Apparently you can't edit? God this forum is THE WORST lol
It is bad, but you can edit your own posts. Not sure why you could not.
Anyway, I do appreciate you putting the work in to resurrect your ratings. We have lost so many rankings (apparently) over the years, especially Rutter and Loch. It's great that you took the time to get all of yours flying again. Many thanks!
What I've never really understood, or maybe a better word is endorsed, regarding GRANT ratings is what you do with goal differential. Let's say that two teams with perfect records have each played 10 games. Team A won the first nine of those games by 1-0 shutouts, before winning one by a score of 2-1. Its ratio of goals scored versus goals allowed would be 11 to 1. Team B won all 10 of its games by 4-1 scores. Its ratio of goals versus goals allowed would be 40 to 10, or 4 to 1.
I don't remember for sure which way you make your assumption, but I think you said that a team with an 11 to 1 ratio is clearly superior to a team with a 4 to 1 ratio. As with most rankings, we're hampered in the hypothetical case above because we don't know anything about the caliber of competition, but I can argue that either Team A or Team B is stronger, depending on which point I'd like to make.
One way... Defense wins championships. A team that is allowing one goal through 10 games is going to be able to shut down its opponent and eventually will score. So Team A is clearly stronger.
Or ... Team B always has its games wrapped up by the final buzzer, while Team A is a bounce away from heading to OT. Hockey is a contest ultimately decided by relatively few plays, so not many bounces need to change to flip a result. For Team A, we're dependent on every bounce along the way. Team B has given itself a little more slack to survive some variance in its results. This suggests that Team B is stronger and in better shape moving forward.
We seldom see a team in women's D-I win a title when it doesn't have a great offense, and in particular, an offense that can score against strong competition. Wisconsin had great defensive teams in the ARD era, but its offense was more boa constrictor than cobra and had trouble delivering the lethal strike against top adversaries once we reached the "win or go home" point of the season.
It would be nice if a model included some dampening of the assumption that GRANT makes, allowing it to recognize a goal differential advantage in addition to a goal ratio advantage. For the A/B example above, it would be possible that Team A could have the same goals & goals allowed shown above for A and its opponents, but include a 2-0 win along with a 0-1 loss. That better highlights the vulnerability of Team A. Yes, goal ratio may matter, but I've seen nothing in my observations to convince me that it vastly outweighs goal differential. GRANT ignores that Team B has a goal differential of 30 while Team A has a goal differential of only 10. You can do all sorts of calculations from that point on, but if a fundamental assumption made early on is unsupported at best, and possibly even errant, there isn't any way to save the comparison.
I used to place the GRANT ratings alongside the other math-based models to see how it compares. Now, that Lock's WCHODR looks to be dormant, KRACH is about all that we have for comparison. RPI/PWR, for all of its attempts at tweaks and corrections over the years, is still about as fair as the federal income tax code.
