2024 top 20

[rolldodge]

I just can not understand this logic. There is a huge difference between losing to ND by 1 and beating BU by 1. Rankings are all about relative performance. Ignoring that is just wrong

Yeah, the "huge difference" is that one is a win and one is a loss.

Simple view. Try answering this scenario.
Team A loses to ND in OT
Team B beats BU in OT..
Team A plays Team B next week. what’s your line?

Do you really think Team A is worse than Team B?

First, its never as simple as evaluating one game versus one other when making a poll.

Second, as a rough measure, any team in the top 30 (lets use RPI despite its flaws) is capable of losing by 1 to any team in the top 10 on a given game day.

Third, good teams find a way to still win despite a bad day or extenuating circumstances. And beating a team in the top 20 (BU) on any given day is not easy for any team.

Losses can never be a positive. They can only be more or less of a negative. Given other data points, I might not take anything away from a team for losing to ND in OT, but it wouldn't add to their resume. They need to prove themselves through wins.

figured you wouldnt answer the question. Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isnt going to play either to a goal. Nobody has said losses were a positive, merely that you need to judge every performance for what it was. Its impressive to take a top 3 team in the country to OT. It gives you a glimpse to the ceiling of the team. Beating a middle of the pack team in OT, doesnt tell you much. Beating them by 10 actually may give you some insight. MOV is important.

Your question is irrelevant because we are talking about ranking teams in a poll, not gambling. If I had to rank those two teams in your scenario, and I had no other data, I would rank team B above team A.

Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isn't going to play either to a goal.

This is good-ole boys logic, that sets certain teams apart based on history and tradition and not results. We can't know how Quinnipiac will do until they play those teams. 1-7 Brown lost to then #3 Maryland by 1. Unranked Colgate beat Penn State. Unranked Georgtown beat Notre Dame. Top 20 Penn beat Duke.

Nobody has said losses were a positive

You did, in your scenario.


Losses are not irrelevant. They are useful to evaluate after you evaluate the strength of wins. But losses should be evaluated against another teams losses. Never against another teams wins, IMO.

[coda]

I just can not understand this logic. There is a huge difference between losing to ND by 1 and beating BU by 1. Rankings are all about relative performance. Ignoring that is just wrong

Yeah, the "huge difference" is that one is a win and one is a loss.

Simple view. Try answering this scenario.
Team A loses to ND in OT
Team B beats BU in OT..
Team A plays Team B next week. what’s your line?

Do you really think Team A is worse than Team B?

First, its never as simple as evaluating one game versus one other when making a poll.

Second, as a rough measure, any team in the top 30 (lets use RPI despite its flaws) is capable of losing by 1 to any team in the top 10 on a given game day.

Third, good teams find a way to still win despite a bad day or extenuating circumstances. And beating a team in the top 20 (BU) on any given day is not easy for any team.

Losses can never be a positive. They can only be more or less of a negative. Given other data points, I might not take anything away from a team for losing to ND in OT, but it wouldn't add to their resume. They need to prove themselves through wins.

figured you wouldnt answer the question. Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isnt going to play either to a goal. Nobody has said losses were a positive, merely that you need to judge every performance for what it was. Its impressive to take a top 3 team in the country to OT. It gives you a glimpse to the ceiling of the team. Beating a middle of the pack team in OT, doesnt tell you much. Beating them by 10 actually may give you some insight. MOV is important.

Your question is irrelevant because we are talking about ranking teams in a poll, not gambling. If I had to rank those two teams in your scenario, and I had no other data, I would rank team B above team A.

Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isn't going to play either to a goal.

This is good-ole boys logic, that sets certain teams apart based on history and tradition and not results. We can't know how Quinnipiac will do until they play those teams. 1-7 Brown lost to then #3 Maryland by 1. Unranked Colgate beat Penn State. Unranked Georgtown beat Notre Dame. Top 20 Penn beat Duke.

Nobody has said losses were a positive

You did, in your scenario.


Losses are not irrelevant. They are useful to evaluate after you evaluate the strength of wins. But losses should be evaluated against another teams losses. Never against another teams wins, IMO.

No, i did not. I stated it is a data point for performance. This is you throwing up a strawman fallacy. If you you just answered my original question, I doubt we would still be discussing this

[rolldodge]

I just can not understand this logic. There is a huge difference between losing to ND by 1 and beating BU by 1. Rankings are all about relative performance. Ignoring that is just wrong

Yeah, the "huge difference" is that one is a win and one is a loss.

Simple view. Try answering this scenario.
Team A loses to ND in OT
Team B beats BU in OT..
Team A plays Team B next week. what’s your line?

Do you really think Team A is worse than Team B?

First, its never as simple as evaluating one game versus one other when making a poll.

Second, as a rough measure, any team in the top 30 (lets use RPI despite its flaws) is capable of losing by 1 to any team in the top 10 on a given game day.

Third, good teams find a way to still win despite a bad day or extenuating circumstances. And beating a team in the top 20 (BU) on any given day is not easy for any team.

Losses can never be a positive. They can only be more or less of a negative. Given other data points, I might not take anything away from a team for losing to ND in OT, but it wouldn't add to their resume. They need to prove themselves through wins.

figured you wouldnt answer the question. Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isnt going to play either to a goal. Nobody has said losses were a positive, merely that you need to judge every performance for what it was. Its impressive to take a top 3 team in the country to OT. It gives you a glimpse to the ceiling of the team. Beating a middle of the pack team in OT, doesnt tell you much. Beating them by 10 actually may give you some insight. MOV is important.

Your question is irrelevant because we are talking about ranking teams in a poll, not gambling. If I had to rank those two teams in your scenario, and I had no other data, I would rank team B above team A.

Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isn't going to play either to a goal.

This is good-ole boys logic, that sets certain teams apart based on history and tradition and not results. We can't know how Quinnipiac will do until they play those teams. 1-7 Brown lost to then #3 Maryland by 1. Unranked Colgate beat Penn State. Unranked Georgtown beat Notre Dame. Top 20 Penn beat Duke.

Nobody has said losses were a positive

You did, in your scenario.


Losses are not irrelevant. They are useful to evaluate after you evaluate the strength of wins. But losses should be evaluated against another teams losses. Never against another teams wins, IMO.

No, i did not. I stated it is a data point for performance. This is you throwing up a strawman fallacy. If you you just answered my original question, I doubt we would still be discussing this

Strawman? Your scenario. One team lost, the other team won. You said the team that lost is better than the team that won. How is that not using a loss as a positive?

[coda]

I just can not understand this logic. There is a huge difference between losing to ND by 1 and beating BU by 1. Rankings are all about relative performance. Ignoring that is just wrong

Yeah, the "huge difference" is that one is a win and one is a loss.

Simple view. Try answering this scenario.
Team A loses to ND in OT
Team B beats BU in OT..
Team A plays Team B next week. what’s your line?

Do you really think Team A is worse than Team B?

First, its never as simple as evaluating one game versus one other when making a poll.

Second, as a rough measure, any team in the top 30 (lets use RPI despite its flaws) is capable of losing by 1 to any team in the top 10 on a given game day.

Third, good teams find a way to still win despite a bad day or extenuating circumstances. And beating a team in the top 20 (BU) on any given day is not easy for any team.

Losses can never be a positive. They can only be more or less of a negative. Given other data points, I might not take anything away from a team for losing to ND in OT, but it wouldn't add to their resume. They need to prove themselves through wins.

figured you wouldnt answer the question. Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isnt going to play either to a goal. Nobody has said losses were a positive, merely that you need to judge every performance for what it was. Its impressive to take a top 3 team in the country to OT. It gives you a glimpse to the ceiling of the team. Beating a middle of the pack team in OT, doesnt tell you much. Beating them by 10 actually may give you some insight. MOV is important.

Your question is irrelevant because we are talking about ranking teams in a poll, not gambling. If I had to rank those two teams in your scenario, and I had no other data, I would rank team B above team A.

Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isn't going to play either to a goal.

This is good-ole boys logic, that sets certain teams apart based on history and tradition and not results. We can't know how Quinnipiac will do until they play those teams. 1-7 Brown lost to then #3 Maryland by 1. Unranked Colgate beat Penn State. Unranked Georgtown beat Notre Dame. Top 20 Penn beat Duke.

Nobody has said losses were a positive

You did, in your scenario.


Losses are not irrelevant. They are useful to evaluate after you evaluate the strength of wins. But losses should be evaluated against another teams losses. Never against another teams wins, IMO.

No, i did not. I stated it is a data point for performance. This is you throwing up a strawman fallacy. If you you just answered my original question, I doubt we would still be discussing this

Strawman? Your scenario. One team lost, the other team won. You said the team that lost is better than the team that won. How is that not using a loss as a positive?

It does not mean the loss was good. It simply judges the performance of both teams on that day. Based on your refusal to answer the scenario, I am guessing you agree.

[rolldodge]

I just can not understand this logic. There is a huge difference between losing to ND by 1 and beating BU by 1. Rankings are all about relative performance. Ignoring that is just wrong

Yeah, the "huge difference" is that one is a win and one is a loss.

Simple view. Try answering this scenario.
Team A loses to ND in OT
Team B beats BU in OT..
Team A plays Team B next week. what’s your line?

Do you really think Team A is worse than Team B?

First, its never as simple as evaluating one game versus one other when making a poll.

Second, as a rough measure, any team in the top 30 (lets use RPI despite its flaws) is capable of losing by 1 to any team in the top 10 on a given game day.

Third, good teams find a way to still win despite a bad day or extenuating circumstances. And beating a team in the top 20 (BU) on any given day is not easy for any team.

Losses can never be a positive. They can only be more or less of a negative. Given other data points, I might not take anything away from a team for losing to ND in OT, but it wouldn't add to their resume. They need to prove themselves through wins.

figured you wouldnt answer the question. Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isnt going to play either to a goal. Nobody has said losses were a positive, merely that you need to judge every performance for what it was. Its impressive to take a top 3 team in the country to OT. It gives you a glimpse to the ceiling of the team. Beating a middle of the pack team in OT, doesnt tell you much. Beating them by 10 actually may give you some insight. MOV is important.

Your question is irrelevant because we are talking about ranking teams in a poll, not gambling. If I had to rank those two teams in your scenario, and I had no other data, I would rank team B above team A.

Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isn't going to play either to a goal.

This is good-ole boys logic, that sets certain teams apart based on history and tradition and not results. We can't know how Quinnipiac will do until they play those teams. 1-7 Brown lost to then #3 Maryland by 1. Unranked Colgate beat Penn State. Unranked Georgtown beat Notre Dame. Top 20 Penn beat Duke.

Nobody has said losses were a positive

You did, in your scenario.


Losses are not irrelevant. They are useful to evaluate after you evaluate the strength of wins. But losses should be evaluated against another teams losses. Never against another teams wins, IMO.

No, i did not. I stated it is a data point for performance. This is you throwing up a strawman fallacy. If you you just answered my original question, I doubt we would still be discussing this

Strawman? Your scenario. One team lost, the other team won. You said the team that lost is better than the team that won. How is that not using a loss as a positive?

It does not mean the loss was good. It simply judges the performance of both teams on that day. Based on your refusal to answer the scenario, I am guessing you agree.

Read again. I answered your scenario as it relates to this conversation. I'm not a betting person. Betting wouldn't exist if it wasn't easy to lose.

And yes, you are judging the performance of both teams on that day. And you are judging the loss as better than the win. Are we now talking about the semantics of "good"?

[coda]

I just can not understand this logic. There is a huge difference between losing to ND by 1 and beating BU by 1. Rankings are all about relative performance. Ignoring that is just wrong

Yeah, the "huge difference" is that one is a win and one is a loss.

Simple view. Try answering this scenario.
Team A loses to ND in OT
Team B beats BU in OT..
Team A plays Team B next week. what’s your line?

Do you really think Team A is worse than Team B?

First, its never as simple as evaluating one game versus one other when making a poll.

Second, as a rough measure, any team in the top 30 (lets use RPI despite its flaws) is capable of losing by 1 to any team in the top 10 on a given game day.

Third, good teams find a way to still win despite a bad day or extenuating circumstances. And beating a team in the top 20 (BU) on any given day is not easy for any team.

Losses can never be a positive. They can only be more or less of a negative. Given other data points, I might not take anything away from a team for losing to ND in OT, but it wouldn't add to their resume. They need to prove themselves through wins.

figured you wouldnt answer the question. Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isnt going to play either to a goal. Nobody has said losses were a positive, merely that you need to judge every performance for what it was. Its impressive to take a top 3 team in the country to OT. It gives you a glimpse to the ceiling of the team. Beating a middle of the pack team in OT, doesnt tell you much. Beating them by 10 actually may give you some insight. MOV is important.

Your question is irrelevant because we are talking about ranking teams in a poll, not gambling. If I had to rank those two teams in your scenario, and I had no other data, I would rank team B above team A.

Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isn't going to play either to a goal.

This is good-ole boys logic, that sets certain teams apart based on history and tradition and not results. We can't know how Quinnipiac will do until they play those teams. 1-7 Brown lost to then #3 Maryland by 1. Unranked Colgate beat Penn State. Unranked Georgtown beat Notre Dame. Top 20 Penn beat Duke.

Nobody has said losses were a positive

You did, in your scenario.


Losses are not irrelevant. They are useful to evaluate after you evaluate the strength of wins. But losses should be evaluated against another teams losses. Never against another teams wins, IMO.

No, i did not. I stated it is a data point for performance. This is you throwing up a strawman fallacy. If you you just answered my original question, I doubt we would still be discussing this

Strawman? Your scenario. One team lost, the other team won. You said the team that lost is better than the team that won. How is that not using a loss as a positive?

It does not mean the loss was good. It simply judges the performance of both teams on that day. Based on your refusal to answer the scenario, I am guessing you agree.

Read again. I answered your scenario as it relates to this conversation. I'm not a betting person. Betting wouldn't exist if it wasn't easy to lose.

And yes, you are judging the performance of both teams on that day. And you are judging the loss as better than the win. Are we now talking about the semantics of "good"?

Betting and ranking are essentially the same thing. Vegas has some of the best models out there. Not the semantics of good. Its about relative performance. Generally speaking it takes a better performance to take the #1 team to OT and lose, than to take the #20 to team to OT and win. I dont think that is controversial statement. It is just logical. (we can come up with injuries and conditions that can mitigate that, but it holds true on the majority of games)

[rolldodge]

Betting and ranking are essentially the same thing. Vegas has some of the best models out there. Not the semantics of good. Its about relative performance. Generally speaking it takes a better performance to take the #1 team to OT and lose, than to take the #20 to team to OT and win. I dont think that is controversial statement. It is just logical. (we can come up with injuries and conditions that can mitigate that, but it holds true on the majority of games)

This is where we disagree. Betting is attempting to predict the future. Ranking is evaluating the present. Betting is two teams in a given location on a given day. Ranking is all teams over time and space.

Here's the logical endpoint of your argument: An 0-10 team with OT losses to all top 10 RPI teams is ranked higher than a 10-0 team with OT wins over Teams in the 11-20 RPI.

If you can't apply your criteria consistently, its not sound.

[OCanada]

Betting can also fluid as the spreads are adjusted to react to events and hedge risk tes?

[rolldodge]

Betting can also fluid as the spreads are adjusted to react to events and hedge risk tes?

Yes, and betting models are not designed to predict which team is better. They are designed to extract the most money from punters.

[coda]

Betting and ranking are essentially the same thing. Vegas has some of the best models out there. Not the semantics of good. Its about relative performance. Generally speaking it takes a better performance to take the #1 team to OT and lose, than to take the #20 to team to OT and win. I dont think that is controversial statement. It is just logical. (we can come up with injuries and conditions that can mitigate that, but it holds true on the majority of games)

This is where we disagree. Betting is attempting to predict the future. Ranking is evaluating the present. Betting is two teams in a given location on a given day. Ranking is all teams over time and space.

Here's the logical endpoint of your argument: An 0-10 team with OT losses to all top 10 RPI teams is ranked higher than a 10-0 team with OT wins over Teams in the 11-20 RPI.

If you can't apply your criteria consistently, its not sound.

We can say the same thing about a team that loses every game in OT vs the top 5 and a team that wins every game in OT vs the 20-25.. You take the first one. Both scenarios are silly. The point of modeling games is to track the performances, figure out the ceiling and floor, and the mean performance. Over the course of the season the mean performance becomes clear. If you simply say all wins are a good performance and all losses are a bad performance, you missing most the data and have garbage rankings.

[coda]

Betting can also fluid as the spreads are adjusted to react to events and hedge risk tes?

Yes. The more in-depth your model the more variables it can handle. Injuries are huge portion of the game. Recency bias can be very helpful or harmful.

[Farfromgeneva]

I just can not understand this logic. There is a huge difference between losing to ND by 1 and beating BU by 1. Rankings are all about relative performance. Ignoring that is just wrong

Yeah, the "huge difference" is that one is a win and one is a loss.

Simple view. Try answering this scenario.
Team A loses to ND in OT
Team B beats BU in OT..
Team A plays Team B next week. what’s your line?

Do you really think Team A is worse than Team B?

First, its never as simple as evaluating one game versus one other when making a poll.

Second, as a rough measure, any team in the top 30 (lets use RPI despite its flaws) is capable of losing by 1 to any team in the top 10 on a given game day.

Third, good teams find a way to still win despite a bad day or extenuating circumstances. And beating a team in the top 20 (BU) on any given day is not easy for any team.

Losses can never be a positive. They can only be more or less of a negative. Given other data points, I might not take anything away from a team for losing to ND in OT, but it wouldn't add to their resume. They need to prove themselves through wins.

figured you wouldnt answer the question. Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isnt going to play either to a goal. Nobody has said losses were a positive, merely that you need to judge every performance for what it was. Its impressive to take a top 3 team in the country to OT. It gives you a glimpse to the ceiling of the team. Beating a middle of the pack team in OT, doesnt tell you much. Beating them by 10 actually may give you some insight. MOV is important.

Your question is irrelevant because we are talking about ranking teams in a poll, not gambling. If I had to rank those two teams in your scenario, and I had no other data, I would rank team B above team A.

Team like Quinnipiac is not playing UVA or ND to 1 goal, most the top 20 isn't going to play either to a goal.

This is good-ole boys logic, that sets certain teams apart based on history and tradition and not results. We can't know how Quinnipiac will do until they play those teams. 1-7 Brown lost to then #3 Maryland by 1. Unranked Colgate beat Penn State. Unranked Georgtown beat Notre Dame. Top 20 Penn beat Duke.

Nobody has said losses were a positive

You did, in your scenario.


Losses are not irrelevant. They are useful to evaluate after you evaluate the strength of wins. But losses should be evaluated against another teams losses. Never against another teams wins, IMO.

No, i did not. I stated it is a data point for performance. This is you throwing up a strawman fallacy. If you you just answered my original question, I doubt we would still be discussing this

Strawman? Your scenario. One team lost, the other team won. You said the team that lost is better than the team that won. How is that not using a loss as a positive?

It does not mean the loss was good. It simply judges the performance of both teams on that day. Based on your refusal to answer the scenario, I am guessing you agree.

This devolved quickly into can one crate a spectrum the brain is supposed to work to solve for ordinal ranking of groups or does to have to be binary or otherwise the model can’t work. Because one is arguing to align all raises on a spectrum however they are valued and the other is completely binary. No wat you can get past this if your working with different arrays.

[Farfromgeneva]

Betting can also fluid as the spreads are adjusted to react to events and hedge risk tes?

You had me until you used betting and hedging int he same sentence. Unless your in prime brokerage sales…

[rolldodge]

We can say the same thing about a team that loses every game in OT vs the top 5 and a team that wins every game in OT vs the 20-25.. You take the first one. Both scenarios are silly.

Not silly. Improbable, but possible. Your model needs to be able to be handle the possible. Your model would rank a team with 0 wins. That's silly.

If you simply say all wins are a good performance and all losses are a bad performance, you missing most the data and have garbage rankings.

My model would not say this. Just that a team is first evaluated on the quality of its wins before losses can even come into the picture. And then, losses are evaluated against losses.

[coda]

We can say the same thing about a team that loses every game in OT vs the top 5 and a team that wins every game in OT vs the 20-25.. You take the first one. Both scenarios are silly.

Not silly. Improbable, but possible. Your model needs to be able to be handle the possible. Your model would rank a team with 0 wins. That's silly.

If you simply say all wins are a good performance and all losses are a bad performance, you missing most the data and have garbage rankings.

My model would not say this. Just that a team is first evaluated on the quality of its wins before losses can even come into the picture. And then, losses are evaluated against losses.

So your model would look at the quality of wins first and then evaluate the quality of losses.. sighs.

[rolldodge]

We can say the same thing about a team that loses every game in OT vs the top 5 and a team that wins every game in OT vs the 20-25.. You take the first one. Both scenarios are silly.

Not silly. Improbable, but possible. Your model needs to be able to be handle the possible. Your model would rank a team with 0 wins. That's silly.

If you simply say all wins are a good performance and all losses are a bad performance, you missing most the data and have garbage rankings.

My model would not say this. Just that a team is first evaluated on the quality of its wins before losses can even come into the picture. And then, losses are evaluated against losses.

So your model would look at the quality of wins first and then evaluate the quality of losses.. sighs.

Yes, some losses detract from a resume more than others.

[44WeWantMore]

Personally, I like the Power Rankings. I might like it even more if, instead of an arbitrary 10-goal cutoff, they used some kind of log function or Fibonacci sequence.

It incorporates margin of victory and home field advantage, but does not trend. That is, an outcome in February is as significant as one in May.

For human polls, I relish the commentary some folks used to write to justify their polling decisions (even if it was nothing more than a "Thud") in the old LasPower cays. But I understand that was a lot of work, and some of the criticism was excessive.

[coda]

Personally, I like the Power Rankings. I might like it even more if, instead of an arbitrary 10-goal cutoff, they used some kind of log function or Fibonacci sequence.

It incorporates margin of victory and home field advantage, but does not trend. That is, an outcome in February is as significant as one in May.

For human polls, I relish the commentary some folks used to write to justify their polling decisions (even if it was nothing more than a "Thud") in the old LasPower cays. But I understand that was a lot of work, and some of the criticism was excessive.

I dont mind the garbage time cut off. There are legitimate arguments for it. Putting a recency bias on a model is very interesting to me. I would almost want to have 2 models, one with it and one without it, and try to figure out the reason for the discrepancy. Was it injury related? Or just trending in a direction?

[Young Warrior]

Basic question - when folks talk about RPI or a certain school has "5 top 10 RPI wins" - the RPI that matters for ALL is end of year RPI, not what they were when played, correct?

[rolldodge]

Basic question - when folks talk about RPI or a certain school has "5 top 10 RPI wins" - the RPI that matters for ALL is end of year RPI, not what they were when played, correct?

Yep.

I've actually contemplated what it would look like if you froze the RPI values at the time of game and used those instead. There's probably just as many problems with that approach -- mainly, there's already too little data for RPI to really be meaningful at the end of season, so certainly its garbage two games in.