Preseason Rankings: AP vs Pomeroy

[gary-7]

No need. You are not correct.

You might want to refresh on significance.

Good Lord man. what I tried to tell you is obviously 100% correct and it is not arguable, and BTW, it has zero to do with "signifigance".

I've been trying to be nice about this, and I'm really trying to help you here, but you are just embarrassing yourself now.

One last time: Reread what you posted and look up what "random" means and perhaps you'll catch your error.

It's rudimentarily simple. Seriously...

 

[What Would Jesus Do?]

Good Lord man. what I tried to tell you is obviously 100% correct and it is not arguable, and BTW, it has zero to do with "signifigance".

I've been trying to be nice about this, and I'm really trying to help you here, but you are just embarrassing yourself now.

One last time: Reread what you posted and look up what "random" means and perhaps you'll catch your error.

It's rudimentarily simple. Seriously...

I know you're wrong, but if you disagree (as obviously you do) perhaps you'll explain. I laid out my thought experiment. I made it really simple. Maybe you can do the same?

Here's my thought experiment again, in case anyone wants to see what we're talking about.

1. Create a ratings list of all 364 teams by selecting them at random.​

2. Get Pomeroy's ratings.​

3. Compare the random list with Pomeroy's list at different points throughout the season.​

4. The null hypothesis is that neither will be appreciably better at predicting reality.​

My bet is that over repeated samplings, Pomeroy's list will outperform the random list with an extremely high level of significance. Seems obvious, doesn't it?

 

[DSouthr]

Sorry, but LMAO. That is patentaly absurd (not to mention flunks basic face validity).
Come on, man. If you select teams at random it's literally a coin-flip as to whether ANY alternative selection order will be better or worse, even if it's chosen by a trained chicken.

Moreover, by that logic, every single basketball fan of even the most rudimentary acumen will do better than random because they all know (just for example) that Gonzaga will be better than Elon.

I look at stats --- hell, as a "Dean disciple" I used them when I coached (and probably more than my contemporaries). You can certainly look back and say we need to be better this season at, say, "TO ratio" or the like, but as a fan until this season's games are played, all those "advanced metrics" from the past aren't gonna tell you any more or less than a reasonably aware eye will as to eventual records, any more than any one preseason poll is destined to be any more accurate than another.

I don't know what your obsession with this pomeroy stuff is about, but again hey, enjoy what you enjoy. And so with that said, my advice is to let it go and look forward to Monday.

LMAO, sorry but I have to give it up for gary in this one, on the "trained chicken" line alone!

gary actually nailed my rub with ranking at this time of the season, they do very little good, they are some what educated guesses that mean very little. WE were ranked pre-season #1 with a ton of stats to prove that we deserved in season before last and we wet the bed, didn't even make the post season?

But what the ranking do is very clearly propagate a narrative that does not easily go away. This conference is the best in the land therefore we should not weigh the losses in that power conference as heavy as losses in other less powerful conferences? The ACC has been devalued over the last few years, a narrative that at very least has limited ACC teams ability to play in the NCAAT or negatively effected their seeding.

It makes absolutely ZERO sense to me to weight what happens early season = or even close to the out comes late season when teams have seasoned and come together. But the narrative drives the early hype that one conference is greater than another, for example, the ACC we tend to see our natural hated opponents twice in the regular season and no matter what, beating a team twice in the same season is a hard thing to do. So maybe we slip up against a GT in a game they play way above their heads and we pay a great seeding price for that because the pre-season narrative dictated the ACC as weak so a bottom level program is especially weak. Yet Kansas could lose to a BYU and because the Kansas conference is pumped as the strongest conference they are not hurt seeding wise nearly as much as we would be.

 

[gary-7]

I know you're wrong, but if you disagree (as obviously you do) perhaps you'll explain. I laid out my thought experiment. I made it really simple. Maybe you can do the same?

Here's my thought experiment again, in case anyone wants to see what we're talking about.

1. Create a ratings list of all 364 teams by selecting them at random.​

2. Get Pomeroy's ratings.​

3. Compare the random list with Pomeroy's list at different points throughout the season.​

4. The null hypothesis is that neither will be appreciably better at predicting reality.​

My bet is that over repeated samplings, Pomeroy's list will outperform the random list with an extremely high level of significance. Seems obvious, doesn't it?

LMAO. Yes, please show us once again that you're barking up an extremely wrong tree, not to mention beating a very dead horse.

So... lemme see if I can simplify it for ya:
ANY list that isn't random --- either from pomeroy or literally ANYONE on a basketball forum --- will outperform a random list. Why? Because a random list (which FYI, means every team has an equally likely chance of landing in any ranking position) will inevitably place some elite teams in the lower echelons (that's basic probability*, BTW). In fact, RANDOM means that there's a 50/50 chance that UNC, for example, could land in the bottom half. The same odds apply independently to UConn or dook or any of the other blue-bloods.

*As another math FYI, a Normal distribution of random samples would predict approximately HALF of elite teams landing in the bottom 50 percentile in any given sample iteration.

Thus, ANYONE with even a cursory knowledge of the game will know to rank obviously better teams in the upper echelons, as opposed to those anomolies endemic in random rankings.

In other words, the only thing one can glean from your "thought experiment" is that pomeroy will outperform RANDOM. Well yeah, and DUH!... as I tried to tell you in my initial response --- given that I could go to any sports bar in America and every patron there (who wasn't passed-out drunk) would beat random, so the experiment is basically MEANINGLESS.

Got it? Good, because this is getting tiresome.

 

[What Would Jesus Do?]

LMAO. Yes, please show us once again that you're barking up an extremely wrong tree, not to mention beating a very dead horse.

So... lemme see if I can simplify it for ya:
ANY list that isn't random --- either from pomeroy or literally ANYONE on a basketball forum --- will outperform a random list. Why? Because a random list (which FYI, means every team has an equally likely chance of landing in any ranking position) will inevitably place some elite teams in the lower echelons (that's basic probability*, BTW). In fact, RANDOM means that there's a 50/50 chance that UNC, for example, could land in the bottom half. The same odds apply independently to UConn or dook or any of the other blue-bloods.

*As another math FYI, a Normal distribution of random samples would predict approximately HALF of elite teams landing in the bottom 50 percentile in any given sample iteration.

Thus, ANYONE with even a cursory knowledge of the game will know to rank obviously better teams in the upper echelons, as opposed to those anomolies endemic in random rankings.

In other words, the only thing one can glean from your "thought experiment" is that pomeroy will outperform RANDOM. Well yeah, and DUH!... as I tried to tell you in my initial response --- given that I could go to any sports bar in America and every patron there (who wasn't passed-out drunk) would beat random, so the experiment is basically MEANINGLESS.

Got it? Good, because this is getting tiresome.

Now that you've wasted all our our time reviewing what was NOT the issue - even though you got part of that wrong, too - maybe you can get back to the thought experiment.

I'm sure you won't. Is it because you don't understand? Now that you have sort of agreed that Pomeroy is better than random, what do you think the next step is?