively.
Neither result establishes how frequently a similar tight end should score twice. It identifies a distribution question to test across more games: did these low means adequately represent the chance of a meaningful scoring role? Retrospectively rewriting their forecasts would remove the evidence needed to answer it.
Close calls belong in the report, too
Bijan Robinson’s official log against San Francisco shows 63 rushing yards and five catches for 59 yards, without a touchdown or lost fumble. His 17.20 DraftKings points were 0.79 below the 17.99 forecast. His 14.70 FanDuel points were 1.83 below 16.53.
Hunter Henry’s seven catches for 71 yards against Chicago produced 14.10 DraftKings points, 1.15 above the saved 12.95. His 10.60 FanDuel points were 0.66 below 11.26.
These examples balance the largest misses without turning the article into a selection of flattering outcomes. Every eligible matched record remains in the aggregate calculations. Close final scores are useful results, but they are not independent evidence that the underlying workload forecasts were correct.
Position-level results and saved ranges
| Position | DraftKings error (records) | FanDuel error (records) |
|---|---|---|
| QB | 6.0 (31) | 6.1 (31) |
| RB | 5.4 (66) | 5.3 (63) |
| WR | 5.0 (110) | 4.4 (110) |
| TE | 4.8 (62) | 4.2 (62) |
The quarterback samples include backups, not only starters. These are raw-point errors, so a lower figure at a lower-scoring position does not automatically mean better relative accuracy. The article is not a salary-weighted test or a reconstruction of one contest’s player pool.
Among 267 DraftKings records with saved ranges, 113 results landed inside them: 42.32%. Another 101 finished below the floor and 53 above the ceiling. FanDuel had 104 inside out of 264, or 39.39%, with 106 below and 54 above. Two matched records on each platform lacked a saved range.
The stored boundaries are labeled P25 and P90. That interval nominally spans 65 percentage points, not the full range of possible outcomes. The observed coverage merits review, but one partially covered week with correlated player results cannot establish long-run calibration. A ceiling is not a promise that a player cannot exceed it.
Methodology: what this audit includes and excludes
We inspected all five archive pages for both platforms. Each held 409 timestamped pre-kickoff forecasts, including 28 defensive entries excluded from this offensive audit. Of 381 offensive forecasts, DraftKings matched 269 scores and excluded 112 unavailable statistical matches. FanDuel matched 266, excluding 111 unavailable matches and four unresolved identities.
Missing rows were not converted to zero. Explicit zero and negative results remained included. All matched records had complete scoring components under the reader’s checks; none required an exceptional-scoring override for this report. This does not establish that the excluded players scored zero or that the evaluated subset represents every lineup decision.
Results use FSN’s imported statistics and the DraftKings Classic and FanDuel NFL scoring rules checked for this audit. Official team sources corroborate the featured examples. We use the reader’s full-score aggregate metrics, not its provisional core-only figures; we did not independently reconcile every ordinary record against settled contest totals. Later statistical corrections can change results.
The forecasts are saved means, not medians. Internal timestamps are not independent external attestations, and these legacy records do not prove exact public-builder equivalence. Current values on the Week 7 DFS builder were not substituted. This is a retrospective, not a backdated prediction article.
FSN takeaway
Week 7’s strongest lesson is to keep workload, yardage, scoring opportunities and final points distinct. Gibbs had substantial involvement and still missed a high forecast. Pollard’s touchdowns magnified an underestimate that existed even without them. Concepcion broke out without scoring, while Nix’s rushing contribution transformed his result.
Those are specific questions for model evaluation, not proof of a single cause or a justification for changing old forecasts. Keeping both close calls and large misses gives the archive value beyond a weekly highlight reel.
Continue with FSN’s Week 8 report card and Week 9 report card. Their different player samples mean the results should not be treated as a controlled test of model improvement.
