Accuracy Report #1: Weeks 5 - 8, 2026

November 4, 2026

Time again to reflect on the predictability of the 2026 season so far!

If you’re new to these: Examining predictive accuracy has been a long-standing tradition, underpinning a key purpose of Subvertadown: To give us an understanding of how predictable things are— and have been. 

• We want to know "Was this season more or less predictable than normal?"

• And then we want to know if Subvertadown models are “keeping up” with top sources out there— the very best track records. If there’s good comparative accuracy, then we reassure ourselves that the future week projections are valid considerations for planning and strategy.

Compared to Other Seasons

As usual, here’s a look at how each individual model is doing, compared to other seasons.

This part isn’t meant to tell "how good the models are".  It tells us how predictable the current season is, compared to the historical norm.

The results show that, in 2026, most models are performing at a “normal” predictability level. (Not wildly different from other seasons.)

• In particular, QB, RB, and TE have been as “easy” to predict as the historical average.

• Even game scores (and Vegas lines) are close to normal levels of randomness, too. That surprised me once again, because we’ve seen a good share of upsets to Survivor pathways (see section below).

The main exceptions to good predictability are Kicker, WR (which must be related to a number of injuries) and D/ST (which was less predictable than the first month). Again kickers always start out each season looking worse, so we expect their predictability to catch up later in the season.

Comparative Accuracy Assessment

Reminder / for newbies: I don’t expect to be “#1” all the time— that would be plain unrealistic. Especially since I work continuously to find other experts for comparison who have been consistently good for at least a few years. (Whereas many sources look great one year but then poor the next.) My goal is rather to check that the models are still performing at a similar level to these. From year to year, I find we naturally trade places at the top of accuracy. Knowing that the models perform at least at a "similarly" to top sources lends confidence and gives reason to trust forecasting, when we extrapolate models to future weeks.

High-level Summary:

• Exceeded my expectations: Kicker and Tight-End. (Also WR, and also D/ST for ESPN settings.)

• Met my high expectations: D/ST and QB. Also RB.

• Below my expectations: Survivor and Betting Lines.

Here’s the Kicker

Kicker has continued exceptionally for the Subvertadown model, on a comparative basis.

Month 1 had shown a clear distinction, since all other sources had negative values (or near zero).

In the month 2, the other rankers did slightly better than before— but still notably worse than normal.

I’m really pleased that month 2 (taken alone) also saw the Subvertadown model perform the highest:

Regardless of which accuracy metric I could choose, they point to the same conclusion:

• Picking kickers might not be easy”, but the Subvertadown models have stood out: the rankings have not been nearly as bad as others’.

  • For context, the value “0.15”, shown above, is the level that other rankers usually perform, in other seasons.

    • So you might say I’m the only one performing “normally”!

  • A correlation coefficient of 0.15 means (very roughly speaking) you’re going to see 2-3 disappointing kickers among the top 8. Taken alone, you might call that “bad”, even if it’s normal in fantasy football. But compared against other sources, this should instill trust that the model follows predictive patterns in the data that others might not see.

Fingers crossed for the remaining of the season!

Defensive Maneuvers

The second month was tougher at D/ST for all rankers. A couple other sources managed to do better than Subvertadown models, within the last period. This makes 2 years in a row that specifically “weeks 5 - 8” challenge the model. It’s maybe something for me to look at more closely, but I suspect it has to do with the level of trust in emerging (new, revitalized) defensive lines.

On the season, my model slid from #1 down to #3, by the correlation metric.

Here’s the comparison for the industry standard D/ST settings (which I call the “Yahoo” setting):

This is a very normal comparative result: It marks a “good” performance, which should maintain trust in how the models are operating.

Why the slide from #1 spot since the first month?
I don’t see an answer in the model trends, but from inspecting the model results, it was mostly from ranking the Bengals and Chiefs too high in week 8. And the weeks before that, I didn’t rank the Broncos high enough. These calls were enough to describe the slight slip in my standings.

And it’s worth pointing out (not depicted here): The Subvertadown models do perform at #1 accuracy for the other D/ST scoring settings: ESPN and “YDB”.

• Therefore: it’s still the case that you should use Subvertadown rankings if your league uses these other scoring settings.

Two Cents for a Quarterback

Technically, my accuracy slid from #1 to #2 in the standings.
In reality, the top 3 of us are extremely close in accuracy— and that’s true when accuracy’s measured in different ways.

This is all to say: the Subvertadown QB model has performed exceptionally well, too— at least in relative terms.

I’m really happy that accuracy has been at this level for 3 years straight (being among the top 2-3), whereas earlier years the model was in the middle.

The top 3 of us all have smaller rank differentials, of 7.95 (technically mine leads with 7.94, but it’s too tight to have any meaning).

[If you’re interested in the historical averages reported in the website’s new QB histograms (the range of outcomes graphs, which depict similar historical games): Those projections also perform with “okay” accuracy. However, it would still sit at #9 or #10 in the chart above, so that means it’s useful as a tool, but it shouldn’t override the rankings themselves.]

FLEX Position Accuracy (RB / WR / TE)

This is the first year that I launched my RB/WR/TE projections for individual players. I launched the models at the end of week 2.

It’s been going pretty well!

I have mostly been measuring the accuracy by the method of the FantasyPros “Accuracy Gap”, because it’s an accuracy metric that’s more conducive to the large pool of players (and their distribution of projections and results).
This time I have charts for you, and remember a LOWER accuracy gap is better.

Here’s what I can observe in my results (black):

• RB (left): A bad week 5, but noticeably better than the other sources most other weeks, including during weeks 6 - 8.

• WR (middle): Accuracy was surprisingly great during the span of weeks 2 - 6. But my accuracy gap was sadly higher these last couple weeks.

• TE (right): Amazing results. My TE model measured better than the 7 sources I compare against, and very noticeably so during weeks 2 - 7. Only in week 8 did it finally have a worse results than others (mostly by ranking Hunter Henry too high).

I have mostly just been happy to see that I’m not falling behind the average of these good sources. I’m definitely at least “keeping up”, and that was my goal.

On the whole, the only clear “standout” behavior is my own TE rankings— That would be awesome if it could continue like that!

Survivor

Based on probability alone, we should normally expect more than half the pathways to lose twice, during 8 weeks of Survivor. 
That means a normal amount would be 4-6 losses total, across the 3 pathways, from the season start until now (8 weeks).

In fact, we’ve seen 4 total losses out of 12 (3 pathways x 4 weeks) this past month (4 weeks). 
That makes a low win prediction rate of 67%. By the way, that’s the same rate as the 4 losses during month 1.

These results are worse than normal. It implies this season is more chaotic than usual.

Here are losses, with their back-ups indicated:

• Pathway 1: Cowboys lost in week 5 (replaced by Lions), Rams lost week 6 (replaced by Chiefs), Texans lost week 7 (replaced by Vikings)

• Pathway 2: No losses this month.

• Pathway 3: Colts lost week 6 (replaced by Bills, 49ers, Seahawks (replaced Bills loss)

Betting Lines

As always: Our baseline expectations should normally be to lose about -40% of the weekly pot, by this time (after 8 weeks).
So, if you routinely bet $100 per week, then the sum of your 8 weeks of losses would be $40.

That’s because a normal “coin flip” betting process would lose us about -5% per week, from the target weekly bet amount. (If we were just monkeys shooting darts.)

This second month— after an especially lucky month #1— we’ve just experienced an especially dismal period that brings us close to this “low case” scenario.

It’s particularly unlucky, on the month, and it seems to be related to the Survivor failures mentioned above.
I say that because the issue is not about losing “at the margin”. It’s not losing by “just a field goal”. It’s more about wildly deviating results. For example, betting on the Cowboys to cover -8 in week 5, but then they have a blow-out loss in the opposite direction.
This is extremely frustrating, and it makes the final ROI far from where we were last year.

The yields a net weekly ROI of -3%, which is above the -5% “break-even”, but far from the +5% average target.

Here are the recorded bet picks and their W/L outcomes, for each of the recent 4 weeks:

Once again:

High-level Summary:

• Exceeded my expectations: Kicker and Tight-End. (Also WR, and also D/ST for ESPN settings.)

• Met my high expectations: D/ST and QB. Also RB.

• Below my expectations: Survivor and Betting Lines.

Looking forward to the next 4 weeks!

/Subvertadown