(01-04-2016 09:59 AM)RiceLad15 Wrote:  

(01-04-2016 09:55 AM)OptimisticOwl Wrote:  Out of curiosity, what kind of record would you demand from the next coach over a similar 40 game stretch, with 13 games against higher rated opponents and 27 against lower rated?

Ok, we are all going to say 27-0 for the latter, so just add in what would acceptable/unacceptable for the former.

PS: generic question for everybody.

On a win %, probably something similar in the teams ranked above. But that's because, ideally, we wouldn't be playing as many games against a team ranked above us because we would be ranked higher.

Using Sagarin is useful because it provides a metric to evaluate teams on more than a head-to-head basis, but when you start using it for comparative statistics you run into problems like this. The Sagarin rankings should, inherently, skew so that most teams have bad records against the teams ranked above them, and very good, to near perfect, records of teams ranked below them.

The only problem I have with using Sagarin is that the gaps are not equal. Number 58 is not necessarily better than #59 equally as much as number 70 is to #71. maybe some sort of power rating would be more enlightening.

(01-04-2016 10:02 AM)RiceLad15 Wrote:  

(01-04-2016 09:39 AM)waltgreenberg Wrote:  

(01-04-2016 09:15 AM)Antarius Wrote:  To add more details,

2-11, the average Sagarin of the wins was 59. Average Sagarin of the losses 37

23-4, the average Sagarin of the wins was 133. Average Sagarin of the losses 114

Again, my apologies for the data screwup. Too many calculated fields daisy chained and missed it.

Again, my issue is not so much the 2-11 record against teams better than us, but the reality that in all 9 of those losses, we were woefully non-competitive (with the lone exception being, perhaps, this year's UT game), and blown off the field...and at least half of those games were against teams ranked outside the Top 50.

I agree with you that the margin of victory, and not the record, is of most importance, but are you sure that half of the losses were to teams ranked outside the Top 50? Because if that is the case, we had a decent amount of games against teams ranked rather high in Sagarin to have an average Sagarin ranking of 37 in the losses. I'd expect that a majority of the losses were to teams in the Top 50, based solely on the mean residing above 50.

Marshall, LaTech x 2, Western Kentucky, USM, UT. I believe only WKU this year was ranked inside the Top 50 when we played them, and just barely so.

(01-04-2016 10:11 AM)waltgreenberg Wrote:  

(01-04-2016 10:02 AM)RiceLad15 Wrote:  

(01-04-2016 09:39 AM)waltgreenberg Wrote:  

(01-04-2016 09:15 AM)Antarius Wrote:  To add more details,

2-11, the average Sagarin of the wins was 59. Average Sagarin of the losses 37

23-4, the average Sagarin of the wins was 133. Average Sagarin of the losses 114

Again, my apologies for the data screwup. Too many calculated fields daisy chained and missed it.

Again, my issue is not so much the 2-11 record against teams better than us, but the reality that in all 9 of those losses, we were woefully non-competitive (with the lone exception being, perhaps, this year's UT game), and blown off the field...and at least half of those games were against teams ranked outside the Top 50.

I agree with you that the margin of victory, and not the record, is of most importance, but are you sure that half of the losses were to teams ranked outside the Top 50? Because if that is the case, we had a decent amount of games against teams ranked rather high in Sagarin to have an average Sagarin ranking of 37 in the losses. I'd expect that a majority of the losses were to teams in the Top 50, based solely on the mean residing above 50.

Marshall, LaTech x 2, Western Kentucky, USM, UT. I believe only WKU this year was ranked inside the Top 50 when we played them, and just barely so.

None of these numbers appear to come from when we played the teams - they appear to be year end rankings.

That changes the WKU lose from this year (ranked 35), but keeps La Tech, USM, and UT as teams ranked outside the Top 50.

But if the other La Tech lose and Marshall lose are from 2014, they were both ranked inside the top 50 (34 and 23, respectively).

(01-04-2016 05:49 AM)Antarius Wrote:  here is a comparion of Sagarin ratings for C-USA teams (and included UH since they are the closest we have to a rival and are in direct competition for Houston press coverage etc.).



We are the only school to not have changed coaches since 2007. Next up, is to include markers on the sagarin graph about coaching changes so we can normalize for new staff etc (which is why originally I did not have 2007 up on the first set of graphs).

I haven't had much time to do any real analysis on this, but first impressions are
1. we are a mediocre team in a very bad conference. Not the worst (not by a long shot) but not good
2. The plateau term does have some merit as for several years our slope was relatively constant while others embarked on a roller coaster ride.
3. FAU, FIU and UNT should stop playing football
4. FAU, FIU and UNT all have nice and relatively new stadiums. Yet they suck. Further data to back up the argument that the EZF will not make a material difference to Rice without other major changes.
5. Coaching changes do not guarantee success. But not changing the coach guarantees mediocrity/failure.

Looking very roughly at just Rice's curve on your top chart from a Fibonacci (logarithmic) perspective, I seem to see a few things:

1. after the initial slide in 2009, and the poor start in 2007, Rice did steadily increase from 2009 to 2013, and the drop-off on your chart in 2014 was only slight. So that gives me perspective on why some, including the admin might feel it was not ready to panic.
2. 2015 is a significant change. The question seems to be: are we starting a new long crawl up again under Bailiff from the perceived new low point (2015) or are has something fundamentally changed and we have already hit our peak in 2013, never to return?
3. If another long crawl up, how long would it be this time? If a false positive, how long do we wait to know? If a change for the worse, we might need to see at least one more year of data if not two or three, to confirm this. But if we wait that long, what is the true cost to Rice and its fanbase (not just money) of forgoing making that decision sooner rather than later?
4. From the Fibonacci (logarithmic) perspective, we should be about to start the crawl up and the next peak should be higher than 2013. If so, how high would that purported new peak be and if it comes, would it be significantly high enough (it would seem to have to be much higher than our 2013 peak) to make a difference? And most importantly, how do we really know that this is actually the case?

Now, the problem I have with the above is that, while your charts do allow me insight to better see a perspective about the past 9 years I just could not see before (that Bailiff is not as bad a coach as I have maintained, and that he actually has been doing okay for the most part) these charts do not encompass what it is like to actually watch the actual games. In other words, there is no way to "graph the eye(or gut) test".

No chart of how often a viewer's stomach may turn or a fans eyes may melt during those charted games, and what that individually and cumulatively does to growing/shrinking our small fan base, etc... which is important after all from a revenue standpoint. Butts in seats and actual attendance provides insight in to those questions, and if they are charted on a curve against the Rice curve on your chart, I don't know that they would show things have been as rosy.

This is why I maintain that when hiring or retaining a head coach (or a player or anyone for that matter) metrics and statistics are useful, but even Billy Beane needs to account for a human factor to be ultimately successful. People are not machines or robots. There is a necessary human factor that statistics don't show.

The other thing missing from such charts would be a criticality factor of how important it is to get the situation right due to the shrinking or closing window of moving back up vs the P5 conferences, if that window has not already closed. I don't know how you'd chart that.

I'd still be interested in seeing such comparisons and curves graphed coincidentally from coaching tenure start to end against Bailiff's tenure for the previous Rice coaches to better get perspective on Bailiff's tenure vs other coaches. The Toad's would be just a point, but I'd be interested to see that season's breakdown as you previously did with Bailiff's seasons.

Some observations:

23-4

We get upset once every 6.75 games we are 'favored' (who knows whether we were actually favored in all 27 of those games, some are on the road).

2-11

We upset teams once every 6.5 games we are underdogs.

Is this really abnormal? If we are favored by 2, is it really an 'upset'? Vice-versa?

What is the national average?

(01-04-2016 12:48 PM)OptimisticOwl Wrote:  Some observations:

23-4

We get upset once every 6.75 games we are 'favored' (who knows whether we were actually favored in all 27 of those games, some are on the road).

2-11

We upset teams once every 6.5 games we are underdogs.

Is this really abnormal? If we are favored by 2, is it really an 'upset'? Vice-versa?

What is the national average?

That's what I was getting at earlier - if we were 11-2 against teams ranked higher than us over that period, wouldn't we be ranked higher and maybe some of those would end up being in the other column?

(01-04-2016 12:55 PM)JustAnotherAustinOwl Wrote:  

(01-04-2016 12:48 PM)OptimisticOwl Wrote:  Some observations:

23-4

We get upset once every 6.75 games we are 'favored' (who knows whether we were actually favored in all 27 of those games, some are on the road).

2-11

We upset teams once every 6.5 games we are underdogs.

Is this really abnormal? If we are favored by 2, is it really an 'upset'? Vice-versa?

What is the national average?

That's what I was getting at earlier - if we were 11-2 against teams ranked higher than us over that period, wouldn't we be ranked higher and maybe some of those would end up being in the other column?

Yep. This is where this sort of a comparison has limitations, especially when using the end of year numbers, which would have taken the outcome of a game where we were ranked against Team X into consideration.

The problem with using these rankings on a per-week basis, which would avoid that issue above, is that, at some point, you are over or under-estimating the quality of a team.

Where these types of evaluations are beneficial, IMO, is when we look at the data that Walt brings up a lot, the data about how we perform against teams ranked above/below a specific threshold. And that information has been hashed out a number of times, and has been contained in Ant's numbers. This way we help to reduce the data conforming to a known result, which would be a bad record against teams ranked above, and a good record against teams ranked below.

Quote:No chart of how often a viewer's stomach may turn or a fans eyes may melt during those charted games, and what that individually and cumulatively does to growing/shrinking our small fan base, etc... which is important after all from a revenue standpoint. Butts in seats and actual attendance provides insight in to those questions, and if they are charted on a curve against the Rice curve on your chart, I don't know that they would show things have been as rosy.

I agree. The issue here is that the improvement and all that jazz still drags us up to the 69th ranked team, which is not good at all. This ties back to a point I made a while ago that 69, 79, 89, 129 are all the same to 99% of CFB fans. Relevance is a step function and Rice under Bailiff has not even come close to making it to the stage of relevance. The graphs are merely an academic discussion of the degree of irrelevance.

I will have to see if I can write a program to compile the weekly numbers from Sagarin and produce a weekly and aggregate graph. I do not know if this data is available historically, however.

(01-04-2016 10:02 AM)RiceLad15 Wrote:  

(01-04-2016 09:39 AM)waltgreenberg Wrote:  

(01-04-2016 09:15 AM)Antarius Wrote:  To add more details,

2-11, the average Sagarin of the wins was 59. Average Sagarin of the losses 37

23-4, the average Sagarin of the wins was 133. Average Sagarin of the losses 114

Again, my apologies for the data screwup. Too many calculated fields daisy chained and missed it.

Again, my issue is not so much the 2-11 record against teams better than us, but the reality that in all 9 of those losses, we were woefully non-competitive (with the lone exception being, perhaps, this year's UT game), and blown off the field...and at least half of those games were against teams ranked outside the Top 50.

I agree with you that the margin of victory, and not the record, is of most importance, but are you sure that half of the losses were to teams ranked outside the Top 50? Because if that is the case, we had a decent amount of games against teams ranked rather high in Sagarin to have an average Sagarin ranking of 37 in the losses. I'd expect that a majority of the losses were to teams in the Top 50, based solely on the mean residing above 50.

3 out of the 11 were outside the top 50 between 2012 and 2014

2012 LaTech, 2013 UH and 2013 UNT. This excludes the 4 losses to teams ranked worse. In these 3 games, the average margin of defeat was 12 points.

Regarding wins, Rice has not beaten a top 50 team between 2009 and now (still need to compile data for all the 2007, 2008 opponents).

Best win, 2012 vs SMU (SMU was ranked 57)

(01-04-2016 02:25 PM)Antarius Wrote:  

(01-04-2016 10:42 AM)GoodOwl Wrote:  No chart of how often a viewer's stomach may turn or a fans eyes may melt during those charted games, and what that individually and cumulatively does to growing/shrinking our small fan base, etc... which is important after all from a revenue standpoint. Butts in seats and actual attendance provides insight in to those questions, and if they are charted on a curve against the Rice curve on your chart, I don't know that they would show things have been as rosy.

I agree. The issue here is that the improvement and all that jazz still drags us up to the 69th ranked team, which is not good at all. This ties back to a point I made a while ago that 69, 79, 89, 129 are all the same to 99% of CFB fans. Relevance is a step function and Rice under Bailiff has not even come close to making it to the stage of relevance. The graphs are merely an academic discussion of the degree of irrelevance.

I will have to see if I can write a program to compile the weekly numbers from Sagarin and produce a weekly and aggregate graph. I do not know if this data is available historically, however.

Thanks for acknowledging the point.

I'd still be most interested to see the curve comparison laid out on one chart between Bailiff and recent Rice coaches Graham, Hatfield, Goldsmith, Berndt, Brown in addition to the one you did above between Bailiff's Rice tenure and the other C-USA schools. To help put in perspective how Bailiff's performance compares with some of our other recent coaches. I'm not familiar with how to do the kinds of analysis you are doing myself, and if it's too much to ask, I understand. I appreciate the analysis you are doing for those who are more pure numbers people. Thank you.

(01-04-2016 02:34 PM)Antarius Wrote:  

(01-04-2016 10:02 AM)RiceLad15 Wrote:  

(01-04-2016 09:39 AM)waltgreenberg Wrote:  

(01-04-2016 09:15 AM)Antarius Wrote:  To add more details,

2-11, the average Sagarin of the wins was 59. Average Sagarin of the losses 37

23-4, the average Sagarin of the wins was 133. Average Sagarin of the losses 114

Again, my apologies for the data screwup. Too many calculated fields daisy chained and missed it.

Again, my issue is not so much the 2-11 record against teams better than us, but the reality that in all 9 of those losses, we were woefully non-competitive (with the lone exception being, perhaps, this year's UT game), and blown off the field...and at least half of those games were against teams ranked outside the Top 50.

I agree with you that the margin of victory, and not the record, is of most importance, but are you sure that half of the losses were to teams ranked outside the Top 50? Because if that is the case, we had a decent amount of games against teams ranked rather high in Sagarin to have an average Sagarin ranking of 37 in the losses. I'd expect that a majority of the losses were to teams in the Top 50, based solely on the mean residing above 50.

3 out of the 11 were outside the top 50 between 2012 and 2014

2012 LaTech, 2013 UH and 2013 UNT. This excludes the 4 losses to teams ranked worse. In these 3 games, the average margin of defeat was 12 points.

Regarding wins, Rice has not beaten a top 50 team between 2009 and now (still need to compile data for all the 2007, 2008 opponents).

Best win, 2012 vs SMU (SMU was ranked 57)

Pretty certain I can answer the 2008 question. Western Michigan was inside the top 50 before the bowl game, but after getting thrashed, they dropped out of the top 50. Houston was barely outside the top 50 - around 53, if I recall. (Just looked it up - they were #52, or #50 if you exclude the 2 1-AA teams ahead of them. I'm guessing from the Sagarin numbers being discussed earlier that 1-AA is being kept in all analyses.)

UPDATE: also looked up the Western Michigan number, at least by interpolation. While Sagarin doesn't show weekly numbers, Massey's composite does, and Western Michigan was 52nd in Sagarin before the bowl game - presumably 1-A only. Not sure if that's the Sagarin ELO or Sagarin composite number - for Sagarin, it had significant divergence between ELO and Predictor ratings and rankings (20 positions in the final ranking). So maybe not in top 50 before the bowl, but close.

The Massey composite actually had them in the top 50 - so a bunch of other rating engines had them as top 50.

I'm not following why this is such a big deal. We were 2-11 against teams ranked better than us and 23-4 against teams ranked worse than us. That's 25-15 total. As long as we're 25-15, then any additional win against somebody ranked ahead of us would have to be offset by a loss to somebody ranked worse than us. So if we upset one more team ranked ahead of us to improve to 3-10, that would mean we lost to one more team behind us to be 22-5. Either way, we're 25-15 against pretty much a cupcake schedule. What we need to do is be better than 25-15. Be 30-10, or 35-5, and then we've got something. The problem is not that we lost 11 to teams ranked ahead of us; it's the we lost 15 total. That's the number that needs to change.

(01-04-2016 11:14 PM)Owl 69/70/75 Wrote:  I'm not following why this is such a big deal. We were 2-11 against teams ranked better than us and 23-4 against teams ranked worse than us. That's 25-15 total. As long as we're 25-15, then any additional win against somebody ranked ahead of us would have to be offset by a loss to somebody ranked worse than us. So if we upset one more team ranked ahead of us to improve to 3-10, that would mean we lost to one more team behind us to be 22-5. Either way, we're 25-15 against pretty much a cupcake schedule. What we need to do is be better than 25-15. Be 30-10, or 35-5, and then we've got something. The problem is not that we lost 11 to teams ranked ahead of us; it's the we lost 15 total. That's the number that needs to change.

whenever I see 35-5 I think of Detroit.

(01-04-2016 11:14 PM)Owl 69/70/75 Wrote:  I'm not following why this is such a big deal. We were 2-11 against teams ranked better than us and 23-4 against teams ranked worse than us. That's 25-15 total. As long as we're 25-15, then any additional win against somebody ranked ahead of us would have to be offset by a loss to somebody ranked worse than us. So if we upset one more team ranked ahead of us to improve to 3-10, that would mean we lost to one more team behind us to be 22-5. Either way, we're 25-15 against pretty much a cupcake schedule. What we need to do is be better than 25-15. Be 30-10, or 35-5, and then we've got something. The problem is not that we lost 11 to teams ranked ahead of us; it's the we lost 15 total. That's the number that needs to change.

I agree. And this is the peak years of the Bailiff tenure that we are looking at. Including 2011 and 2015 and the picture looks significantly worse.

Ultimately, you are correct that Rice needs to go 27-0 (or close) before we can start chipping away at those 11 games where we were underdogs.

(01-04-2016 12:48 PM)OptimisticOwl Wrote:  Some observations:

23-4

We get upset once every 6.75 games we are 'favored' (who knows whether we were actually favored in all 27 of those games, some are on the road).

2-11

We upset teams once every 6.5 games we are underdogs.

Is this really abnormal? If we are favored by 2, is it really an 'upset'? Vice-versa?

What is the national average?

I will have to run the numbers.

Also, this 2-11 and 23-4 is the best 3 year period under Bailiff. When we include the rest, the numbers look far, far worse.

(01-04-2016 09:59 AM)RiceLad15 Wrote:  

(01-04-2016 09:55 AM)OptimisticOwl Wrote:  Out of curiosity, what kind of record would you demand from the next coach over a similar 40 game stretch, with 13 games against higher rated opponents and 27 against lower rated?

Ok, we are all going to say 27-0 for the latter, so just add in what would acceptable/unacceptable for the former.

PS: generic question for everybody.

On a win %, probably something similar in the teams ranked above. But that's because, ideally, we wouldn't be playing as many games against a team ranked above us because we would be ranked higher.

Using Sagarin is useful because it provides a metric to evaluate teams on more than a head-to-head basis, but when you start using it for comparative statistics you run into problems like this. The Sagarin rankings should, inherently, skew so that most teams have bad records against the teams ranked above them, and very good, to near perfect, records of teams ranked below them.

Exactly. Saying that Bailiff is 23-4 against teams ranked worse than his team and 2-11 against teams ranked better than his are excellent examples of meaningless statistics. The entire purpose of a rating system is to minimize if not eliminate so-called "ranking violations", wherein a team with a lesser rating defeats one with a greater rating. In a quality rating system, you'd expect a team to lose every game to a higher-ranked team and win every game against a lower-ranked team. The fact that Bailiff has a poor record against higher-ranked teams doesn't necessarily mean he's a bad coach; by a rating system's very nature, he's supposed to have a poor record against these teams. By way of example, if you think Bailiff should be canned for only winning two games in the past three seasons against better teams, that's still two wins more than Nick Saban at Alabama has over that time period (he's 0-3).

I'll post some more comments about Antarius's work when I have the time. Basically, it means less than he seems to think it does, although much of that isn't his fault.

Jonathan - it doesn't mean anything supremely insightful nor was it meant to. It was merely a deep dive to support several assertions made over the years about Bailiff being mediocre.

To me the peak of 69 in Sagarin in 2013 was enough to confirm that Bailiff has been a 9 year mistake. But that isn't enough for several others, so giving it some conte x t and graphs was intended to make the point abundantly clear

They more important point in these statistics is that Bailiff has managed 4 ranking violations against worse teams. Which is pretty poor considering we were ranked in the 70's. Further, it shows exactly how poor the quality of teams we have beaten up on are.

As a whole, 2-11 and 23-4 is meaningless. When taken in context of Rice being ranked in the 70's, it paints a picture that our 2012-2014 stretch wasn't nearly as impressive as people here want us to believe

(01-05-2016 01:17 AM)Antarius Wrote:  Jonathan - it doesn't mean anything supremely insightful nor was it meant to. It was merely a deep dive to support several assertions made over the years about Bailiff being mediocre.

Then why bring it up?

Quote:To me the peak of 69 in Sagarin in 2013 was enough to confirm that Bailiff has been a 9 year mistake. But that isn't enough for several others, so giving it some context and graphs was intended to make the point abundantly clear

The more important point in these statistics is that Bailiff has managed 4 ranking violations against worse teams. Which is pretty poor considering we were ranked in the 70's. Further, it shows exactly how poor the quality of teams we have beaten up on are.

How do you know that four ranking-violation losses is poor? Have you done the research to establish that four such losses in three years is poor? Maybe that's typical; maybe that's even good. How do you know this? What evidence do you have? As to beating poor-quality teams, would you rather be losing to them? This can be arranged (see the 1980s). In the end, Bailiff can only play the teams on his schedule, and any drop in the quality of teams being beaten may simply be a function of the quality of teams being on the schedule to be beaten.

One of the big problems about this debate (and why I stay mostly out of it) is that many of the participants are simply asserting without evidence, turning such posts into giant exercises in confirmation bias.

Quote:As a whole, 2-11 and 23-4 is meaningless.

Again, why bring up something that was meaningless to begin with?

Quote:When taken in context of Rice being ranked in the 70's, it paints a picture that our 2012-2014 stretch wasn't nearly as impressive as people here want us to believe

It's not chopped liver, either. In Massey's ratings, it's the best three-year stretch of Rice football since Ken Hatfield's first four seasons; it's a stretch comparable to Fred Goldsmith's tenure, and it isn't surpassed again until we get back to the Jess Neely era. Given the performance of Rice football in the past half-century, that's not saying a whole lot, but it is a start and should be acknowledged as such.