• You do not need to register if you are not going to pay the yearly fee to post. If you register please click here or log in go to "settings" then "my account" then "User Upgrades" and you can renew.

TeamRankings

BGRed

Starter
15 Year Member
Which is fine but that doesn't explain the parameters that told them that Nebraska would lose 9 games. Where did that number come from?
They currently had 1 loss. Based on the simulations and likelihood for winning or losing each game, they would lose 8 of the remaining games...which ones is not exactly known, but more likely that they'll lose to on the road to Michigan State and/or Michigan than at home against Wisconsin.

This is based on simulations using actual prior game statistics. The probability of winning each game is determined as an outcome of the simulations.

To think of it differently, here is the distribution of outcomes of the simulations. In a small number of simulations, we ended up 28-2. In another small number, we ended up 13-17.

Note that we are now most likely at 22-8 as of today versus 21-9 from Tuesday's simulations. And, also note that even though 22-8 is 'most likely', it is still only a 14.2% occurrence in the simulations.

upload_2018-12-1_9-58-5.png
 

huskernut

Heisman
10 Year Member
I'll throw out there this additional explanation for anyone still wondering how they get to 9 losses (1 loss we already have + 8 more they are forecasting going forward), which might be helpful to anyone reading this who hasn't seen the statistical concept of Expected Value before. Not everyone has but it's pretty useful, so it may be worthwhile to explain. It's not complicated really, so I think most people do or can understand it pretty readily after it's been explained. If anybody wants more info about it or has questions you can post or PM me.

To make an estimate of the overall Expected Value for a forecast of the outcome of a group of events, like to estimate the number of wins and losses in the group of remaining games in a basketball season, you simply add together your estimate of the probabilities of the outcome for each event.

For example, lets say there were only 5 games remaining and this was your estimate of the probability of winning each of those 5 games

Game Prob of Winning
1 90%
2 70%
3 50%
4 50%
5 30%
Total 290%

To convert that total of the probabilities of 290% into number of wins, you just divide by 100%. So 290% divided by 100% = 2.9 wins. So the Expected Value of the group of % forecasts you made for those 5 games is 2.9 wins. Of course you can't win or lose fractions of games, so they will generally round up and predict to win 3 of the games, and lose 2.

That's how they got to losing 8 more games going forward, simply by adding up all those individual game probability predictions. It gave a record of 16-8 for the remaining 24 games. *

The basis for this approach is simply that overall you are likely to win half the games you think you have a 50% chance of winning, win a few you think you have little chance of winning going in, and lose a few you think you have a great chance of winning going in. Overall, if your % forecasts were pretty good going in, you should end up with about the number of wins and losses that calculating the Expected Value predicted.






* If you want to eyeball check it, you can do it in about a minute, simply by taking all those winning %s they estimated, rounding them to the nearest 10%, and then adding them. That would give you only 13.6 wins, not the 16 wins they predict for the remaining 24 games. But you'll notice the list of least and most likely wins only have 20 games total between them. They left 4 games out, probably because these are lists of the 10 hardest and 10 easiest games, a nice round number.

But when you do an Expected Value calculation like this you must have a value for each event in order to be accurate. Since the remaining 4 games are not on either the most or least likely win lists, it's reasonable to assume the likelihood of each of those games is a number between the two lists of predictions. The lowest winning % on the list of easiest games is 76%; the highest winning % on the list of hardest games is 68%. So it would be reasonable to assume the forecast for the 4 games they left out was between 68% and 76%, or, say, about 70%. The Expected Value of wins from those 4 additional games would be 70% times 4 games, or 280%, or 2.8 wins. So you would add those to the total you eyeball added for the two lists, giving a total of 16.4, which rounds down to 16 wins, or a forecast of 16-8.

(Note: Basil liked this post when I had temporarily edited it to say "nm" after I first posted it, but then decided to take it down and fix it. So now that I put it back up, he has the right to say he hates it. I'd be surprised if he or anyone even reads it, so I'm fine either way. :D)
 
Last edited:

BGRed

Starter
15 Year Member
Last night's loss at Minnesota was expected to be a close win. The loss knocks down our current tournament bid odds from 84% to 70%.

upload_2018-12-6_7-20-34.png
 

BGRed

Starter
15 Year Member
This morning's update moved us from a 7 seed (25th overall) to a 6 seed (24th overall). I started a bracket projection process yesterday using real-life bracketing principles and based on these simulations and came up with the folllowing pod for the Huskers.

Kansas City Region.

Hartford, CT Pod.

#6 Nebraska vs. winner of #11 Washington/#11 Utah State
#3 North Carolina vs. #14 Vermont
 

BGRed

Starter
15 Year Member
Using updated projections from the TeamRankings site, I came up with the following potential bracket spot for the Huskers.

Anaheim Region (West)

Salt Lake City Pod:
#1 Gonzaga
#16 Montana

#8 Arizona
#9 Texas

Tulsa Pod:
#4 Houston
#13 Lipscomb

#5 Nebraska
#12 Louisville
 

BGRed

Starter
15 Year Member
Updating a bracket through games of yesterday (still way too early to take this seriously...).

Louisville Region:

Columbia SC Pod:
#1 Duke
#16 TX Southern/N Kentucky

#8 Creighton
#9 Florida

San Jose Pod:
#4 Oklahoma
#13 Murray St

#5 Nebraska
#12 Wofford
 

Rainbow Dash

Embrace yourself.
2 Year Member
Updating a bracket through games of yesterday (still way too early to take this seriously...).

Louisville Region:

Columbia SC Pod:
#1 Duke
#16 TX Southern/N Kentucky

#8 Creighton
#9 Florida

San Jose Pod:
#4 Oklahoma
#13 Murray St

#5 Nebraska
#12 Wofford
That wouldn’t suck. Visit San Francisco.
 
Last edited:

BGRed

Starter
15 Year Member
Last night's game didn't change much in the TeamRanking projections as the metrics put us only slightly over 50% to get the win.

We still project out to be in a close race for spots #2-7 in the B1G standings and a #5 seed in the Big Dance. Getting into the top four of conference would be huge with a double bye.

My wife is an Iowa grad and we watch a lot of their games. They are very overrated and we must get a road win in Iowa City this weekend. Will see how our family schedule shakes out and maybe get into the arena for the game.

upload_2019-1-3_7-33-6.png


Building a new meaningless bracket also didn't change much, just moves us down a couple of notches into the "Anaheim" regional and way east to Hartford instead of Louisville Region/San Jose Pod.

Anaheim Region
Salt Lake City Pod
Gonzaga vs. N Kentucky
Arizona St vs. Texas

Hartford Pod
Virginia Tech vs. Yale
Nebraska vs. Wofford
 

Attachments

Top