(08-11-2015 06:23 PM)talon owl Wrote:  

(08-10-2015 11:52 PM)baker-13 Wrote:  As with last year, Ken Massey has predicted results for our schedule this year.

I took the liberty of throwing these percentages into a trusty Excel spreadsheet (based on one JOwl built for this last year) to see what happened. It's attached for anyone who wants to play along.

Quick summary of probabilities of given win totals:
0 wins: 0.0000% (taking this out to ten decimals still yields zero's)
1 win: 0.0002%
2: 0.0061%
3: 0.0994%
4: 0.8350%
5: 4.0858%
6: 12.3686%
7: 23.7070%
8: 28.5493%
9: 20.6881%
10: 8.1672%
11: 1.4196%
12: 0.0738%

There are two scenarios tied for the most likely individual set of outcomes: 9-3 (losses to Texas, Baylor, and LaTech), and 10-2 (flip the LaTech result).

I plan on keeping this updated as the season goes along. If nothing else, it's something to amuse myself with as time goes on.

I don't see the spreadsheet attachment. ?

Doesn't look like the attachment worked. Revising now.

ETA: Quasi-fixed. Had to shift to Google doc instead. Upside is, no need to just take my word for it on updates.

(08-11-2015 12:52 PM)Antarius Wrote:  

(08-11-2015 11:27 AM)OptimisticOwl Wrote:  

(08-11-2015 11:22 AM)owl95 Wrote:  OO I want to win as well. 1 in 5 chance to win is not bad. It depends on how the Notre Dame game goes. If UT gets beat up, that could be good for us, sometimes losing badly is contagious. I feel like that is part of what happened vs ODU last year. I think might tune into the UT-Notre Dame game a bit to see how they look.

Yeah. I was referring to the people who will be satisfied with a good showing. That would be better than nothing, but as a goal it stinks.

I don't think that's the goal anyone has. Its just that if we HAVE to lose, please let it be in such a way that we are competitive.

Actually, I have seen a good showing in a loss to Baylor and/or Texas listed as a goal. Not recently, but it would not surprise me to see it again in the next month.

Usually couched (coached?) as something like " I want us to win 8 games, challenge for the division (conference) championship, and show well in the games again Baylor and UT. take them well into the fourth quarter, give 'em a scare."

maybe not a goal, per se. maybe more a statement of minimum acceptable.

(08-11-2015 01:41 PM)OwlAndSparrow Wrote:  

(08-11-2015 09:28 AM)OWLmanz Wrote:  WHAT?? I hear what you're saying, but I find it hard to believe.............. I'm probably too "simplisticly rational" I guess! Also, if we are truly favored in 9 games, why isn't that the highest probability result????????????

(08-11-2015 08:40 AM)baker-13 Wrote:  

(08-11-2015 07:46 AM)OWLmanz Wrote:   OK I am "just" an "academ," but how can 9 and 10 wins be the most likely scenarios, since the data table appears to show 8 wins as the highest probability outcome, with 7 wins as the second highest probability????? Makes no sense.....

Of course, we learned even in Econ and Finance, that you can interpret statistics any way that you want...........

"Most likely scenario" = most specific exact combination of things. You get higher probabilities for 7 and 8 wins because there are more different combinations that lead to 7 or 8 wins than there are 9 or 10 wins, even if those individual combinations have a lower probability that the most likely 9- and 10-win combinations.

Even if we were favored in all 12 games, 12 wins wouldn't be the most likely result.

Let's call the games A, B, C, D, E, F, G, H, I, J, K, and L. We'll call the probability of winning each W(X) for game X. So, if A is against Wagner, then W(A) is close to 100%, so probably something like .99. For each game, the probability of losing, L(X), is 1 - W(X), so L(A) is about .01.

Since each game is DIFFERENT, then the probability of having a particular season is the product of the probabilities for each game. Kinda like the probability of flipping a coin heads and then tails is .5 * .5 = .25. Now, the probability of getting one heads and one tails is .5, since you get HT 25% of the time and TH 25% of the time.

So, the probability of winning all 12 games is:

W(A)*W(B)*W©*W(D)*W(E)*W(F)*W(G)*W(H)*W(I)*W(J)*W(K)*W(L)

The probability of getting 11 wins is not so easy to write, because there are 12 ways it could happen, with some much more likely than others (e.g., it's EXTREMELY unlikely we'll beat Texas and Baylor but lose to Wagner).

So, we'd take L(A)*W(B)*W©* ... * W(K)*W(L), the first way to win 11 games, then add

W(A)*L(B)*W©* .. *W(K)*W(L),

then add

W(A)*W(B)*L©*W(D)* ... *W(K)*W(L),

and so on. Add those all up to get the probability of winning 11 games.

For 10, 9, 8, and so on, there are a lot more terms that need to be added up.


In the end, games B, D, and H are the hardest games. W(H) is about .5, it sounds like. W(B) is about .20, and W(D) is something small. These are the ones where L(B) and L(D) are higher than W(B) and W(D), and W(H) is about the same as L(H).

Thus,

W(A)*L(D)*W©*L(D)*W(E)*W(F)*W(G)*L(H)*W(I)*W(J)*W(K)*W(L)

and

W(A)*L(D)*W©*L(D)*W(E)*W(F)*W(G)*W(H)*W(I)*W(J)*W(K)*W(L)

are the most likely seasons.


Now, looking at the other nine games, we're favored in all of them, BUT that doesn't mean it's likely we'll actually win all of them. It's like, roll a die nine times. Count how many times you roll higher than a 2. Each individual time you roll, you're probably getting higher than a 2, but the chances of ALL NINE rolls being higher than 2 are very small (about .00005).

So, the season

W(A)*L(D)*L©*L(D)*W(E)*W(F)*W(G)*L(H)*W(I)*W(J)*W(K)*W(L)

is less likely than the one where we just lose to Texas and Baylor, but if we take into account ALL of the other nine games (and LaTech is a coin flip), we're more likely to lose one or two of them than we are to win all of them.

THIS is what I love about Rice, our alums, and the Parliament.

However, I am a simple person, and don't care about the probabilities for the season. I want us to go into each game, one at a time, win all the ones we are supposed to win(favored in), and steal as many of the others as we can.

hasn't it been said here many times that we need to win the ones we are supposed to win? Aren't we supposed to win the ones we are favored in? Am I missing something here?

One at a time.

FWIW: Massey recently re-updated some of his percentages...we went from a 68% to a 69% chance to beat UTSA, with the downside of going from a coin-flip against LaTech to a 49% chance. Mostly likely result is still 8 wins, but the most probable scenario is now the 9-3 scenario mentioned in the original post.

This will likely change some once he re-updates to take today's results into account (NDSU losing is a bit of a big deal in the 1-AA world)...I plan to update again Saturday morning.

How accurate have Massey's preseason ratings been in the past, from a predictive standpoint?

The spreadsheet up top is updated after the results of yesterday's games. For convenience, all scenarios involving a loss to Wagner are now filtered out, so you only have 2048 possible scenarios to look through.

Highlights:

• We now have a 0.1330% chance of winning out.
• Our chances of losing out are 0.0000321%
• Our most probable regular-season win total remains at 8 (28.8676%), followed by 9-3 (23.7524%) and then 7-5 (21.1696%).
• Our most probable single scenario is 10-2, losing to Texas and Baylor (3.826%).
• We have a 96.6216% chance of bowl eligibility.