re: Percentage chance that you give LSU to win in the Regular Season?

North Carolina 70%
@ Vanderbilt 98%
Miss. State 85%
West Virginia 75%
Tennessee 50%
@ Florida 50%
McNeese St. 99%
@ Auburn 50%
Alabama 50%
UL-Monroe 99%
Ole Miss 70%
@ Arkansas 60%

quote:

---

As each game is an individual event, and a win against North Carolina has no bearing on whether we beat Vanderbilt, this does not apply.

---

This isn't about outcomes, this is about predictions. If you have a possible sequence of events, and there is a 60% chance of the first occurring, an 80% chance of the second occurring, a 70% chance of the third occurring, and a 75% chance of the fourth occurring, then is foolish to predict that all 4 will occur even when the events are unrelated. There is in fact only a 25.2% chance that all four will occur.

Similarly, given a situation where LSU will be playing a series of games where we are more likely than not to win each individual game, that does not necessarily mean the most likely prediction is that we win all of the games.

quote:

---

My argument. If a team has a greater chance than not at winning ONE GAME, then you can write it down that they win that game.

---

So is it your predictive model to assume that any event with a likelihod greater then 50% is always assumed to occur?

(no message)

quote:

---

but it doesn't explain the probability of that outcome, which is only 30%...which obviously means there is a 70% probability that the final record won't be 8-2...

---

Yes, but even at 30%, that is a higher probability than any other individual potential outcome, making it the most likely (which is the point).

quote:

---

A 50% chance of victory can be counted as half a win

---

quote:

---

Huh? A win is a win, no half wins.
With two possible results, you have either 50% or no chance in the case of forfeit or disqualification.
Chance has to do with finite possibilities where Pr(E), with Event being between 0 and 1 (coin = 1/2, 2headed coin =2/2), usually a ratio of occurance to all possibilities, such as lottery tickets (1/1,000,000, a card (1/52), die roll (1/6)etc., not prowess on the field, yds/game or home field adv.
I explained it earlier as a horse in a 9 horse race has a 1 in 9 chance. The likelihood of Win, Place or Show would fall on handicapping, statistics and a whole different ballgame.

---

Holy frick dude, you're as bad as that other guy. You most certainly can add probabilities of victory together to predict a win-loss record. I explained it at a 3rd grade level earlier in the thread, and you still don't get it? Let's see if you can understand it this way. You will roll a regular 6-sided die to determine a win or a loss. Even number you win (2,4,6), odd number you lose (1,3,5). You have to agree that you will win 50% of the time, right? Now roll it 12 times. You might roll 12 evens in a row and go 12-0. Roll another 12 times and you might roll 12 odds and go 0-12. Most likely you will fall somewhere in between. If you could roll 1000 sets of 12 rolls, the most common record will be 6-6. The next most common will be 7-5 and 5-7. It will be a "normal distribution" (look it up).

Now let's change the rules so that you win if you roll 1,2,3 or 4, and lose with a 5 or 6. Would you agree that your probability of winning is now 66.7%? Please please please tell me you understand that. Now roll 12 times again and again. The most common win-loss record this time will be 8-4. Why? Because .667 + .667 + .667 + .667 + .667 + .667 + .667 + .667 + .667 + .667 + .667 + .667 = 8.

This is true no matter how you mix the probabilities, adding them up will tell you what your record will most likely be.

quote:

---

AnActualFan

---

I like your optimism and, though I don't share it, I can hope along with you. This is the part that confused me, though...

quote:

---

Tennessee 50%

@ Florida 50%

---

I'ts hard to see how our chances would be the same at home against Tennessee as they are in the Swamp against Florida. I could maybe see this if the order of the games were reversed and you were expecting a letdown after the huge game at Florida. Playing UT first, though, I have hard time understanding how they would be as likely to beat us as Florida is. Maybe you're expecting us to be looking ahead when we play them? I guess that's possible, but I don't think UT has fallen off to the point where they are an opponent that anyone would overlook. Judging from your comment...

quote:

---

At this point in the season, a loss to Dooley is not unthinkable!

---

...it sounds like part of your assessment here includes a very healthy respect for Dooley. I agree that he's a good coach, but I'm not sure anybody is so good that they could immediately transform Tennessee into a team that is an equal threat at our place to Florida in theirs.

quote:

---

It takes more than 12 rolls to make frequency probability work. That is the flaw.
At least that's what I remember from quite awhile back.

---

What does this have to do with converting a series of win probabilities into a predicted record? There is no flaw in the way Nuts4LSU did the math. That's just the way it is.

I hope you aren't taking chilge1's side in this, because according to him if you give LSU a 51% chance of winning each of 12 games, you just predicted a 12-0 record. This is obviously not the case, as it's pretty apparent that you would be predicting a 6-6 record.

North Carolina 80%
@ Vanderbilt 90%
Miss. State 85%
West Virginia 75%
Tennessee 70%
@ Florida 45%
McNeese St. 98%
@ Auburn 60%
Alabama 50%
UL-Monroe 99%
Ole Miss 65%
@ Arkansas 55%