PDA

View Full Version : Achieving a fundamental base line



siouxperirish
03-19-2013, 05:20 PM
To begin, let me make it clear, this rationale I am suggesting should not consider game statistics post regular season, whereby a team is able to win themselves a bye because of dominant play. This also would not apply to the automatic seed given to those teams like Air Force, Army, etc. because those numbers would cause too much turbulence because of their inherent inconsistency.

What I am suggesting is a discussion around what quantifies a game played for teams that don't amass the same amount of tangible (actual) games played throughout the season. The affect that playing less games ~vs~ a full schedule can be answered better by players and coaches of the game. Consider what I'm suggesting analogous to combat, the more times you are exposed to danger, the better chance you have of getting hurt; more opportunity to have your collective ability as a team assessed and that assessment reflect against overall wins ~vs~ losses.

Teams (not counted what I first mentioned), that are able to cruise into the Frozen Four with a lighter season load will be fresher, less hurt and have potentially more players available for all games; those teams carrying a full complement of games are more exposed to among other things, poor calls by refs that are not able to reverse calls in retrospect so the affect of this could cause a team to lose a key player if he were to get that third misconduct. Playing a full season may allow individuals to collect better stats but overall, the teams with more ice time increase the likelihood they'll at the very least, be affected by a poor penalty.

With that out of the way, I'm suggesting a conversation on how to quantify that "non-game" for those teams with the lighter schedules. Again, not including those in the first paragraph.

A quick and easy way would be to remove outliers from those that play more games. Outliers removed would be based on the likelihood those teams with lighter schedules, would have fared against those same opponents. But how do you compare and qualify the ability of those teams with lighter schedules? You consider each of them individually against the teams they did play. That is, if Yale played Minnesota during a given year and Yale beat Minnesota, then an argument could be made that Yale "might" have beat the other teams Minnesota played. Accumulate the teams they both have in common for a given year, determine which team had more success with those teams they have in common and if the Minnesota ~vs~ Yale comparison comes out in Yale's favor, then Yale's factor for that comparison can be +1. Every team that doesn't play the same amount of games in a year (excluding 1st paragraph), should have their "non-game" games weighted to even this out.

The benefit would still be in the favor of those teams with less games because of injuries, bad penalties, etc. but I believe a method like this for starters, would help. There are going to be more methods to approach this but the most important consideration is how to better flatten out the baseline.

Another (future) discussion needs to be on a historical representation and how a team's intangibles should be accounted for. Obviously this is going to get even more squishy with team's recently added to the Category, "Men's Division 1". The art will be in how to create that history for those teams and then once that baseline can be established, certain unstable prognostication came become another topic for consideration. But just for fun really because then you would need to draw in coaching history, school support, etc.

SanTropez
03-19-2013, 09:41 PM
http://www.youtube.com/watch?v=s2xauk4l_Hg

Patman
03-19-2013, 09:52 PM
Teams will schedule games as a matter of filling seats. There are small ways to game the pairwise in the sense you can game your RPI with a strong schedule. Otherwise, it is not readily apparent that there is a way to do this. In theory you could try to figure what the TUC line is and schedule those teams, but doing that 1-2 years in advance is sincerely unlikely and hard.

As such, in the short term, I do not see a particular way to absolutely "game the system" and while there are imperfections there is enough dust in the air that its hard to take advantage of the imperfections.

As to teams being "fresher"... sincerely unlikely. I have seen some real physical AHA games over my time at UConn. I don't think currently there is a way to sandbag your way through life. It could be feasible down the line if the NCAA hockey pool weakens (say, double current sport sponsorship) then its possible one could trawl the chum to a high record and that might be a way to get by and a weakness in the method... but that goes back to the iffy-ness of the RPI and not so much the other components. The other components attempt to compare like with like in some manner.

So, in the end, you're asking for an omnibus system that some how represents the brain dump of human decision. You aren't going to get that system and even if you could approach such an object via modeling (and you won't... maybe a few lifetimes to pull that off) then it would be so intense to not be worthwhile.

We use simple measures in our life because they are easy to come by, easy to understand, and often sufficiently approximately correct. Don't expect any pure divination out of the data any time soon.

LTsatch
03-19-2013, 11:07 PM
I will hit this one from the street level. Couldn't it be more of an advantage to start practicing earlier and having more game experience? In your model, maybe the teams should start hiring risk managers, and remove players on the basis that they have a better chance to be injured in a certain game. It is an interesting thought, but most if not all hockey people will tell you that the Ivies shorter season is a disadvantage, hence the poor performance in national tournaments. These teams with shorter schedules are not "gaming" the system, they are playing with the hand they are dealt. In the case of Yale, some significant OOC wins are the only reason they are still in the NCAA picture.

Patman
03-19-2013, 11:17 PM
I will hit this one from the street level. Couldn't it be more of an advantage to start practicing earlier and having more game experience? In your model, maybe the teams should start hiring risk managers, and remove players on the basis that they have a better chance to be injured in a certain game. It is an interesting thought, but most if not all hockey people will tell you that the Ivies shorter season is a disadvantage, hence the poor performance in national tournaments. These teams with shorter schedules are not "gaming" the system, they are playing with the hand they are dealt. In the case of Yale, some significant OOC wins are the only reason they are still in the NCAA picture.

I think its unlikely that such moneyball tactics would ever be necessary. If you really wanted to get into it, it would be more advantageous for a team who will make the tournament and won't get a bad draw to sandbag... Quinny, Minny... sandbag... they have nothing to play for other than a league title right now.