Not sure how many of y'all remember as it's been a few years, but it's about that time of year when I put together my massive Coastal Division spreadsheet. Heading into tonight's game, there are 14 conference games remaining featuring at least one Coastal Division team, leading to a total of 16,384 possible permutations of the season from this point on. My spreadsheet looks at every one of these permutations and calculates a champion for each one as well as the relative probability of that permutation and spits out totals for each team.
It's no surprise that the Hokies have struggled a bit so far this year, and (spoiler alert) the numbers follow that same narrative. In the table below, the middle column shows how many of those 16,384 permutations each contender would win. The rightmost column shows the relative probability of each contender winning (using the SP+ rankings), which doesn't match the middle column for the sole reason that not every permutation is created equally (for example, 1,024 of them involve Duke winning out). You'll note that VT has much lower odds than the raw numbers would indicate, and that is mostly due to the fact that VT has the lowest SP+ rating of all 5 remaining contenders plus plays 3 of the last 4 weeks on the road.
| Team | Odds | # Outcomes |
|---|---|---|
| Pittsburgh | 5,909 | 55.7% |
| Virginia | 5,632 | 22.0% |
| Miami | 964 | 12.7% |
| Virginia Tech | 2,168 | 4.7% |
| North Carolina | 1,618 | 4.2% |
| Tie | 17 | 0.4% |
| Unknown | 76 | 0.2% |
That "Tie" line above is the same as in prior iterations of this report. Those are all 3-way ties that cannot be resolved until Step #7 in the ACC Football Tiebreaker Policy: "The tied team with the higher ranking by the Team Rating Score metric provided by SportSource Analytics following the conclusion of regular season games." Since there's no way to know who would be ranked first by "SportSource Analytics" unless and until it happens, I just have them listed under "Tie".
Something new from 2019 is that "Unknown" line. At one point recently, one individual pointed out that they interpreted Step #3 in the tiebreakers differently than I did. We had a back-and-forth discussion for a bit, and while they made a good point, I (a) didn't like their attitude, and (b) still think that my original interpretation has merit. So, in the couple years that have passed since then, I have reached out to a number of different people who might have connections to someone who could authoritatively state how that step would be applied, but none of them have responded with anything definitive, so I have decided that any outcome that changes based on the interpretation of that rule should be marked as "Unknown", and if/when it gets closer to being relevant we might get a more definitive answer.
Anyway, here are some charts:
I don't know about y'all, but I'm looking forward to watching those blue wedges shrink over the next few weeks.
Comments
Excellent work again.
I would like to request more information about the two sides of the unknown option.
In the ACC Football Tiebreaker Policy, Step #3 on both the 2 team and 3+ team lists reads (with emphasis added by me):
The bolded sentence is the part that is seemingly ambiguous, with two fairly distinct (and potentially legitimate) interpretations that I have heard so far.
Interpretation #1: rephrase as if to say "if there are multiple ties within the division, apply all prior steps to all ties in order from first to last before applying this step." The best argument against using this interpretation is that it would make more sense for them to place a note at the start, saying something like "apply each step to all ties in order from first to last before advancing to the next step".
Interpretation #2: rephrase as if to say "if there are multiple ties within the division, attempt to break a tie for first before breaking any other tie below it". This means that if there are, for example, 2 teams tied for 4th behind the tie for first, you would apply this step to the pair of teams tied for 4th instead of any single team. The best argument against using this interpretation is that it could more frequently result in an unresolvable tie that is only broken by the mythical "Team Rating Score metric provided by SportSource Analytics" in Step #7.
... In any case, I have probably spent entirely too much time dwelling on what might happen in a rare circumstance that (a) has never even come close to happening, and (b) is not particularly likely to happen at any point in the next 50 years, even if we are still using the same tiebreaker procedure that far in the future.
It seems logical to me that if 3 teams are tied for first, and 3 teams are tied for fourth that the complete set of tiebreaker rules would be applied to the 1st-place-tie, resolving who is 1st, 2nd and 3rd, and then the tiebreaker rules would be applied to the 4th-place tie, resolving who is 4th, 5th and 6th. Just moving down the chain.
Not sure how that aligns with your interpretation #1 or #2...
this is also how I interpreted it
Essentially, interpretation #1 boils down to this:
- Apply step 1 to tie for first
- Apply step 1 to tie for fourth
- Apply step 2 to tie for first
- Apply step 2 to tie for fourth
- etc
Interpretation 2 would then be:
- Apply step 1 to tie for first
- Apply step 2 to tie for first
- Apply step 3 to tie for first (looking at the 3 teams tied for 4th as one whole group)
- Apply step 1 to tie for fourth
- Apply step 2 to tie for fourth
- etc
My initial interpretation was #2, and it is still my preferred interpretation, but since I see some merit in #1 (and no grammatical reason why that absolutely can't be the intended interpretation) I've reached out to a number of people for further information, but I'm afraid that we won't actually get a definitive answer until such a scenario that would result in a split champion becomes a likely (or at least plausible) result.
What did our odds go to after that shot show tonight?
Fell from 4.7% to 3.3%. Pending the upcoming downward adjustment of our SP+ rating, that is.
You need to grab negative plaid for that chart. I am far enough into post game I keep thinking the blue is Duke. I don't think it will matter what color you use for the good guys for a while now, but FUCK LOLUVA.
Late Friday night update:
The biggest surprise after putting in tonight's result is that apparently the vast majority (74 of 76) of the "unknown" permutations depended on us beating BC. There are now only 2. Both scenarios include a 3-way tie between Pitt, Miami, and UVA at 5-3, and require a specific result in each of the 13 remaining ACC games involving Coastal teams besides BC @ GT, so it's pretty hilariously unlikely at this point.
As far as VT goes, we dropped from 2,168 to only 616 permutations as the Coastal champ, and now have only a 3.3% chance of pulling it off. All other teams' odds improved, but by less than 1% each.
So now we can get back to rooting against UVa, right?
Now that the new SP+ ratings are out, and I've had a chance to plug in the numbers, I figured it's time for an update. Predictably, with the loss to BC coupled with Pitt's win over Duke, our chances of winning the Coastal have dropped significantly but we still aren't mathematically eliminated yet. The only other significant changes to the chart are that Miami rose from 12% to 17% with their win over GT, UNC fell from 4.2% to 2.6% despite winning their non-conference game against Wake Forest, and all of the potentially ambiguous finishes are now off the table entirely.
At this point, the possible tiebreaker situations we could win are:
- a 2-way tie with UVA
- a 3-way tie with Pitt and any of Miami, UNC, or UVA
- a 3-way tie with UVA and either of Miami or UNC
- a 4-way tie with UNC and any two out of Pitt, Miami, and UVA
- a 5-way tie with Pitt, Miami, UNC, and UVA
And this is the easiest way to describe the current list of scenarios that VT could win:
- VT needs to win out
- There is no single other game where VT needs a specific outcome
- That said, VT still needs a lot of help from other teams, but mostly Pitt
The most interesting thing to point out at this point is that if VT wins out, the Coastal race is down to VT and Pitt, and that VT wins every possible tiebreaker combination including Pitt and at least one other Coastal team.
And as for rooting interests this coming week, there are certain results that make it more likely for VT to win the Coastal, such as UNC beating Pitt and Miami beating FSU, but if VT beats Duke we are guaranteed to remain in contention for at least another week, regardless of what other combinations of results happen.
I couldn't believe how many scenarios go our way when I posted them in the Wheel of Destiny. I was also surprised that there was only one tie combination that was impossible due to games remaining (VT/UNC/Miami). Although I didn't bother to figure out if any of the individual 2-way ties were possible or not.
What's boosting us right now is that we still have more divisional games in front of us than behind us.
Yeah, it is a bit strange that we would win literally any tie other than a 2-way tie with Pitt, but that comes with the territory of having lost to both crossover opponents and otherwise only to Pitt, plus Pitt having already beaten one of their crossover opponents.
The path forward is becoming somewhat clear: win out, hope that Pitt collapses down the stretch, and hope one other Coastal contender wins out except for our game (if applicable). Not likely, but there's still a chance.
I'm so depressed about the state of Hokie football that I can't even chime in with the traditional "so you're telling me there's a chance" dot gif
Yeah, I feel you on that. I probably wouldn't have bothered this year, but the numbers are kinda my thing, and the process helps to soothe the pain somewhat.
Atlantic division summary coming up later this week, though.
Funny how the kid in that gif is wearing a sweater that looks like a UNC end zone.
Ummm, sorry to Debbie Down, but I am pretty sure an L to Duke is coming given the plethora of worrisome circumstances revolving around the program and available roster.
I don't think you have to be worried about being a Debbie Downer.
We already have an Amy Apathetic, Melissa Melancholy, and Bridget Bummer.
Worst Teen Girl Squad Ever
Fuente has managed to lose to teams that are not as good as Duke; so anything is possible.
Thursday night update: Pitt's win still doesn't eliminate the Hokies, but UNC was the most likely opportunity for a Pitt loss (and we need 2 to get back into contention), so our odds have now dropped to only 0.79%, and not only would we need to win out but we would need Pitt to lose out.
In short..... our chances are much better next year. Or in basketball. Or probably close to every other sport in the coming year.
Pitt's win effectively ended anybody else's chances of representing the Coastal. And to be honest, they have looked the best of the teams in our division and deserve to be the CC.
Ugh, now I need to go wash my hands they feel dirty from typing that.
They didn't end anything except UNC's chances of winning the division. UVA still controls their own destiny, Miami just needs them and UVA to each lose one, and we never really had a decent chance to begin with.
I was about to say "it really sucks that the 3 most hated Coastal programs are the ones still in contention", but then I realized that UNC, Duke, and GT are the other 3 and that I hate them all pretty much equally except UVA. Any time anyone else but us wins the Coastal it sucks.