The way to compute Expected Value (EV) is to take the product (multiply) of win probability (W%) and opponent NPI Win Points (WP), then add it to the product of lose probability (1-W%) and opponent NPI Lose Points (LP). For example, consider the W% & EV Table above with Win Pts (WP) & Lose Pts (LP) offered directly above that. The T100 has MSOE & Nichols rated nearly equal, why they are 50% to defeat each other, and explains the same 63%-win probability to defeat Lancaster Bible.
These facts make it so either MSOE or Nichols presently has an EV = 58.5 NPI Points if they’d scheduled Lancaster Bible. The calculation determining entries in Column 1 of the EV Table is:(63% Win Probability) * (66.4 LB WP) + (37% Lose Probability) * (45.1 LB LP).
Lancaster Bible’s EV for playing MSOE & Nichols computed the same way becomes 50.2 & 51.7, respectively, as seen above on the first row of the EV Table. Regarding present NPI which suggests Lancaster Bible is already rated significantly stronger, it would have little incentive to play either when they’d be a dog to earn 8.3 & 6.8 less points on average than MSOE & Nichols would earn. Arguably, this was true 10 months previous before the 2025 season began too, though likely wasn’t considered then, either. This directly opposes the competitive ideal for a team to want to seek marginally better competition in order to arrive at the best version of themselves down the road. Seems to be problematic.
Here is a list of all their opponents played and the results of the matches through the month of March. Common opponents are highlighted, and all opponents are sorted by using Massey ranks from the strongest to the weakest:
On Tuesday I offered a slide showing systematic Midwest Bias perpetrated by the NPI. I have zoomed into the section where Lancaster Bible & MSOE can be seen on this plot to take note of where their data points lie and how it relates to the entries found on LMN tables at the top.
First level thought mandates looking at the three teams relative to each other as I have attempted to do. Neither of them is in consideration for at-large bids because none is better than 20th in any ranking realm. It is crystal clear the last at-large will come from those among the first 10 NPI ranked teams. (All rating models are usually in agreement for the top half of these 10 before beginning to slightly diverge right near the “bubble.”) Second level thoughts pertaining to the “LMN Effect” mandate taking a look at how the 3 teams influence the bubble, in particular.
I would argue the bubble presently constitutes teams ranked between 9th & 13th with two weeks remaining in the 2025 season before the final NPI is released. One of those 5 teams will most likely be “The Last One In,” except for something unforeseen happening. (Like multiple conference favorites not winning titles, etc.) The midpoint of the bubble just happens to be St. John Fisher at NPI #11 – The one and only common opponent to all 3 teams of the LMN cohort seen highlighted in yellow a little earlier. How fortuitous is that? LOL
The LMN Effect, as I have now coined the phrase, certainly gets its name from the 3 teams put under the microscope to make the points in this post. Be very careful if thinking it is the exception rather than the rule for teams ranked across the NPI landscape. The LMN Effect permeates the NPI with varying severity across its whole spectrum. Upon first look, that severity increases exponentially moving from the strongest to the weakest teams – Not to be confused with the damage it can produce. Sure, there can be compounding residual effects by the bottom 60+ teams in the NPI, a collective I’d describe as an abomination, but the most confounding and potential to create damage occurs in a sweet spot roughly between 25th and 55th – This is because their positive influence with teams among the best 25 is far greater. i.e. The best 25 teams with wins against the 55th ranked and better often contribute while those against the bottom 60+ typically won’t as their team’s move beyond their 13th victory and the win being unhelpful to its metric. As the win probabilities for top 25 teams shrink competing with others of the same, the influence exerted by their poorer EVs is far more often detrimental. See below:
The #11 curve would be flowing right through the middle of “Bubble Teams Region” on the graph above – Half-way between #9 & #13. How many teams see themselves with worthy BP (Bubble Potential) before next season as they are putting some touches on their 2026 schedules right now? Maybe 25 are in that “boat” with one eye already focused on the 2026 season? Which opponents “serve” these 25 programs best by offering potential to place them in an at-large position should they not win their conference championship game next year? (Certainly, a lot depends on their conference foes and how many of they will be among the better 55 in the landscape.)
The most critical section of the graphs above for any who contemplates such things is where “TEAMS REGION” is wedged in between expected values of 62 & 63-points by playing against those among the 18th to 34th best opponents. Not so surprising that the tail end of that coincides with the inception of win bonuses in the algorithm, either. However, all of this is only valid given two assumptions: [1] You know who’ll be among the 18th to 34th best teams in 2026 and [2] You can count on the NPI getting the best 55 in an accurate order. The first is far more plausible than the second, I am sorry to say. Just look above at the permeating LMN effect described in this post to know it’s true.
Coaches and players are blessed with optimism bias. They have to be because that is what fuels competitive fire and allows both to be at their best on the court. Off it, a “Bubble Want-To-Be” can’t be enamored by 76.5-win points for defeating the 7th best with a 36% chance to get it done, all the while minimizing a 64% chance to walk away with a mere 51.5 points, if they don’t! That’s 60.5 points expected, in the long run a couple points less than competition against the 18th through 34th ranked squads I was writing about just a moment ago. That is a very influential two points as many now understand the “cut-off” is right there! Risk simply outweighs the reward. And when you consider the NPI will hardly ever get #7 too wrong, it will make a habit of reversing #22 with #32 with some predictability. Let the gamesmanship begin! (As I choke down the vomit in the back of my throat writing that last line because it’s supposed to be about the games and not the gamesmanship!)
All this from a process-oriented individual interested in the likely incentive driven endgame of the NPI, as opposed to being a “bottom-line” type. You know the one I am referring. He’s the “Volley-Talker” who was told 3 years of empirical back testing looks pretty good and will want to wait and see which 5 teams make it in the tournament and how the bracket is seeded in 10 days’ time so he can say “It’s all Good!” That is until it’s NOT! When at that time he’ll throw a couple of darts at the “Sound Byte Board” for reasons why it isn’t.
A quote from one of my favorite books comes to mind. “The Black Swan” by Nassim Nicholas Taleb, who wrote:
“Consider a turkey that is fed every day. Every single feeding will firm up the bird’s belief that it is the general rule of life to be fed every day by friendly members of the human race “looking out for its best interests,” as a politician (athletic administrator?) would say. On the afternoon of the Wednesday before Thanksgiving, something unexpected will happen. It will incur a revision of belief.”
Don’t be a TURKEY!
You must be logged in to post a comment.