Key finding
Liz Loza’s ranking product led the 2026 Week 3 half-PPR DEF comparison with a skill score of 20.0, followed by Draft Sharks at 10.6. The displayed field contains 14 eligible tracked ranking products, with a 41.2-point spread from highest to lowest. Behind the leaders, tied positive scores give way to results below the random-order reference. That shape makes this more than a story about the winner: the same weekly score can sit above one product’s through-week record and below another’s. These results describe named ranking products in this weekly sample, not overall services. [1][2]
What the scores and publication status mean
Skill score measures how much better than a random order a ranking matched how players actually scored. A score of 0 is random and 100 is perfect; most rankers land between about 5 and 35 because weekly fantasy scoring is noisy. It is not a prediction success rate. That distinction matters when reading Loza’s leading 20.0: the number measures the ordering’s match with actual scoring, rather than the share of predictions that succeeded. Negative results elsewhere in this field fall below the scale’s random-order reference. [2][1]
None of the 14 displayed products has a provisional result. Provisional status flags a result as still subject to the publication and correction process; this cohort’s results are published rather than provisional. Published does not mean immune to revision, however. The comparison uses the weekly lock policy and permitted published observations, and corrections or permission changes can alter the available evidence. Readers can therefore distinguish the current published ordering from a pending result without treating the publication status as a guarantee that the record can never change. [1][2]
The leaders also exceeded their own records
Loza finished 9.4 skill points ahead of the next distinct displayed product, Draft Sharks. Her lead establishes the ordering within this cohort, but her own through-week record adds another perspective. Loza’s average through Week 3 is 0.8 across 3 eligible weeks out of 3 published weeks, with this week’s result 19.2 points above that average. The leading performance was therefore above her observed record through the same week, rather than simply matching it. The average includes Week 3; it is not a separate pre-week baseline. [1]
Draft Sharks presents a similar contrast between weekly placement and accumulated results. Its 10.6 score is 17.3 points above its through-week average of -6.7, based on 3 eligible weeks out of 3 published weeks. The leading products thus both posted positive weekly scores while their through-week averages sat nearer to, or below, the random-order reference. This helps explain how to read the top of the board: the weekly leaders were above their own observed averages, but those comparisons do not establish that either result will recur. [1][2]
Weekly ties conceal different through-week records
Daniel Dopp and FantasyPros Consensus each scored 7.1, giving them the same weekly result behind Draft Sharks. Their records through Week 3 differ: Dopp averages -2.0, with this result 9.0 points above that average, while FantasyPros Consensus averages -3.9, with its result 11.0 points above. Both averages cover 3 eligible weeks out of 3 published weeks. The tie answers which product scored better this week—neither did—but it does not make their accumulated records interchangeable. Weekly placement and performance relative to a product’s own average answer different questions. [1]
Matt Bowen follows at 5.9, which is 11.4 points above his -5.5 through-week average. Field Yates, Eric Moody and Tristan H. Cockcroft then share a weekly score of 2.4. Their respective averages are -2.4, -4.3 and -5.1, with stated above-average comparisons of 4.7, 6.7 and 7.5 points. Each average covers 3 eligible weeks out of 3 published weeks. This part of the field reinforces the distinction: identical weekly scores need not reflect identical records, and a positive result can coexist with a negative through-week average. [1]
Below zero does not always mean below average
Eric Karabell, FFToday and Mike Clay illustrate why the random-order reference and a product’s own average should remain separate. Their weekly scores were -4.7, -7.1 and -9.4 respectively, all below zero. Yet Karabell finished 3.5 points above his -8.2 average, FFToday finished 9.8 above its -16.9 average, and Clay finished 6.3 above his -15.7 average. All cover 3 eligible weeks out of 3 published weeks. These results were below the random-order reference while still exceeding the products’ own observed averages; neither comparison cancels the other. [2][1]
At the lowest displayed score, ESPN app projections and Footballguys (PPR board) tied at -21.2. ESPN app projections was 2.7 points below its -18.4 average, while Footballguys was 0.8 above its -22.0 average; both averages cover 3 eligible weeks out of 3 published weeks. FF Toolbox (PPR board) finished above that tie at -17.6, or 4.7 points above its -22.4 average. Its average has a different participation scope: 2 eligible weeks out of 3 published weeks. That shorter eligible record should stay visible alongside the comparison. [1]
What fantasy managers can take from the spread
The 41.2-point highest-to-lowest spread describes the extent of this week’s field, while the 9.4-point leader gap describes the separation at the top. Neither gap establishes statistical significance or predicts another week’s ordering. The through-week comparisons add useful context without supplying that missing certainty: Loza and Draft Sharks exceeded their own averages, while the lowest weekly tie contained results on opposite sides of their respective averages. The report therefore supports comparisons within the weekly field and within each product’s record, not an explanation of what caused those differences. [1][2]
For a fantasy manager, this sample identifies which tracked DEF ranking products best matched actual scoring in the stated week and format, and whether their results exceeded their own through-week averages. It cannot establish season-long or future superiority; the seasonal participation gate is separate from this weekly view. Nor should a named ranking product’s result become an overall rating of its associated service. The useful conclusion is specific: Loza led this published weekly comparison. Whether that ordering persists remains unanswered by this sample. [1][2]
Sources
- Fan50 evidence record · Checked October 29, 2026
- Fan50 methodology · Checked November 5, 2026