Illinois #23 in 2/3 AP Poll

Status
Not open for further replies.
#76      
NarrowJ
the juiceman cometh said:
Pretty sure nobody has suggested that predictive metrics don't use data.
Click to expand...

It really seemed to be insinuated, although perhaps unintentionally.

I said that the predictive metrics are results-based (which the sites themselves explain exactly this) and seemed folks disagreed with that.

Edit: If the explanation is the value of wins vs losses then that does help explain the differences in those numbers. I'm still interested in knowing if the prolonged absences of our two best players are taken into account. I'm guessing that is not factored in.
 
Last edited:
#77      
the juiceman cometh
NarrowJ said:
It really seemed to be insinuated, although perhaps unintentionally.

I said that the predictive metrics are results-based (which the sites themselves explain exactly this) and have been told I am wrong about that.
Click to expand...
The result of a game is a win or a loss. "Results based" means based on wins and losses. Predictive metrics are not based on wins and losses.
 
#79      
NarrowJ
theNewGuy said:
So it seems:
Results based = only looking at the end result
Metric based = looking at what happened during the game
Click to expand...

So, which one is more accurate / better?

After this entire conversation my conclusion is we've lost a lot of games we shouldn't have. But again, my next question is now: are key players missing games accounted for in the metrics?
 
#80      
the juiceman cometh
theNewGuy said:
So it seems:
Results based = only looking at the end result
Metric based = looking at what happened during the game
Click to expand...
I think more accurately it would be:

Results based: Primary goal is a resume based on wins/losses and quality of opponents (looking backwards)

Predictive: Primary goal is to predict future results by determining offensive and defensive strength of a given team (looking forwards)

They all obviously give different weight to different factors or KenPom and Torvik would rank everyone the same.

Both are valuable in their own way. Take North Carolina v. Memphis. SOR has Memphis at #16. Seems reasonable. They are 18-4 and have beaten some good teams. SOR has UNC at #51. Again, seems reasonable. They've had a tough schedule but 13-10 is not a great record. KenPom though has Memphis at #40 and UNC at #39, and their scores are practically the same (separated by .01). Why? Because KenPom's algorithm has seen something that makes it think UNC is capable of better results going forward than its record suggests and Memphis is expected to do worse going forward than its record suggests. I think there's value in that, but also value in knowing that Memphis has a good resume right now, and UNC does not.
 
#81      
Snoopy
NarrowJ said:
So, which one is more accurate / better?

After this entire conversation my conclusion is we've lost a lot of games we shouldn't have. But again, my next question is now: are key players missing games accounted for in the metrics?
Click to expand...
You could ask google, which returns: KenPom doesn't take injuries into account, while sportsbooks do. If one of the teams loses a key player right before a match, bookmakers have a tendency to overreact and push the lines, while the KenPom prediction remains unchanged.

Quoting pointspreads.com

I suspect this is a manpower issue. As tracking injuries for every NCAA game would be costly and over the course of the season it probably averages out anyway. I believe the tournament committee does consider injury. Hence Self's blatant lies last year while working the committee for a better seed.

Edit, I should add, despite their limitations, not taking into account injuries, refs, student section on holiday, time between games, etc. The efficiency based models have proven more accurate than schedule based models. But in any endeavor, it is important to understand your models and their strengths and weaknesses when using them to make a decision.
 
Last edited:
#82      
NarrowJ
Snoopy said:
You could ask google, which returns: KenPom doesn't take injuries into account, while sportsbooks do. If one of the teams loses a key player right before a match, bookmakers have a tendency to overreact and push the lines, while the KenPom prediction remains unchanged.

Quoting pointspreads.com

I suspect this is a manpower issue. As tracking injuries for every NCAA game would be costly and over the course of the season it probably averages out anyway. I believe the tournament committee does consider injury. Hence Self's blatant lies last year while working the committee for a better seed.
Click to expand...

Last time I googled and reported the answer back to this message board, I was told I was incorrect lol

But thanks for the pro tip (y)

I did know the selection committee considers player absences, but wasn't sure about the ratings sites because they don't seem to say one way or another (unless I just missed it previously).
 
#83      
skyIdub
200w.gif
 
#84      
battle
I have wondered if KenPom reflects starters being pulled (I have searched). There has been an increased tendency keep starters in the game due to Net. Unlike data issues with player absences, the data is there to know that 3-5 starters were pulled for bench and then for non-rotation players and put less weight on those plays for both teams.
 
#85      
the juiceman cometh
battle said:
I have wondered if KenPom reflects starters being pulled (I have searched). There has been an increased tendency keep starters in the game due to Net. Unlike data issues with player absences, the data is there to know that 3-5 starters were pulled for bench and then for non-rotation players and put less weight on those plays for both teams.
Click to expand...
I believe KenPom and Torvik have some kind of garbage time adjustment where if the score of a game gets lopsided enough, the subsequent scoring for the rest of the game is weighted less. Has nothing to do with which players are on the court though.
 
#87      
pweo
Animated GIF


Is this predictive or results based?
 
#88      
Fighter of the Nightman
Well, even if we beat Minnesota by 80 points on Saturday, we have finally convinced the pollsters to dump us from the rankings, lol. Would probably have to go 3-0 in our next 3 games to even have a prayer at being ranked again anytime soon.
 
#90      
Illini2010-11
Fighter of the Nightman said:
Well, even if we beat Minnesota by 80 points on Saturday, we have finally convinced the pollsters to dump us from the rankings, lol. Would probably have to go 3-0 in our next 3 games to even have a prayer at being ranked again anytime soon.
Click to expand...
If (and a big one at that) the Illini go 3-0 the next 3 games, I would not call it a prayer to be ranked. A lot of mediocre teams this year vying for the 20-30 range. I would think the Illini would be a near lock to be ranked again. Of course, the results of the calendar year says this is implausible to say the least.
 
#91      
Old Farm Illini
Fighter of the Nightman said:
Well, even if we beat Minnesota by 80 points on Saturday, we have finally convinced the pollsters to dump us from the rankings, lol. Would probably have to go 3-0 in our next 3 games to even have a prayer at being ranked again anytime soon.
Click to expand...
I want the players to see that nice number next to Illinois go away for a week. I mean, they should have had a wake-up call multiple times already this season, but it seems like they still need a check on their egos. That being said, if we win on the road even at Minny, there's a chance we don't drop much. It's like Groundhog Day.
 
#92      
Fighter of the Nightman
Illini2010-11 said:
If (and a big one at that) the Illini go 3-0 the next 3 games, I would not call it a prayer to be ranked. A lot of mediocre teams this year vying for the 20-30 range. I would think the Illini would be a near lock to be ranked again. Of course, the results of the calendar year says this is implausible to say the least.
Click to expand...
Yeah, I really kind of misspoke there. If we beat Minnesota and sweep two ranked teams next week, we will indeed be ranked when the Monday 2/17 poll comes out ... we'd be 18-8 and would APPEAR to have turned a corner (who knows with this group, lol...). What I should have said is that if we don't win the next three, it would be a "prayer" to enter the top 25 again anytime soon.

I'm quite positive that we will not just drop out of the top 25 for next week but also fall pretty far down the RV category, and I don't think splitting with UCLA and MSU would be enough to dig us out (especially if the win is against a lower-ranked UCLA). And if we aren't sweeping UCLA and MSU at home in one week, then we for sure are not sweeping Wisconsin and Duke away from home the next!
 
Status
Not open for further replies.
Back