|
Replies: 33
| visibility 360
|
Webmaster [∞]
TigerPulse: 100%
∞
Posts: 46743
Joined: 2012
|
FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 11:11 AM
|
|
|
|
|
|
 |
Rival Killer [3072]
TigerPulse: 100%
33
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 11:13 AM
|
|
I wonder what the same Computer predicted in the the Ohio St. game??
|
|
|
|
|
|
Standout [313]
TigerPulse: 91%
15
|
|
|
|
|
|
CU Guru [1222]
TigerPulse: 100%
27
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 11:35 AM
|
|
So BLUF......I like how their prediction now looks.
|
|
|
|
|
|
Paw Master [17038]
TigerPulse: 100%
51
Posts: 21917
Joined: 2007
|
|
|
|
|
|
All-American [555]
TigerPulse: 100%
20
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 12:27 PM
[ in reply to Re: FB Update: Computer predicts winner of Clemson-Alabama ] |
|
Okay, since the computer overestimated Ohio State by 35, and underestimated us by 3, let's apply that to the Alabama simulation:
Computer simulated results: Alabama 34, Clemson 27
Adjusted results: Alabama -1, Clemson 30
I'd settle for that.
Okay, okay. Alabama can't have a negative score, so we can adjust both up one point.
31-0 Clemson. Hmmm. Sounds right to me.
|
|
|
|
|
|
Tiger Cub [11]
TigerPulse: 100%
2
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 2:12 PM
|
|
I think I just heard a mic hit the floor...
|
|
|
|
|
|
Standout [201]
TigerPulse: 79%
13
|
|
|
|
|
|
Rival Killer [3078]
TigerPulse: 100%
33
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 11:28 AM
|
|
If Gallman has 90 yards and DW4 has no picks, I don't see us losing
|
|
|
|
|
|
Orange Blooded [2841]
TigerPulse: 100%
33
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 11:29 AM
|
|
If Gallman rushes for 90 yards I don't see how we lose the game. Alabama doesn't give up much on the ground, so 90 yards on 25 carries would not be the greatest game ever, it would take a good bit of pressure off of Watson.
|
|
|
|
|
|
110%er [3821]
TigerPulse: 82%
35
|
if the OLine doesn't step up some, we don't score 30...
Jan 6, 2017, 11:29 AM
|
|
whatever the skill level.
|
|
|
|
|
|
Ultimate Clemson Legend [102961]
TigerPulse: 100%
64
Posts: 98975
Joined: 2009
|
Re: if the OLine doesn't step up some, we don't score 30...
Jan 6, 2017, 11:42 AM
|
|
Our QB, WRs and RB are going to have to help the linemen out. You can't expect a line to hold firm on every play. The key for the QB is early reads and lightning fast decisions. All others are just expected to execute their typical duties.
We don't need a monster offense to win but I would be delighted if they came out like a machine.
|
|
|
|
|
|
Paw Warrior [4697]
TigerPulse: 100%
37
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 11:32 AM
|
|
Didn't the computer let Ohio State score 28 points!?
|
|
|
|
|
|
Ring of Honor [22921]
TigerPulse: 100%
53
Posts: 15724
Joined: 2012
|
gallman only rushed for 90 yards in 501 simulations????
Jan 6, 2017, 11:36 AM
|
|
That's only .1796 yards per game....
I guess we are facing the best defense ever.
|
|
|
|
|
|
Clemson Icon [24212]
TigerPulse: 100%
54
Posts: 24535
Joined: 2003
|
There are too many variables in the game of football, and
Jan 6, 2017, 12:23 PM
|
|
not enough game time for them all to be relevant in an individual game. This means a computer simulation can't reasonably predict an outcome of a game to the level of specificity demonstrated in this article.
Maybe if you stretch it out to predict season trends... but not a single game.
There are variables that the programmer can't possibly know to include, and there is too much virtual randomness in a single football game to trust a computer.
One of the anomalous simulations is just as likely to occur as the median or the mean.
The computer simulation relies on human input, thus making the simulations subjective. People forget that.
Scientism sucks.
|
|
|
|
|
|
Orange Phenom [14610]
TigerPulse: 100%
49
Posts: 23679
Joined: 2004
|
You are reading a lot into this. The takeaway here is that
Jan 6, 2017, 12:31 PM
|
|
we won 42% of the time. That's actually pretty good...
|
|
|
|
|
|
Clemson Icon [24212]
TigerPulse: 100%
54
Posts: 24535
Joined: 2003
|
You missed the point of my post. "We" didn't do anything.
Jan 6, 2017, 12:36 PM
|
|
The simulation did something on its own that has very little to do with how "we" do. A group of people judged us, used their perception to input data into a program... and let that play out.
They didn't upload the team.
|
|
|
|
|
|
Orange Phenom [14610]
TigerPulse: 100%
49
Posts: 23679
Joined: 2004
|
Assumedly, they put real data in there. They have control
Jan 6, 2017, 12:43 PM
|
|
over how much various things get emphasized, but it's not like they have a line of code that says "SEC SPEED = +3 pts."
You can choose to take away whatever you want from these things, but if you were to tell me right now we had a 42% chance of winning on Monday i would take it. You aren't going to get much better than something close to a coin flip against Bama.
|
|
|
|
|
|
Clemson Icon [24212]
TigerPulse: 100%
54
Posts: 24535
Joined: 2003
|
They don't have access to "real" data. We can't quantify
Jan 6, 2017, 12:48 PM
|
|
much of the variables in college football. We can't upload Deshaun Watson's actual skills or performance. We can only program based on metrics that we think matter more than others. Then humans tune those metrics based on incomplete understanding of how things work.
Certain unknown things can effect players on certain days that would make obsolete their career stats.
This is why "shocking" games happen literally every week. People think they have a solid grasp on the data, and then reality often washes it away.
Computer simulations are more about entertainment than predictions. Thinking otherwise is the result of scientism. Like labeling high school athletes with "stars." It's an illusion.
|
|
|
|
|
|
Orange Phenom [14610]
TigerPulse: 100%
49
Posts: 23679
Joined: 2004
|
Shocking results are perfectly normal. They are just the
Jan 6, 2017, 1:10 PM
|
|
more improbable results at the ends of the bell curve.
|
|
|
|
|
|
Clemson Icon [24212]
TigerPulse: 100%
54
Posts: 24535
Joined: 2003
|
How can they be called improbable? They were always 100%
Jan 6, 2017, 1:12 PM
|
|
likely to happen. It's not a game of chance. It's a game of predicting what is certain to happen.
We can't verify that anything that has ever actually happened was itself "improbable." Only that it was not expected by most.
|
|
|
|
|
|
Orange Phenom [14610]
TigerPulse: 100%
49
Posts: 23679
Joined: 2004
|
Explain the NBA and MLB playoffs then.
Jan 6, 2017, 1:16 PM
|
|
4-3 series happen all the time and there are often blowouts within those results.
|
|
|
|
|
|
Clemson Icon [24212]
TigerPulse: 100%
54
Posts: 24535
Joined: 2003
|
I don't understand how that is countering anything I said.***
Jan 6, 2017, 1:16 PM
|
|
|
|
|
|
|
|
Orange Phenom [14610]
TigerPulse: 100%
49
Posts: 23679
Joined: 2004
|
|
|
|
|
|
Clemson Icon [24212]
TigerPulse: 100%
54
Posts: 24535
Joined: 2003
|
That kind of thinking is putting the cart before the horse
Jan 6, 2017, 1:37 PM
|
|
Whatever is going to happen, is 100% likely to happen.
The predictions are not based on 100% perfect knowledge. They are based on a partial amount of data being processed subjectively. There is no 33% chance we win. That is not a real thing. It's simply not logical to think of predictions that way. There is a 33% we win, according to some equation that does not fully know anything.
|
|
|
|
|
|
Orange Phenom [14610]
TigerPulse: 100%
49
Posts: 23679
Joined: 2004
|
I'm not claiming this (or any model) is perfect.
Jan 6, 2017, 2:04 PM
|
|
It's just about getting an idea of what could happen. Like last week we were 3 pt dogs in Vegas but the consensus was that Clemson was at least even with OSU, if not better. That obviously played out on the field and to a largely degree that just about anyone could have expected. I don't think we are 31 points better than OSU but we were that night and i do feel like we are just a better team overall.
I'm kind of getting a similar sense this week where we are 6-6.5 pt dogs to Bama but this game is actually several points closer than that. I think we might have a better shot at beating them this year than last year. I think our D is peaking late this year whereas we saw clear regression on that side of the ball late in the year last season.
|
|
|
|
|
|
Tiger Cub [11]
TigerPulse: 100%
2
|
Re: There are too many variables in the game of football, and
Jan 6, 2017, 2:25 PM
[ in reply to There are too many variables in the game of football, and ] |
|
Models are imperfect, and they are programmed by perception of capability not measured capability. What this really is reflective of is the reliance on and trust of crowd sourced info. The impression by most is that Alabama or tOSU is the superior team. So you ask a large group of people about expected results and they end up saying Alabama or tOSU is the favorite. We trust it because it looks like a bunch of independent thinkers say the same thing...but the thing is that they all get their info from the same place. So it's really just one opinion often repeated. "Alabama is bigger, stronger, faster" "tOSU is unbeatable". We don't have a way to weigh the opinions of some as better than others. You saw it with Reece Davis after the Fiesta bowl. "we all had the impression that tOSU was the better team, but as soon as they started playing it was clear they were in trouble, that 'that guy can't handle that guy'". You see it in pre-season rankings. Also, Clemson hasn't really done anything to change the perception this year, squeaking by every moderately talented team so, assuming they will play to a different standard in this game, the line looks very good, the models look very good. This team IS different than the one that played NC State, though.
In other words, use it to your advantage and bet the line.
|
|
|
|
|
|
110%er [7719]
TigerPulse: 100%
42
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 12:43 PM
|
|
So with 57 seconds left at mid field and 1 time out we are going to run the ball....
0:57 2/5 41 27 - 34 140 Gallman is caught behind the line for a loss of 3 yards.
Yea.... This simulation is garbage.
|
|
|
|
|
|
Rival Killer [2610]
TigerPulse: 90%
33
|
And so how did our fair computer
Jan 6, 2017, 1:33 PM
|
|
do when simulating the Clemson OSU game? Or did it go into hiding after that?
|
|
|
|
|
|
TigerNet Immortal [174069]
TigerPulse: 100%
69
Posts: 29179
Joined: 2012
|
Does the computer work for Espn?***
Jan 6, 2017, 1:47 PM
|
|
|
|
|
|
|
|
Varsity [104]
TigerPulse: 92%
11
|
Re: FB Update: A computer cannot predict BYOG!
Jan 6, 2017, 2:27 PM
|
|
Tweet that!
|
|
|
|
|
|
Legend [6933]
TigerPulse: 99%
41
|
A computer cannot predict BYOG! ~this~***
Jan 6, 2017, 9:31 PM
|
|
|
|
|
|
|
|
Walk-On [149]
TigerPulse: 63%
11
|
Re: FB Update: Computer predicts winner of Clemson-Alabama
Jan 6, 2017, 7:25 PM
|
|
DUMB-A__ computer!
|
|
|
|
|
|
Tiger Spirit [9173]
TigerPulse: 100%
44
|
Well, I have 501 computers that each made one simulation
Jan 6, 2017, 9:50 PM
|
|
And Clemson actually won 58.1% of the time. Go figure!
|
|
|
|
|
|
Replies: 33
| visibility 360
|
|
|
50
