Guest Posted August 28, 2022 Posted August 28, 2022 4 minutes ago, jsb235 said: If you are fine with games counting differently, that's on you. In the Trey Lance example, you are counting in my games differently. His qb rating is 97.3, your method says it is 104.65, as you are giving to much weight to his game with a 127.1 rating.
Guest Posted August 28, 2022 Posted August 28, 2022 1 minute ago, jsb235 said: Are we counting all the games evenly or counting the game he did poorly in not as much? I want you to tell me the rating. According to the NFL it is 78.9. According to you it is 70.5. Which is it ??
jsb235 Posted August 28, 2022 Author Posted August 28, 2022 1 minute ago, downundermike said: In the Trey Lance example, you are counting in my games differently. His qb rating is 97.3, your method says it is 104.65, as you are giving to much weight to his game with a 127.1 rating. Honestly, I am not paying much attention to you. There is only one issue - do all games count the same or not. That's it.
Guest Posted August 28, 2022 Posted August 28, 2022 1 minute ago, jsb235 said: Honestly, I am not paying much attention to you. There is only one issue - do all games count the same or not. That's it. I am going to keep pounding this thread with examples showing your method of calculating QB rating does not align with every single NFL statistic site. Even people who don’t like me on this board, who I have had knock down drag outs with have said you are wrong. You will not find one person to back up your false notion that adding rating and dividing by games played is accurate.
Guest Posted August 28, 2022 Posted August 28, 2022 15 hours ago, jsb235 said: Good luck. You need it. Nope, the facts is all I needed.
ToastJenkins Posted August 28, 2022 Posted August 28, 2022 7 minutes ago, jsb235 said: Honestly, I am not paying much attention to you. There is only one issue - do all games count the same or not. That's it. No the issue seems to be you dont understand stats and cant admit it 1 1
jsb235 Posted August 28, 2022 Author Posted August 28, 2022 2 minutes ago, downundermike said: I am going to keep pounding this thread with examples showing your method of calculating QB rating does not align with every single NFL statistic site. Even people who don’t like me on this board, who I have had knock down drag outs with have said you are wrong. You will not find one person to back up your false notion that adding rating and dividing by games played is accurate. Have fun. But until you address the basic question of whether all games should or should not count equally you are wasting your time. Just say no they shouldn't. It's what you believe, so why not say it?
Guest Posted August 28, 2022 Posted August 28, 2022 1 minute ago, jsb235 said: Have fun. But until you address the basic question of whether all games should or should not count equally you are wasting your time. Just say no they shouldn't. It's what you believe, so why not say it? I counted the 4 starts of Jalen Hurts equally and determined his QB rating is 78.9. You are not and saying it is 70.5. Why you doing Jalen dirty like that.
Guest Posted August 28, 2022 Posted August 28, 2022 14 minutes ago, jsb235 said: Have fun. But until you address the basic question of whether all games should or should not count equally you are wasting your time. Just say no they shouldn't. It's what you believe, so why not say it?
brkmsn Posted August 28, 2022 Posted August 28, 2022 57 minutes ago, downundermike said: That is doing a huge disservice to Jalen Hurts. You are taking away 8.4 of his actual QB rating because you are wrong. Fact, Jalen Hurts QB rating for the last 4 games of 2020 is 78.9, that is 100% correct based on how the NFL calculates QB rating. @brkmsn, are you going to let jsb not give Jalen Hurts the appropriate credit for what he did at the end of 2020 ?? I have my own opinions and arguments. I'm sure you and I agree on some and I agree with other people on some others. I do believe that each game has it's own set of circumstances which can affect context in an argument. However, if you are going to string a set of games together, I believe, to be consistent you should use the standard practice of combining the stats for the stretch of games as you would for a season. Even then, stats don't tell the whole story, but they can help support an argument based on opinion. Honestly, I'm surprised this debate is still ongoing. 2
Guest Posted August 28, 2022 Posted August 28, 2022 7 minutes ago, brkmsn said: However, if you are going to string a set of games together, I believe, to be consistent you should use the standard practice of combining the stats for the stretch of games as you would for a season. Thank you
jsb235 Posted August 28, 2022 Author Posted August 28, 2022 4 minutes ago, brkmsn said: I have my own opinions and arguments. I'm sure you and I agree on some and I agree with other people on some others. I do believe that each game has it's own set of circumstances which can affect context in an argument. However, if you are going to string a set of games together, I believe, to be consistent you should use the standard practice of combining the stats for the stretch of games as you would for a season. Even then, stats don't tell the whole story, but they can help support an argument based on opinion. Honestly, I'm surprised this debate is still ongoing. That's reasonable. But I think you can also agree that this would also count some games more than others, right? So while accounting for that would be a different way of doing the calculation, that doesn't make it wrong. There's no math principle being violated. It's just a different way of looking at the data that corrects for an obvious flaw.
Next_Up Posted August 28, 2022 Posted August 28, 2022 13 minutes ago, jsb235 said: That's reasonable. But I think you can also agree that this would also count some games more than others, right? So while accounting for that would be a different way of doing the calculation, that doesn't make it wrong. There's no math principle being violated. It's just a different way of looking at the data that corrects for an obvious flaw. You might want to read up on both linear independence and statistical independence in terms of using statistics in a rigorous way. By grouping games that are "similar" you are invoking a kind of set dependence between them. If each game is a unique and independent event, then statistical analysis is easier, otherwise you need to account for the dependencies that result in weighting some events in a different way than others. So, maybe not violating a math principle but its sloppy analysis on a purely statistical basis. IMO
brkmsn Posted August 28, 2022 Posted August 28, 2022 18 minutes ago, jsb235 said: That's reasonable. But I think you can also agree that this would also count some games more than others, right? So while accounting for that would be a different way of doing the calculation, that doesn't make it wrong. There's no math principle being violated. It's just a different way of looking at the data that corrects for an obvious flaw. It is a different way of doing it, but what if we did the same thing with rushing stats? Let's say a RB in a 3 game stretch has: 20 - 100 (5.0) average 20 - 80 (4.0) 30 - 180 (6.0) If you average the 3 averages together, his average average was 5.0 yards per carry during that stretch. But his real average per carry during that stretch was 5.14 ypc.
jsb235 Posted August 28, 2022 Author Posted August 28, 2022 16 minutes ago, brkmsn said: It is a different way of doing it, but what if we did the same thing with rushing stats? Let's say a RB in a 3 game stretch has: 20 - 100 (5.0) average 20 - 80 (4.0) 30 - 180 (6.0) If you average the 3 averages together, his average average was 5.0 yards per carry during that stretch. But his real average per carry during that stretch was 5.14 ypc. Passer rating is not an average. It is a measure of performance. If the rb was getting credit for tds, dinged for fumbles, and other factors were being considered, and then they were getting a performance score, you could average them, right?
jsb235 Posted August 28, 2022 Author Posted August 28, 2022 26 minutes ago, Next_Up said: Otherwise you need to account for the dependencies that result in weighting some events in a different way than others. So, maybe not violating a math principle but its sloppy analysis on a purely statistical basis. IMO That's what passer rating does, no? It assigns a performance score to account for those dependencies. But you do understand that the nfl uses raw data and not per-game data not because the math is flawed, but because there are too many times a qb gets hurt, or otherwise has games that are severe outliers that would corrupt the data, right? They have to account for dozens of players each season across multiple seasons, and figuring out how to count games with anomalies would be a nightmare. No such issue exists in examining the games Hurts was in, so I don't see why you would call that sloppy. It would be sloppy to do it over a wide range of data, but not six similar games.
Next_Up Posted August 28, 2022 Posted August 28, 2022 3 minutes ago, jsb235 said: That's what passer rating does, no? It assigns a performance score to account for those dependencies. But you do understand that the nfl uses raw data and not per-game data not because the math is flawed, but because there are too many times a qb gets hurt, or otherwise has games that are severe outliers that would corrupt the data, right? They have to account for dozens of players each season across multiple seasons, and figuring out how to count games with anomalies would be a nightmare. No such issue exists in examining the games Hurts was in, so I don't see why you would call that sloppy. It would be sloppy to do it over a wide range of data, but not six similar games. Actually, the entire thesis of statistics is to use as broad a data set as possible so that outliers are normed our and general trends can be determined. I was responding to your claim that no math theory was being violated and now I think you just don't understand statistical analysis. Why don't you just pick the data sets that make your point, do your "analysis" and leave it at that.
jsb235 Posted August 28, 2022 Author Posted August 28, 2022 5 minutes ago, Next_Up said: Actually, the entire thesis of statistics is to use as broad a data set as possible so that outliers are normed our and general trends can be determined. I was responding to your claim that no math theory was being violated and now I think you just don't understand statistical analysis. Why don't you just pick the data sets that make your point, do your "analysis" and leave it at that. So how do we norm out outliers in a data set of six games? Is it better to use raw data that clearly discounts some of those games or treat every game as equal, since we are comparing six similar events? Again, it comes down to the idea that for a six-game data set, should we be weighting those games differently? Yes or no.
Next_Up Posted August 28, 2022 Posted August 28, 2022 29 minutes ago, jsb235 said: So how do we norm out outliers in a data set of six games? Is it better to use raw data that clearly discounts some of those games or treat every game as equal, since we are comparing six similar events? Again, it comes down to the idea that for a six-game data set, should we be weighting those games differently? Yes or no. I am really not interested in teaching you statistics. I was simply responding to your premise that using less data makes something more statistically significant. It does not. See MLB batting averages if you want an introductory example.
Aerolithe_Lion Posted August 28, 2022 Posted August 28, 2022 1 hour ago, jsb235 said: Honestly, I am not paying much attention to you. Did you not just make an entire thread to call him out? 2
ToastJenkins Posted August 28, 2022 Posted August 28, 2022 1 hour ago, jsb235 said: That's reasonable. But I think you can also agree that this would also count some games more than others, right? So while accounting for that would be a different way of doing the calculation, that doesn't make it wrong. There's no math principle being violated. It's just a different way of looking at the data that corrects for an obvious flaw. No you are still wrong
Aerolithe_Lion Posted August 28, 2022 Posted August 28, 2022 37 minutes ago, jsb235 said: So how do we norm out outliers in a data set of six games? Is it better to use raw data that clearly discounts some of those games or treat every game as equal, since we are comparing six similar events? Again, it comes down to the idea that for a six-game data set, should we be weighting those games differently? Yes or no. Passer rating DOES act like an averaged stat though. 1 for 1 with a 90 yard TD pass is 158.3 10 for 10 with 900 yards and 10 TDs is 158.3 The only way to discern between the two is to weigh them differently when additional attempts are added. 1 more pass attempt should affect rating A far more than rating B. You do that by breaking it back down into the accumulation stats, added them up, and then return to the new rating. By the way you are attempting to use passer rating, a game where Jalen throws 11 passes and ruptures his Achilles in the second quarter is worth exactly the same as a game where he has 39 attempts in a monster come-from-behind overtime win. A 1 quarter performance equals a 5 quarter performance with your assessment. By blanketedly averaging out the 3 passer ratings without diving deeper into each individual game and weighing them differently, you’re intentionally removing context.
jsb235 Posted August 28, 2022 Author Posted August 28, 2022 6 minutes ago, Aerolithe_Lion said: By the way you are attempting to use passer rating, a game where Jalen throws 11 passes and ruptures his Achilles in the second quarter is worth exactly the same as a game where he has 39 attempts in a monster come-from-behind overtime win. A 1 quarter performance equals a 5 quarter performance with your assessment. I am specifically not doing that. In fact, i cited that as a reason why the nfl uses the formula it uses. But here is the issue. As mike pointed out, Hurts had an average rating of 71 in four games, two each against the Giants and WFT. But the raw data from those four games had him with a rating of 79. So the bad game against the giants counted for significantly less than his good one. Thevquestion becomes, is the 79 figure a more accurate representation of his performance over those four games, or is the 71 figure? Essentially, should the first game be weighed less? I don't see a good reason for a yes answer, but if anyone has one, i am all ears. 26 minutes ago, Aerolithe_Lion said: Did you not just make an entire thread to call him out? I made this thread to keep the discussion out of the blog.
jsb235 Posted August 28, 2022 Author Posted August 28, 2022 39 minutes ago, Next_Up said: I am really not interested in teaching you statistics. I was simply responding to your premise that using less data makes something more statistically significant. It does not. See MLB batting averages if you want an introductory example. I doubt you could. Your contribution so far has been pretty basic.
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now