- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
re: CFB Advanced Metrics and Computer Polls, Week 13
Posted on 11/24/19 at 2:32 pm to Y.A. Tittle
Posted on 11/24/19 at 2:32 pm to Y.A. Tittle
quote:
I saw what you posted. It simply illustrated the flaws of it.
Elaborate.
Posted on 11/24/19 at 2:34 pm to KosmoCramer
Comparing relative intergame stats is not really necessarily indicative of anything.
Posted on 11/24/19 at 2:34 pm to Y.A. Tittle
quote:
Comparing relative intergame stats is not really necessarily indicative of anything.
what?
Posted on 11/24/19 at 2:42 pm to KosmoCramer
Point being, if you have an algorithm that is trying to add some sort of intergame stat component and also incorporate components that account for wins and losses, I would contend the stat component contains nothing really relevant with regard to what you are actually wanting to analyze. Compile rankings analyzing ONLY such stats and see.
Posted on 11/24/19 at 2:46 pm to Y.A. Tittle
quote:
Point being, if you have an algorithm that is trying to add some sort of intergame stat component and also incorporate components that account for wins and losses, I would contend the stat component contains nothing really relevant with regard to what you are actually wanting to analyze. Compile rankings analyzing ONLY such stats and see.
SP+ for instance has been updated over time as the link said. There has been signficant research done to find what stats and metrics have a high correlation to winning. Hence SP+ using those 5 data points as the crux of it's analysis.
I dont think you understand any of this honestly.
You're just using random buzzwords that don't hardly make sense.
Posted on 11/24/19 at 2:47 pm to KosmoCramer
quote:
SP+ for instance has been updated over time as the link said. There has been signficant research done to find what stats and metrics have a high correlation to winning. Hence SP+ using those 5 data points as the crux of it's analysis.
Sounds like it’s chasing its tail.
Posted on 11/24/19 at 2:48 pm to Y.A. Tittle
quote:
Sounds like it’s chasing its tail.
There is definitely that possibility and xiv has even said as such last week.
It's also 55% against the spread this year, so it's doing something very right.
Posted on 11/24/19 at 2:50 pm to KosmoCramer
quote:
I dont think you understand any of this honestly.
How much do these statistical components weigh into the SP rankings you’ve cited? ETA versus analyzing W-L data?
This post was edited on 11/24/19 at 2:52 pm
Posted on 11/24/19 at 2:54 pm to Y.A. Tittle
quote:
How much do these statistical components weigh into the SP rankings you’ve cited? ETA versus analyzing W-L data?
Wins and Losses dont factor into SP+ at all that I know of. If that's what you're asking.
More info:
quote:
The S&P+ Ratings are a college football ratings system derived from the play-by-play and drive data of all 800+ of a season's FBS college football games (and 140,000+ plays). The components for S&P+ reflect opponent-adjusted components of four of what Bill Connelly has deemed the Five Factors of college football: efficiency, explosiveness, field position, and finishing drives. (A fifth factor, turnovers, is informed marginally by sack rates, the only quality-based statistic that has a consistent relationship with turnover margins.) Here are the components of the ratings shared below:
Second-Order Wins (2ndO Wins): Defined here and discussed in further detail here and here, second-order wins compare the advanced statistical components of a given game, and the single-game win expectancy they create, to the actual results of the game. This projected win total is a cousin of the Pythagorean record, a concept common in many sports. They are presented below, with the difference between a team's wins and second-order wins in parentheses.
S&P+ rating: Using the five-factors concept above, the S&P+ ratings take into account efficiency (Success Rates), explosiveness (IsoPPP), and factors related to field position and finishing drives. It is now presented in two forms: the first is a percentile, and the second is an adjusted scoring margin specific for this specific season's scoring curve.
Off. S&P+ rating: A team's offense-specific S&P+ rating, presented in the form of an adjusted scoring average.
Def. S&P+ rating: A team's defense-specific S&P+ rating, presented in the form of an adjusted scoring average (and since this is defense, the lower the average, the better).
Special Teams S&P+ rating: This is an initial attempt to measure play-for-play special teams efficiency, weighted for overall importance. (Note: special teams ratings not available for 2005.)
Strength of Schedule rating (SOS): S&P+ strength of schedule is determined by projecting the win percentage an average top-five team could expect against a given team's schedule. A lower number, therefore, means a higher ranking.
Use the menu above for pages that give more details on offensive and defensive S&P+, or our other college football ratings, the drive-based FEI.
LINK
Posted on 11/24/19 at 2:58 pm to KosmoCramer
quote:
Wins and Losses dont factor into SP+ at all that I know of. If that's what you're asking.
It’s certainly a component of this.
quote:
Strength of Schedule rating (SOS): S&P+ strength of schedule is determined by projecting the win percentage an average top-five team could expect against a given team's schedule. A lower number, therefore, means a higher ranking.
Posted on 11/24/19 at 2:58 pm to KosmoCramer
Whoops.
This post was edited on 11/24/19 at 2:59 pm
Posted on 11/24/19 at 3:02 pm to Y.A. Tittle
quote:
Strength of Schedule rating (SOS): S&P+ strength of schedule is determined by projecting the win percentage an average top-five team could expect against a given team's schedule. A lower number, therefore, means a higher ranking.
I dont think so. It takes the top five teams and calculates what they would vs that schedule. It doesn't take into account actual game results.
Posted on 11/24/19 at 3:08 pm to KosmoCramer
I see what you’re saying.
Posted on 11/24/19 at 3:33 pm to Y.A. Tittle
quote:
I’m just curious, are they breaking down overall in-game stats or is it all just comparing relative overall rankings? Either would certainly have flaws, but whatever.
S&P+ and FPI are based on in game stats, though there is a blowout component, in which it stops counting if the game is out of hand. Basically, it values efficiency above all else, so it rates teams on how many yards/play it achieves versus the expected yards/play (very simplified).
Despite it's safeguards, S&P+ and FPI loves teams that beat the stuffing out of bad schedules. It's baked into the formula, as it is designed to normalize everyone's numbers. But it's better to destroy Clemson's schedule than, say, win by 7-10 points consistently with LSU's.
Posted on 11/24/19 at 3:54 pm to Baloo
If I’m looking at what Kosmo has posted correctly, it seems to me these algorithms are basically looking at statistical measures that have been previously successful and weighting those accordingly. Therefore, if you have a team like Alabama, who has generated certain statistical measures and been successful and is continuing to generate similar statistics, it should be ranked accordingly unless and until the metric is later adjusted to account for something different.
It has no means to account for some team (it doesn’t have to be LSU, but yeah it is now), who may have found a way to greater success doing something different statistically.
It has no means to account for some team (it doesn’t have to be LSU, but yeah it is now), who may have found a way to greater success doing something different statistically.
Posted on 11/24/19 at 4:22 pm to Y.A. Tittle
quote:
If I’m looking at what Kosmo has posted correctly, it seems to me these algorithms are basically looking at statistical measures that have been previously successful and weighting those accordingly. Therefore, if you have a team like Alabama, who has generated certain statistical measures and been successful and is continuing to generate similar statistics, it should be ranked accordingly unless and until the metric is later adjusted to account for something different.
It has no means to account for some team (it doesn’t have to be LSU, but yeah it is now), who may have found a way to greater success doing something different statistically.
This isn't new data or based on Alabama in any way. Its heavily based on efficiency.
It uses metrics that have a high correlation to make teams successful and win.
LSU didn't reinvent football this year.
Posted on 11/24/19 at 4:33 pm to KosmoCramer
quote:
LSU's defense
LSUs defense has not given up 17 unanswered points at any time this season.
Posted on 11/24/19 at 4:35 pm to Xenophon
Well then, there's one in the LSU column defensively.
Posted on 11/24/19 at 4:36 pm to Xenophon
quote:
LSUs defense has not given up 17 unanswered points at any time this season.
Good for them.
They're also ranked 29th in SP+ defense, 42nd in scoring defense, and 43rd in total defense. I can provide more data as needed.
This post was edited on 11/24/19 at 4:37 pm
Posted on 11/24/19 at 4:38 pm to KosmoCramer
quote:
They're also ranked 29th in SP+ defense, 42nd in scoring defense, and 43rd in total defense. I can provide more data as needed.
lol got em bro
Popular
Back to top
Follow TigerDroppings for LSU Football News