Revisiting Public Projection Models' Expectations for the 2025-26 Sabres...Already

This could’ve been written after One Week, and it almost was, but then they won a couple games so I gave it time to breathe. But now, yeah we’re going to have to get this thing published before the end of the road trip.

The italicized parts were what I had written a month ago.

I wonder what the number nerds think about their #Sabres predictions at this point.

Need to stuff their spreadsheets and heat maps into a sock right about now.
— Mike Harrington (@ByMHarrington) October 12, 2025

Real Analytics⬇️ https://t.co/s0sb2J6zHv
— After The Whistle (@AfterLeWhistle) October 13, 2025

And before you get upset at me for referencing some of our biggest friends to open up an article on an analytics site — can you blame them?!

Frankly, I don’t feel any need to try and defend another person’s projection model because nothing about the first three games deserves any defending of the world on paper. I’ll even provide the socks (I promise they’re clean).

I’m not even going to correct Andrew and/or Craig by saying that goal scoring rates and goal differentials are the foundation of analytics and don’t live in their own other world, and that, yes, you don’t need a graph to tell you how bad 0 for 11 is on the powerplay (Andrew if you’re reading this thanks for signing an autograph for me the week after the Ottawa brawl that was a good time). Nor am I going to mention that a heat map is simply a display of frequency of datapoints in a plane to provide the opportunity to view trends visually.

But how is this happening?

Remember your old friend the standard deviation

While the histograms and density plots are pretty, it’s always important to remember to consider the standard deviation in alongside the means since that is also part of the story that comes with the results from the millions of replications of the simulations.

With the normal distribution, about 68% of the results fall within one standard deviation of the mean, about 95% within two standard deviations, and about 99.7% within 3.

POV: You're getting to the bottom of this pic.twitter.com/OVlLoAuxZ4
— xBuffalo (@ExpectedBuffalo) October 15, 2025

Percentile	Micah’s Model (Mean = 92)	Twins’ Model (Mean = 98.8)
1	72.2	79.5
2.5	75.3	82.5
16	83.5	90.5
25	86.3	93.2
75	97.7	104.4
84	100.5	107.1
97.5	108.7	115.1
99	111.7	118.1

So 68% of the simulations, just over 2 of every 3, fall between 83 and 100 points in Micah’s model, even with the average result being 92.

16 of every 100 fall below 90 in the Twins model, which is about 1 in 6. Remember your probability classes. Rolling your favorite number on a natural die is just as theoretically likely. Not too likely, but we’ve all played Yahtzee.

It’s clearer now than it was a month ago, but we’re never going to see the aspired opening night lineup this year. Of course that itself does open up the outlier conversation because a lot of things they couldn’t control did happen, but I don’t think anyone is satisfied with ending the conversation there.

All three of us on this website have said that we don’t feel they were built to account for injury and non-linear growth uncertainty, and if the other leaders on the team are unable to perform at a higher gear with the captain away attending to crucial family matters, projections can be incinerated for where things are going to go next for the Sabres.

But there is one more piece that may have gone into that projection, and it is kind of coming to fruition.

With the way that the Atlantic is jammed, even with lower current projections than a month ago, the traffic jam in the Atlantic is matching what those preseason expectations were suggesting: Toronto and Florida taking a step back could move the points around the rest of the division even with the rest of the division staying relatively consistent with their performances from a season prior.