1/28/13 Poll Speculation

[joecct]

Re: 1/28/13 Poll Speculation

How did you ever get out of Clarkson with a 'D' in Advanced Calculus?

Switched majors to Accounting. I got tired of pulling stuff out of thin air (which was happening in Adv Calc II) and dropped the math major in favor of accounting. I was taking accounting anyway as a 2nd major and it just was more comfortable.

I should have paid more attention to the Aptitude test I took in HS. It said my career should be (1) Mech Eng (2) CPA (3) Civil Eng (6) Mathematics. And since my Dad was a CPA, I'm a CPA, and my goalie son was summa as an Accounting major, I guess it's in the genes.

 

[NUProf]

Re: 1/28/13 Poll Speculation

How did you ever get out of Clarkson with a 'D' in Advanced Calculus?

Advanced Calc is a theoretical course. Engineering majors don't have to excel in theoretical courses. On the other hand if you get a D in Diff Eq, you would be toast.

 

[Fishman'81]

Re: 1/28/13 Poll Speculation

Do you know anything about the metrics that the so-called "KRACH" (actually the Bradly-Terry system) uses? There is a lot of smoke and mirrors that goes on there - it uses logistic regression and was actually originally designed to rank chess players. The d3 hockey schedule doesn't meet all of the assumptions that are required for the use of logistic regression, so there are some "cluges" built into it that make it not as reliable as the KRACH fanboys would have you think. It is an interesting system and worthy of conversation, but it is not close to the be-all end all of ranking systems. (In no way am I implying that other well-known computer ranking systems are better, because due to low connectivity, no computer ranking system will ever be as reliable as we would like).

In short, Prof, not so much.

While I took my requisite year of stats in college and did quite well, I don't profess to remember anything much of substance... There's just been too much time elapsed without me having to apply it to anything I do. (BTW:Your later post was appreciated; I kinda-sorta followed it, despite my ignorance.)

Still, having closely followed the D-1 KRACH for years, I believe that it churns-out results which pass the sniff-test better than any other statistical metric I've seen. (I have touted the PWR because it's transparent, yes; but it certainly can be ridiculously volatile, and I hate that TUC "cliff".)

I do enjoy the majority of your posts; you have been a font of information over the years as to the D-3 process. If you forgive me for taking a cheap shot at how inscrutably static two other rankings have been lately in regards to the respective teams you and I follow, I'll turn the other cheek to your oblique but unmistakable "fanboy" remark.

 

[Fishman'81]

Re: 1/28/13 Poll Speculation

Would fishman be touting the KRACH if Utica wasn't #1 in it? That's the $64,000 question./QUOTE]

OK, so which team do you think deserves that rating, if you wonder why I do..?

(That's an honest question.)

 

[PrezdeJohnson09]

Would fishman be touting the KRACH if Utica wasn't #1 in it? That's the $64,000 question./QUOTE]

OK, so which team do you think deserves that rating, if you wonder why I do..?

(That's an honest question.)

I think it should be Utica based on resume alone or Eau Claire right now.

I haven't seen either play though.

Norwich, Oswego, Bowdoin would be in the next tier.

 

[joecct]

Advanced Calc is a theoretical course. Engineering majors don't have to excel in theoretical courses. On the other hand if you get a D in Diff Eq, you would be toast.

A in Diffy Q.

 

[CARDS_rule_the_Burgh]

I got a "C" in sadistics. I followed most of it. When John Whelan and Prof get talking about the higher math stuff, my brain goes into lock (said by the guy who never worked so hard in his life for a "D" in Advanced Calculus).

As a lib arts major, my college math course was "The History of Mathematics".

 

[Ozz]

Re: 1/28/13 Poll Speculation

As a lib arts major, my college math course was "The History of Mathematics".

I took that class! One of the toughest math classes that I took as it wasnt so much about the history of math, but more doing math the way others in the past did it. It was quite difficult trying to do a math problem a certain way when you already know a much easier way to do it!

 

[polarbearfan]

Re: 1/28/13 Poll Speculation

Prof, I was just looking at your computer rankings this week and was curious... do you factor home/away or margain of victory into your rankings or just W/T/L?

 

[NUProf]

Re: 1/28/13 Poll Speculation

I'll turn the other cheek to your oblique but unmistakable "fanboy" remark.

That wasn't solely aimed at you. I have found through the years that there are many, many people who tout the KRACH as some kind of perfect system - and refuse to understand that there are inherent flaws in the system, some statistical, and some more practical. The practical flaw is that to the uninformed, it seems almost inscrutable. It's lack of anything resembling transparency is a major disadvantage.

 

[hawk]

Re: 1/28/13 Poll Speculation

That wasn't solely aimed at you. I have found through the years that there are many, many people who tout the KRACH as some kind of perfect system - and refuse to understand that there are inherent flaws in the system, some statistical, and some more practical. The practical flaw is that to the uninformed, it seems almost inscrutable. It's lack of anything resembling transparency is a major disadvantage.

Prof why isn't Castleton in the top 15 D3 poll. They have wins over both Plattsburgh and Manhattanville and I think there only loss for some time was against Norwich. They just defeated a good St. A's team this past weekend. Maybe if they defeat both UMB and Babson this weekend the will get some notice.

 

[NUProf]

Re: 1/28/13 Poll Speculation

Prof, I was just looking at your computer rankings this week and was curious... do you factor home/away or margain of victory into your rankings or just W/T/L?

I don't consider margin of victory or location. I could, and I've played with incorporating both into my model, but have discarded them as not adding much value to the analysis. The reasons are more philosophical than statistical

1. For the most part home and away will balance out over the course of the season. Some teams play a higher percentage of home games than others, but for the most part that is kind of a "wash" factor.

2. The problem with goal scoring models is that I don't like trying to measure something that someone doesn't try to do. At the outset a team is trying to win - period they aren't trying to win by a set number of goals. Whether Norwich beats Southern Maine by 2 goal or by a "no-class" ten goals is not relevant and it really doesn't necessarily provide insight into the strength of either team. Look at this season's first game between Genny and Oz. Did the blow out nature of that game really reflect the strength of either team.

 

[polarbearfan]

Re: 1/28/13 Poll Speculation

I don't consider margin of victory or location. I could, and I've played with incorporating both into my model, but have discarded them as not adding much value to the analysis. The reasons are more philosophical than statistical

1. For the most part home and away will balance out over the course of the season. Some teams play a higher percentage of home games than others, but for the most part that is kind of a "wash" factor.

2. The problem with goal scoring models is that I don't like trying to measure something that someone doesn't try to do. At the outset a team is trying to win - period they aren't trying to win by a set number of goals. Whether Norwich beats Southern Maine by 2 goal or by a "no-class" ten goals is not relevant and it really doesn't necessarily provide insight into the strength of either team. Look at this season's first game between Genny and Oz. Did the blow out nature of that game really reflect the strength of either team.

Something that I have played with in the past is instead of incorporating pure goal differential, incorporating a metric that differentiates a convincing win from an overtime nail biter. The problem you run into is the subjectivity of it but philosophically, and I understand the goal is to win, I like the idea of a team getting more credit for winning by 3+ on the road than for eeking one out at home. I incorporated the home/away into the same metric. Obviously the model wasn't statistically pure but I found I agreed with the rankings more-so than the straight up W/T/L model at the end of the day.

Thank you for taking the time to do your rankings, I know its quite time consuming!

 

[OSU2010]

I don't consider margin of victory or location. I could, and I've played with incorporating both into my model, but have discarded them as not adding much value to the analysis. The reasons are more philosophical than statistical

1. For the most part home and away will balance out over the course of the season. Some teams play a higher percentage of home games than others, but for the most part that is kind of a "wash" factor.

2. The problem with goal scoring models is that I don't like trying to measure something that someone doesn't try to do. At the outset a team is trying to win - period they aren't trying to win by a set number of goals. Whether Norwich beats Southern Maine by 2 goal or by a "no-class" ten goals is not relevant and it really doesn't necessarily provide insight into the strength of either team. Look at this season's first game between Genny and Oz. Did the blow out nature of that game really reflect the strength of either team.

I'm just curious if you could post the hard numbers you use in your computer rankings, I'd like to see if I can follow the math. I took statistics 9 years ago in College but I think it would be kinda fun to try, I see the reasoning you use on the other site but I'd like to see the formulas as well. You can post it here or the other site.

 

[NUProf]

I'm just curious if you could post the hard numbers you use in your computer rankings, I'd like to see if I can follow the math. I took statistics 9 years ago in College but I think it would be kinda fun to try, I see the reasoning you use on the other site but I'd like to see the formulas as well. You can post it here or the other site.

I will include some more details on the other site in one of my weekly articles when I get a chance. I'm on "vacation" from one of my on line teaching jobs for a couple of weeks. I always try to sound literate so I take care when I do things like that

 

[OZFEST]

Re: 1/28/13 Poll Speculation

Margin of victory statistics would likely be at home in the other poll where the coaches vote (see college football)

 

[Fishman'81]

Re: 1/28/13 Poll Speculation

I think that margin of victory could be a fairly useful stat, if weighted correctly and skewed more heavily toward H2H comparisons. (And I wouldn't include ENGs.)

 

[Fishman'81]

Re: 1/28/13 Poll Speculation

That wasn't solely aimed at you. I have found through the years that there are many, many people who tout the KRACH as some kind of perfect system - and refuse to understand that there are inherent flaws in the system, some statistical, and some more practical. The practical flaw is that to the uninformed, it seems almost inscrutable. It's lack of anything resembling transparency is a major disadvantage.

I agree: I don't understand the nuts and bolts of it any better than I do one other metric... Nor do I consider anything perfect, including the KRACH.

As for the third (PWR), it's nice for everyone to be able to do the math at home. That process is as transparent as glass, but at the cost of being deeply flawed statistically.

My point was that the KRACH consistently shakes-out as the most logical late-season rankings every week, IMO, and I've been following it for years at the D-1 level.

While I don't love it, I certainly agree with its results more often than I do with the others' out there. (...Even when it reflects poorly on my favorite teams, which has been the case more often than not in the past several years.)

 

[bakdraft21]

Re: 1/28/13 Poll Speculation

the D3 site has a poll out for 2/4
1-norwich
2eau claire
3-utica
4-hobart
5-Ozzzzswego

I link if I could,but I think if any of the top 15 teams played each other it would be a 1 goal game,they all look very close,except Utica.....of course

 

[Fishman'81]

Re: 1/28/13 Poll Speculation

What you cite is the "I'll Find a Way to Make Norwich My #1" metric, and it's been that way all year long (maybe longer, I haven't paid any attention to it before early in this season.... And not at all, recently.)

While I haven't taken the trouble to go there in months, I'm guessing that the NCAA will afford it about as much weight as they do the the media poll. (None.) Prof has to re-tool his stats, if he wants to make that thing credible.

UC, Hobart, and Oswego look to be placed accurately in it, in an Eastern context. All three of those teams have much more solid resumes than say, Norwich, at this point.