Jump to content


2020 Quarterback Competition


Recommended Posts

Just now, Husker_Bohunk said:

And yet national polling companies use small sample sizes and are considered accurate. You can look that up for yourself if you'd like, I won't have time today. :)

I don't care what they do. Its basic statistics that a small sample size is likely to produce skewed results.

 

 https://sciencing.com/disadvantages-small-sample-size-8448532.html

And I can find you 100 other links just like that one that say the same thing. A small sample size is unreliable plain and simple.

  • Plus1 1
  • Haha 1
Link to comment

14 minutes ago, Husker_Bohunk said:

100? Lol list them.

It's called exaggeration but here are a few more 

 

https://garstats.wordpress.com/2017/02/04/small-sample-sizes/

 

https://eiko-fried.com/small-samples-can-be-inherently-problematic/

 

https://select-statistics.co.uk/blog/importance-effect-sample-size/

 

There are more but I'm not gonna waste my time. Again, its basic statistics. 

Link to comment
5 hours ago, Nebfanatic said:

It's called exaggeration but here are a few more 

 

https://garstats.wordpress.com/2017/02/04/small-sample-sizes/

 

https://eiko-fried.com/small-samples-can-be-inherently-problematic/

 

https://select-statistics.co.uk/blog/importance-effect-sample-size/

 

There are more but I'm not gonna waste my time. Again, its basic statistics. 

It actually depends on how good the sample is along with data size.  Data size alone will not produce good results.  Garbage in/garbage out.   Pharma companies will do approval studies with 1000 patients and get results that mimic the use after approval on 1 million patients.  

 

Poll 2000 registers voters and the data will be worse than polling 500 likely voters that voted in the previous election. 

 

Its basic statistics!

  • Plus1 2
Link to comment
2 minutes ago, Archy1221 said:

It actually depends on how good the sample is along with data size.  Data size alone will not produce good results.  Garbage in/garbage out.   Pharma companies will do approval studies with 1000 patients and get results that mimic the use after approval on 1 million patients.  

 

Poll 2000 registers voters and the data will be worse than polling 500 likely voters that voted in the previous election. 

 

Its basic statistics!

 

Different statistics stabilize at different numbers, and QBR does not stabilize at 30 or so snaps from the original "McCaffrey would have an 85+ QBR with 95% certainty!" post. 30 is the minimum mentioned in basic statistics, but it depends on what is being measured. In the case of QBR (itself an imperfect measure of QB play), 30 snaps is not nearly enough to draw a conclusion.

 

The bold part isn't really about sample size, it's about choosing the right sample.

  • Plus1 2
Link to comment
5 minutes ago, Husker in WI said:

 

Different statistics stabilize at different numbers, and QBR does not stabilize at 30 or so snaps from the original "McCaffrey would have an 85+ QBR with 95% certainty!" post. 30 is the minimum mentioned in basic statistics, but it depends on what is being measured. In the case of QBR (itself an imperfect measure of QB play), 30 snaps is not nearly enough to draw a conclusion.

 

The bold part isn't really about sample size, it's about choosing the right sample.

Correct.  The person I was replying to said sample size is what matters.  I argue sample size, and data input relative to what you are looking at both matter.  

  • Plus1 2
Link to comment

On 12/1/2019 at 6:18 AM, ZMagers22 said:

QBR leaders for 2019 according to ESPN -


1. Tagovailoa - 94.4

2. Fields - 93.2

3. Burrow - 92.9

4. Hurts - 92.1 

5. Huntley (Utah) - 88.7

6. Morgan (Minn) - 84.3

 

71. Martinez - 58.2
 

McCaffery didn’t have enough games to show up on the list, but he did have enough plays (30+) to have 95% confidence that his QBR would have been roughly the same throughout the season. Luke has an 89.2 QBR

 

As with any set of statistics, there is variability and correlation components that need to be taken into consideration. However; Central Limit Theorem, The Law of Large Numbers and Standard Error suggest that Luke’s sample size is enough for a comparison. Therefore:

 

1. Tagovailoa - 94.4

2. Fields - 93.2

3. Burrow - 92.9

4. Hurts - 92.1 

5. McCaffery - 89.2

6. Huntley (Utah) - 88.7

7. Morgan (Minn) - 84.3


72. Martinez - 58.2

 

Based on inferential statistical analysis, Luke won By a landslide this year (5% error allowed). I’ll leave the eye test arguments to everyone else.

 

 

The Central Limit Theorem requires the sample to be random and it’s not.

 

What statistic would your confidence interval be on? Where would you get your standard deviation? I don’t think it would be on QBR; QBR is calculated from multiple plays.

  • Thanks 1
Link to comment

Just to add to the above now that I'm home, QBR is based on passing stats, so I'm not sure why people are talking about 30 plays unless I'm missing something. McCaffrey has 12 passes.

 

Still dunno how you'd make a confidence interval on QBR for 1 QB, unless you did it by game and then the sample size for McCaffrey would be 3, but if we look at completion % using a Wilson interval, we can be 95% confident McCaffrey would complete between 46.8% and 91.1% of his passes. The ginormous margin of error is due to the tiny sample size.

 

Now if the sample size was 52 instead of 12, and he completed 39 of 52 (still 75%) passes, we could be 95% confident he'd complete between 61.8% and 84.8% of his passes.

  • Plus1 2
Link to comment
43 minutes ago, Moiraine said:

Just to add to the above now that I'm home, QBR is based on passing stats, so I'm not sure why people are talking about 30 plays unless I'm missing something. McCaffrey has 12 passes.

 

Still dunno how you'd make a confidence interval on QBR for 1 QB, unless you did it by game and then the sample size for McCaffrey would be 3, but if we look at completion % using a Wilson interval, we can be 95% confident McCaffrey would complete between 46.8% and 91.1% of his passes. The ginormous margin of error is due to the tiny sample size.

 

Now if the sample size was 52 instead of 12, and he completed 39 of 52 (still 75%) passes, we could be 95% confident he'd complete between 61.8% and 84.8% of his passes.

Yeah, it's hard to tell how McCaffrey is at passing. He's had some real bad passes, so he will need to work on that in the off season. He's definitely a great option QB. 

 

I am still interested to know whether Martinez is injured. He hasn't looked right running or passing the ball for a while now. The good news is that the QB position has some depth now. I just hope whoever gets beat in the spring/fall for the number one spot sticks around. I can't remember the last season Nebraska has made it through the season without the starting QB getting injured. 

Link to comment

We seem to have a leadership problem with Martinez at QB.  There's a lot of writing out there about what 'makes a great leader'.  They all tend to say the same thing.  Good leaders are:

 

      1. Effective problem-solvers

      2. Results-driven

      3. Supportive of others

      4. Encouraging of ideas

      5. Champions of change

 

So how does Martinez look when we go down the list?   He doesn't seem to attract a lot of teammates around him on the sideline like we've seen with McCaffrey or Vedral after a scoring drive.  Superficially it suggests there's a disconnect with the team.  And when Vedral or McCaffrey step into the huddle the team looks excited.  With the social media swirling with JD leaving rumors, it suggests a certain level of players resistant to the way the team is being led on the offensive unit.

Link to comment
15 hours ago, Moiraine said:

 

 

The Central Limit Theorem requires the sample to be random and it’s not.

 

What statistic would your confidence interval be on? Where would you get your standard deviation? I don’t think it would be on QBR; QBR is calculated from multiple plays.

So if you had n as the total number of samples and uses a randomizer to select  up to n/10 capped at no more than 10% from that group, and ran that routine ten times then averaged out the results, would that be a fair assessment?  I have pretty high confidence it would look favorable in ways that would suggest Martinez is not the best option.  All those bad plays would catch up to him, because they happen way too often.  His downside is killing us and his upside is often too little too late.

  • Thanks 1
Link to comment

1 hour ago, madrat said:

We seem to have a leadership problem with Martinez at QB.  There's a lot of writing out there about what 'makes a great leader'.  They all tend to say the same thing.  Good leaders are:

 

      1. Effective problem-solvers

      2. Results-driven

      3. Supportive of others

      4. Encouraging of ideas

      5. Champions of change

 

So how does Martinez look when we go down the list?   He doesn't seem to attract a lot of teammates around him on the sideline like we've seen with McCaffrey or Vedral after a scoring drive.  Superficially it suggests there's a disconnect with the team.  And when Vedral or McCaffrey step into the huddle the team looks excited.  With the social media swirling with JD leaving rumors, it suggests a certain level of players resistant to the way the team is being led on the offensive unit.

We're going full sideline psychologist, huh? These are all huge stretches IMO.

1 hour ago, madrat said:

So if you had n as the total number of samples and uses a randomizer to select  up to n/10 capped at no more than 10% from that group, and ran that routine ten times then averaged out the results, would that be a fair assessment?  I have pretty high confidence it would look favorable in ways that would suggest Martinez is not the best option.  All those bad plays would catch up to him, because they happen way too often.  His downside is killing us and his upside is often too little too late.

 

And there's your problem - despite having almost no data, you have high confidence in something that if true, would mean the coaches are either stupid or knowingly sabotaging the team's success. I know there are people who believe the coaches are sticking with Martinez when they shouldn't be, but I see zero reason they would do that. So maybe instead of being very confident in something with no data to back it up, the answer is the coaches have more data and Martinez was in fact the best option. Just maybe.

  • Plus1 4
Link to comment
7 minutes ago, suh_fan93 said:

 

Well he only had 3 incompletions but those 3 must have been really bad apparently.

 

 

Not all completions are good passes, just like not all incompletions are bad passes.  There are many who still question McCaffrey's passing ability, and it's fair.  We haven't seen enough of him in a wide variety of passes to make judgement on his passing ability.

Link to comment

My biggest issue with Martinez is he isn't making his reads properly - including in the run game. There are countless times he makes the wrong read to give or pull. Albeit, we don't know if its an actual zone read or a straight zone play. However, I know I annoy those I watch with by keeping a tally of bad reads in the running game. 

  • Plus1 1
  • Fire 1
Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Recently Browsing   0 members

    • No registered users viewing this page.

Visit the Sports Illustrated Husker site



×
×
  • Create New...