SOS - What it does and doesn't tell us

[SB Shock]

Seems like a moving target. You talk about SOS, then you give sime example and calculate it based on RPI. One of the hypothetical scenario was team A beats 1 and 299 and team B beats 149 and 151, now you are changing to 10ish and 340ish and 170ish? Instead of black and white it becoming shades of grey.

I think u need to back to grade school thinking and lay out the "givens" and no "ish", and have a "find" and lay out exactly what your are trying to calculate (ie average RPI, average SOS, etc).

Btw, in your previous example you calculated using RPI, u still have the same problem. RPI distribution is a 's-curve' so you will have to change our hypothesis to make it fit whatever your point is - which you tried to do.

[jdshock]

I don’t know if this helps or not, but here is the original hypothetical, restated as best I can to clear up any confusion that has been expressed in the first several pages of this thread.

Team A is 2-0. They beat one of the 10 best teams in the county and one of the 10 worst teams in the country.
Team B is 2-0. They beat two of the most mediocre teams in the country.There are 351 teams in D1, so we are talking about a couple teams in the 170s.

Nothing else is known. We are assuming that all the facts presented are indisputable. Team A beating one of the 10 best teams in the country is to be treated as a fact and is not up for debate.

We are left with a very simply analysis.

1 - Which team do you think is better based on this limited info? Do you think that beating a great team and a terrible team is more impressive, less impressive, or equally impressive than beating 2 mediocre teams?

2 – I would argue that any formula that calculates SOS would say that Team A and Team B have virtually the same SOS despite arriving at it through vastly different means. Does anyone dispute this?

I have been on the record twice now saying that "beating a great team and a terrible team" is more impressive than "beating 2 mediocre teams."

Your data itself doesn't back up your second point, though. You say the 1 351 matchup is just a "tad higher" than the rest. [1 is actually Kansas, not Kentucky. But maybe you meant Kansas benefits from being beaten by a freakishly good Kentucky team... I don't know]. Run the calculations for all of the teams in the top 10 versus their equivalent at the bottom of the pile. They will all be a "tad higher" than similar averages close to the 175 numbers.

But it's not just a tad higher. RPI numbers are ridiculously close. Take any team in the top 10 with its equivalent team at the bottom, and you will get an average with a number similar to beating a 150 team. The average of the 170s are, by definition, close to the 170s. It's not "virtually the same" at all.

Seriously, though, I agree with you. I think beating one incredible team is better than beating two mediocre teams. I am even willing to say that I'm probably more impressed with a win over a top 10 team and a win over a 351 team than two wins over 150 teams. But what do I know. I just think the SOS might be more complicated than you're giving it credit.

Edit to add: I am still interested in seeing where you are going with it ultimately. Why the secrecy and why the start with such small information. Why couldn't you have said "Team A beat teams 1-15 and teams 337-351. Team B beat teams 161-189" or whatever?

[Jamar Howard 4 President]

Sample size too small. There is not enough data to make a valid conclusion.

So you have absolutely no opinion on whether it is more impressive to beat a great team and a terrible team as compared to beating two mediocre teams?

[Shocker-maniac]

Not really. Get back to me after15 games and let me know how they are doing. Outliers happen all the time especially in the early season.

[Jamar Howard 4 President]

Not really. Get back to me after15 games and let me know how they are doing. Outliers happen all the time especially in the early season.

OK, say the hypothetical repeats itself 10 times in a row. At that point, can you make a judgment in favor of team A or B?

If you can, then you should be able to make a judgment for my small sample size. The only difference is your level of confidence that an outlier has/hasn't occurred. For the purposes of a hypothetical, outliers don't matter. We are simply discussing "rules of thumb". Does beating a great team and a terrible team generally impress you more than beating two mediocre teams? If you need to pretend that the sample size is larger to help you feel better, fine I guess, but I disagree that the larger sample size is truly necessary.

[jdmee]

Would this help in the current top 25 there are two teams, both with the same number of losses and the average RPI of the teams they lost to are about the same.

Team A

Average RPI of their Top 5 wins 22
Average RPI of their Bottom 5 wins 233

Team B
Average RPI of their top 5 wins 43
Average RPI of their bottom 5 wins 214.

Which team would you rank higher?

Or these Teams both with the same number of losses and Average RPI of all the wins within 15 of each other.

Team C

Top 5 61
Bot 5 245

Team D

Top 5 39
Bot 5 283

Do either of these reduce the outliers enough for everyone and is there enough to talk about the SOS?

[Jamar Howard 4 President]

I have been on the record twice now saying that "beating a great team and a terrible team" is more impressive than "beating 2 mediocre teams."

I appreciate that. Glad we agree on that point.

Your data itself doesn't back up your second point, though. You say the 1 351 matchup is just a "tad higher" than the rest. [1 is actually Kansas, not Kentucky. But maybe you meant Kansas benefits from being beaten by a freakishly good Kentucky team... I don't know]. Run the calculations for all of the teams in the top 10 versus their equivalent at the bottom of the pile. They will all be a "tad higher" than similar averages close to the 175 numbers.

But it's not just a tad higher. RPI numbers are ridiculously close. Take any team in the top 10 with its equivalent team at the bottom, and you will get an average with a number similar to beating a 150 team. The average of the 170s are, by definition, close to the 170s. It's not "virtually the same" at all.

You are right about Kansas being #1, not Kentucky. I wish I would have never even mentioned 1/351 being a slight outlier this year because ultimately I don't think it matters. Just ignore that statement all together.

Here is why I say that 1/351 compared to 175/176 are "virtually all the same":

The average of 1/100 is 0.6265
The average of 25/150 is 0.5608
The average of 130/300 is 0.4696
The average of 200/330 is 0.4344

Those 4 examples are all just random pairs of teams that could be played in back to back games. What they aren't is pairs of teams that are equally balanced away from the center (175.5) Notice that the results (0.62, 0.56, 0.47, 0.43) are all well outside the .49 to .51 window that I demonstrated with my 1/351, 11/341, 175/176 chart.

My point is that in any given 2 game stretch, there will be "tough" schedules and "easy schedules". Anyone who plays 1/351 or 175/176 will find their SOS to be right smack dab in the middle. A variance of 0.02 is not all that large. That is why I say all the 1/351 type of schedules all tend to lead to "virtually the same" SOS, regardless of if they are spread far apart (1/351) or close together (175/176).

Edit to add: I am still interested in seeing where you are going with it ultimately. Why the secrecy and why the start with such small information.

I shutter to think what would have happened if I hadn't kept my hypothetical as simple as possible. This thread has been all over the place even with a simple starting point.

Why couldn't you have said "Team A beat teams 1-15 and teams 337-351. Team B beat teams 161-189" or whatever?

Maybe I should reconsider something like you proposed. Beating 1-15 and 337-351 simply means expanding my hypothetical to have happened 15 times instead of once. Just as I responded to Shocker-maniac a couple of posts ago, I see no reason why this is necessary. If a conclusion can be drawn when an event happens 15 times, then it can also be drawn when it happens once. The only difference is level of certainty.

I wanted to keep things simple, but if it helps some people to expand the sample size to 30 games instead of 2, maybe we can do that. I just wanted to point out that I feel this is completely unnecessary.

[im4wsu]

So you have absolutely no opinion on whether it is more impressive to beat a great team and a terrible team as compared to beating two mediocre teams?

I would be more impressed by a team ranked #350 beating two teams ranked #170 than I would be by #1 team beating #9 and #351. That is not to say that the former team is better, however.

[Jamar Howard 4 President]

I would be more impressed by a team ranked #350 beating two teams ranked #170 than I would be by #1 team beating #9 and #351. That is not to say that the former team is better, however.

But where do you come up with #1 and #350?! You are introducing new, assumed, information in order to say that! We don't know if Team A is ranked #1. We don't know if Team B is ranked #350. That is just one of a million sub-scenarios that you came up with. Please quit adding new information to the scenario and then factoring your new information into your decision. I made a simple scenario. Let's keep it simple. It is totally unfair for you to assume something positive about team A (that they are #1) and something negative about Team B (that they are #350)

Team A and Team B are just 2 mysterious teams. All you know is the results of 2 of their games. All I want to know is which result (beating 1/351 vs 175/176) is a better resume booster in your opinion. Which result makes you say "I don't know much about these teams, but from what little I do know, I think Team ____ is the better team".

[DUShock]

Jamar, I would go with 1/351 because of the spread and let's face it, who doesn't like beating a #1.

[Jamar Howard 4 President]

Seems like a moving target. You talk about SOS, then you give sime example and calculate it based on RPI. One of the hypothetical scenario was team A beats 1 and 299 and team B beats 149 and 151, now you are changing to 10ish and 340ish and 170ish? Instead of black and white it becoming shades of grey.

I mentioned an example with RPI because it was just that, an example. I was trying to answer some detailed questions. Stepping back to the general scenario, I still maintain that it is RPI free.

I never used 299. I originally said 300+. I eventually clarified 300+ to simply be #341 so we could be more precise.
I never used 1. I originally said "top 10". I eventually clarified top 10 to simply be #10 so we could be more precise.
The closest thing I did to changing anything was going from "mid-100s" to "mid-170s", and I was very clear that this was to help clarify that these were average teams in a D1 environment of 351 total teams.
Calling these clarifications "creating a moving target" seems like a very unfair critique IMO. Nothing of any major substance changed. Only small details and clarifications so we could avoid any possible disagreements based on "technicalities".

Btw, in your previous example you calculated using RPI, u still have the same problem. RPI distribution is a 's-curve' so you will have to change our hypothesis to make it fit whatever your point is - which you tried to do.

I do not understand this statement about RPI distribution being an "s-curve".

[Jamar Howard 4 President]

@ShockTalk: @Rlh04d:

Please help. I understand that some people initially just wanted to be difficult. However, most of the posts now seem to be genuine. I am literally blown away by how difficult some people seem to think this needs to be. I see this as something very simple that we should have been able to all agree on very early.

What do the two of you think? I respect your opinions? Do you agree with all these "but what about..." posts that are being thrown around?

[Shocker-maniac]

I would say that both teams A and B would appear to have about the same SOS ranking because playing a top 10 team and a bottom 10 team averages to be about the same as playing two mediocre team. However, when you look at the details you would say team A is better because they beat a top 10 team.

[im4wsu]

But where do you come up with #1 and #350?! You are introducing new, assumed, information in order to say that! We don't know if Team A is ranked #1. We don't know if Team B is ranked #350. That is just one of a million sub-scenarios that you came up with. Please quit adding new information to the scenario and then factoring your new information into your decision. I made a simple scenario. Let's keep it simple. It is totally unfair for you to assume something positive about team A (that they are #1) and something negative about Team B (that they are #350)

Team A and Team B are just 2 mysterious teams. All you know is the results of 2 of their games. All I want to know is which result (beating 1/351 vs 175/176) is a better resume booster in your opinion. Which result makes you say "I don't know much about these teams, but from what little I do know, I think Team ____ is the better team".

It is a shame you are so locked onto your own narrow set of parameters that you totally miss what others are saying. Be that as it may, I will rephrase with your terminology:

I would be more impressed with a team that was not expected to beat two of the most mediocre teams in the country, but did than I would be with a team that was expected to beat one of the 10 best teams in the county and one of the 10 worst teams in the country and did. There is simply not sufficient information to determine, other than a totally unsubstantiated wild guess rather than a "decision," whether Team A or Team B is "better" or "ranked higher" or whatever criteria you finally land on for your "test."

But assuming the SOS is the same, what does it tell us or not tell us? It tells us nothing and doesn't tell us anything. There, I played you game.

[jdmee]

It is a shame you are so locked onto your own narrow set of parameters that you totally miss what others are saying. Be that as it may, I will rephrase with your terminology:

I would be more impressed with a team that was not expected to beat two of the most mediocre teams in the country, but did than I would be with a team that was expected to beat one of the 10 best teams in the county and one of the 10 worst teams in the country and did. There is simply not sufficient information to determine, other than a totally unsubstantiated wild guess rather than a "decision," whether Team A or Team B is "better" or "ranked higher" or whatever criteria you finally land on for your "test."

But assuming the SOS is the same, what does it tell us or not tell us? It tells us nothing and doesn't tell us anything. There, I played you game.

How about my teams. You know they are all top 25 teams. Which was more impressive. TEAM A or TEAM B. How about between Team C and D?