HomeLog inRegisterFAQSearch

Share | 
 

 Positional Consistency/PPG Data

View previous topic View next topic Go down 
Go to page : 1, 2  Next
AuthorMessage
Charm City Justice
 
 
avatar

Posts : 1304

PostSubject: Positional Consistency/PPG Data   Wed Jun 14, 2017 11:15 am

I was looking at the debate in the other thread about optimal starting lineup, consistency, etc., so being a nerd I decided to do a little digging into the data to see what the optimal lineup would be for IDP if you were looking for high PPG and consistency.  I took the top 12 scorers for each position and labeled them as DE1, DT1, LB1, CB1, S1.  The next 12 and labeled them as DE2, DT2, LB2, CB2, S2, etc. etc. up to the max number of starters for that position.  I then calculated the average ppg for each group and the standard deviation from week to week.  In my mind a higher standard deviation meant less consistent.  One problem with this theory is that it became pretty clear that this overvalued consistently mediocre players (as you can see, the CB4 group was the most consistent but 16th in PPG).  I am open to suggestions on how to correct for this to come up with some kind of rankings, but for now you can all just look at the data if you like and use it however you like.  This data set also assumes even distribution of players across teams, which obviously isn't the case, so optimal lineups will differ based upon roster construction.


Back to top Go down
Alaska Arsenic
 
 
avatar

Posts : 771

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 11:26 am

What if you dropped any outliers? Eliminate the top player and the 12th player of each grouping to see if the SD falls in line to show more consistency in your theory?


Back to top Go down
http://faketeams.com
Bergen Brawlers
 
 
avatar

Posts : 4923

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 11:42 am

So is the optimal lineup 1 DT, 1 DE, 4 LB, 1 CB, and 2 S?

I'm not surprised at the lower guys having more consistency -- they aren't scoring a lot of points so there aren't a lot of high scoring weeks.

I was really surprised to see LB3 being more consistent than LB1 or LB2. That might be caused by players becoming a LB1 or LB2 based on a few high-scoring weeks.

What would be interesting would be to look at the projections going into a season rather than end of season projections. We do have 2 redraft drafts for each of the last two years in the RWTC to use as a basis.


Back to top Go down
Nevada NightHawks
 
 
avatar

Posts : 2487

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 12:43 pm

This is cool stuff just to look at and compare.  Of course I started with Alec Ogletree. laughing  2nd ranked LB in PPG of the Top 12 LB's behind only Kwon Alexander.  In this top group, they have a deviation of 7.26 points.  Ogletree only has a deviation of only 5.33 which is the best out of the Top 5 and 2 PPG below the Top 12 average.  He has the 3rd lowest deviation of the Top 12 group.  So it's cool to compare to other players in the group. 

When I have more time, I'm gonna have to compare this to some of my other defensive players.  Nothing better than playing with Stats. vote yes


Back to top Go down
Charm City Justice
 
 
avatar

Posts : 1304

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 1:17 pm

Bergen Brawlers wrote:
So is the optimal lineup 1 DT, 1 DE, 4 LB, 1 CB, and 2 S?  

I'm not surprised at the lower guys having more consistency -- they aren't scoring a lot of points so there aren't a lot of high scoring weeks.

I was really surprised to see LB3 being more consistent than LB1 or LB2.  That might be caused by players becoming a LB1 or LB2 based on a few high-scoring weeks.

What would be interesting would be to look at the projections going into a season rather than end of season projections.  We do have 2 redraft drafts for each of the last two years in the RWTC to use as a basis.  

Your optimal lineup only has 9 starters, so I wouldn't suggest going with that but I definitely think you start with that base and then could make arguments for the final two positions. I think LB5 and S3 probably make the best argument based on the numbers (which Jon will appreciate since it is what he proposed as the optimal lineup.

Also, a lower standard deviation equals more consistent (I think), so actually LB1 and LB2 are more consistent that LB3 (assuming my calculations are correct).

I'm not quite sure I follow what you mean about looking at projections going into a season. Do you mean ranking the players in tiers based upon their draft position pre-season as opposed to their actual end of season output, but still using the end of season data? If so, that will be a project for another day, but something I could take a look at doing.

Alaska Arsenic wrote:
What if you dropped any outliers? Eliminate the top player and the 12th player of each grouping to see if the SD falls in line to show more consistency in your theory?

I can give that a shot, although I would imagine that this won't actually change the rankings too much. As Russ said, the lower scoring of the lower tiers inherently leads to a lower standard deviation because your range is much smaller. But, I could be wrong, so I will take a look.


Back to top Go down
Bergen Brawlers
 
 
avatar

Posts : 4923

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 1:28 pm

And that's what I get for typing at lunch. Ignore the first two points.

As for my last point, I think there is a difference in looking at numbers when we know how the players will end up performing and looking at them based on pre-season projections. I'm not asking you to do this, but I'd be curious how the numbers would look if we looked at projected LB1s, DE1s, CB2s, etc. How do they end up performing, and how consistent are they? I may try to tackle this later if I get some free time.

Finding rankings for our system, etc. would normally be tough but we do have redraft rankings (2 sets) for the past two years as a result of the RWTC draft. Now, we all aren't great drafters, but it is a decent gauge of consensus on rankings.

Knowing that I should start 5 LBs is one thing, knowing which 5 to start is another. It's one of the reasons that, in standard leagues, kickers and defenses have little value -- they are very inconsistent from year-to-year. So it's tough to know which 5 to pick.


Back to top Go down
Tenafly Vipers
 
 
avatar

Posts : 352

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 1:30 pm

This is great stuff! Thanks!


Back to top Go down
Charm City Justice
 
 
avatar

Posts : 1304

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 1:41 pm

Bergen Brawlers wrote:
And that's what I get for typing at lunch.  Ignore the first two points.  

As for my last point, I think there is a difference in looking at numbers when we know how the players will end up performing and looking at them based on pre-season projections.  I'm not asking you to do this, but I'd be curious how the numbers would look if we looked at projected LB1s, DE1s, CB2s, etc.  How do they end up performing, and how consistent are they?  I may try to tackle this later if I get some free time.

Finding rankings for our system, etc. would normally be tough but we do have redraft rankings (2 sets) for the past two years as a result of the RWTC draft.  Now, we all aren't great drafters, but it is a decent gauge of consensus on rankings.

Knowing that I should start 5 LBs is one thing, knowing which 5 to start is another.  It's one of the reasons that, in standard leagues, kickers and defenses have little value -- they are very inconsistent from year-to-year.  So it's tough to know which 5 to pick.

I will take a look at what I can do if work stays slow sometime in the next two days. There is currently a judicial conference going on so there are exactly 0 criminal courts open in my courthouse for M, T, and W of this week, but they reopen tomorrow so I would imagine that I may actually have work to do tomorrow.



Back to top Go down
Charm City Justice
 
 
avatar

Posts : 1304

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 2:21 pm

Made a new column by running the standard deviation as Ghoji recommended by eliminating outliers in each category. Not a ton of change. LB2 gets a nice bump (Vontaze Burfict was wreaking some havoc in this subset) and so does S3 (Jimmy Ward was the culprit here), but overall not a ton of change.


Back to top Go down
England Dragons
 
 
avatar

Posts : 8509

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 4:01 pm

Great stuff, Bill.

It kinda shows the position is pretty irrelevant once you get past the top 8 or so. I think the important column is this one:

Rank by PPG
Pos Average PPG
LB1 18.33
LB2 16.60
LB3 15.64
S1 15.62
DE1 15.00
DT1 14.24
CB1 13.85
S2 13.35

-------------

LB4 12.70
CB2 12.65
DE2 11.86
LB5 11.50
S3 11.49
CB3 11.48
DT2 11.31
CB4 10.58
S4 10.33
DE3 9.68
DE4 8.34
DT3 8.24

So in an ideal scenario (of course, completely dependent on your roster construction, the players you have, bye weeks, injuries, etc etc) you want DT, DE, LB, LB, LB, CB, S, S for sure. After that, doesn't really make much difference which position you fill out your starting line-up with. You go with your roster strength, BPA, match-ups. Whatever.

Very interesting. Thanks for the work you put in on this Bill.

cheers


Back to top Go down
Norfolk Bombers
 
 
avatar

Posts : 5644

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 4:17 pm

My first thought is to test using CIs, but then I realized that our N might be too small. To keep things simple I assumed an N of (12*16) 192 for each group. This ignores injuries, but fuck it.

DE1 1.2532542293 13.75 16.25
DE2 1.0439070217 10.82 12.90
DE3 0.9378189098 8.74 10.62
DE4 0.8175857162 7.52 9.16
DT1 1.1471661174 13.09 15.39
DT2 1.0071298096 10.30 12.32
DT3 0.8656789936 7.37 9.11
LB1 1.0269329238 17.30 19.36
LB2 1.0340054646 15.57 17.63
LB3 1.0594666115 14.58 16.70
LB4 0.9915702198 11.71 13.69
LB5 0.8713370263 10.63 12.37
CB1 1.1132179215 12.74 14.96
CB2 1.0721971849 11.58 13.72
CB3 0.9448914506 10.54 12.42
CB4 0.789295553 9.79 11.37
S1 1.0495650544 14.57 16.67
S2 0.9194303037 12.43 14.27
S3 0.9463059587 10.54 12.44
S4 0.7963680938 9.53 11.13


Back to top Go down
Bergen Brawlers
 
 
avatar

Posts : 4923

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 4:25 pm

I'm a bit surprised at how high scoring the CB2 and CB3 range is, both coming in above the LB5 in Dave's numbers (Bill and Jon have the CB3 around the same as the LB5).


Back to top Go down
Norfolk Bombers
 
 
avatar

Posts : 5644

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 4:36 pm

Table above sucks, bit what it shows is the CI for each position with an assumed N of 192. Would mean a lot more if we could get the N up, maybe use more than one year.

Here is a graph -

https://i58.servimg.com/u/f58/19/29/64/03/cifori10.jpg

If a positions Low is above another positions High, than that position is considered to be significantly higher of a scorer.


Back to top Go down
Philadelphia Pigskins
 
 
avatar

Posts : 6467

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 4:37 pm



Back to top Go down
Norfolk Bombers
 
 
avatar

Posts : 5644

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 4:40 pm

Bergen Brawlers wrote:
I'm a bit surprised at how high scoring the CB2 and CB3 range is, both coming in above the LB5 in Dave's numbers (Bill and Jon have the CB3 around the same as the LB5).  

The greater SD caused a larger range.


shrug


Back to top Go down
Charm City Justice
 
 
avatar

Posts : 1304

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 4:42 pm

Philadelphia Pigskins wrote:

And you are teaching the children of America. SMH


Back to top Go down
Philadelphia Pigskins
 
 
avatar

Posts : 6467

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 4:47 pm

Charm City Justice wrote:
Philadelphia Pigskins wrote:

And you are teaching the children of America. SMH

Yours mostly made sense.  Dave's though...



Back to top Go down
Riverside Rottweilers
 
 
avatar

Posts : 290

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 5:00 pm

Bergen Brawlers wrote:
And that's what I get for typing at lunch.  Ignore the first two points.  

As for my last point, I think there is a difference in looking at numbers when we know how the players will end up performing and looking at them based on pre-season projections.  I'm not asking you to do this, but I'd be curious how the numbers would look if we looked at projected LB1s, DE1s, CB2s, etc.  How do they end up performing, and how consistent are they?  I may try to tackle this later if I get some free time.

Finding rankings for our system, etc. would normally be tough but we do have redraft rankings (2 sets) for the past two years as a result of the RWTC draft.  Now, we all aren't great drafters, but it is a decent gauge of consensus on rankings.

Knowing that I should start 5 LBs is one thing, knowing which 5 to start is another.  It's one of the reasons that, in standard leagues, kickers and defenses have little value -- they are very inconsistent from year-to-year.  So it's tough to know which 5 to pick.

Another way to do this that doesn't involve rankings is to look at the end of year scoring over multiple years.

For example, the top 5 WRs have been pretty much the same over the past 3 years, but the top 5 DBs have varied dramatically. So it's easier to predict a WRs performance (and safer to draft) than a DB. You can simply track each player's scoring over multiple years and calculate average, deviation, etc.

Only problem with this method is that any drops/spikes in a player's value from a QB change, move to a different team, teammate injury (allowing player to get more touches) will increase a player's deviation, whereas early season projections can often account for these changes.


Back to top Go down
Mohawk Ridge Marauders
 
 
avatar

Posts : 19933

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 5:00 pm

Why is this thread that isn't about pictures of Andy getting so much attention?


Back to top Go down
Norfolk Bombers
 
 
avatar

Posts : 5644

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 5:04 pm

baghead2

Sorry, this thread got me all geeky. I love this stuff.


Back to top Go down
Bergen Brawlers
 
 
avatar

Posts : 4923

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 5:28 pm

So I got a little motivated after reading this and started putting data into Excel spreadsheets. Then couldn't figure out how to run formulas across multiple rows at one time. The one thing that I did learn, though, is the RWTC, as a group, sucks at picking DTs.


Back to top Go down
Charm City Justice
 
 
avatar

Posts : 1304

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 5:56 pm

Bergen Brawlers wrote:
So I got a little motivated after reading this and started putting data into Excel spreadsheets.  Then couldn't figure out how to run formulas across multiple rows at one time.  The one thing that I did learn, though, is the RWTC, as a group, sucks at picking DTs.

I think you just need to type the formula in once, then on the lower right hand corner of that box, hover, click and drag the square down however many rows you want to copy the formula to. If easier and you want to share the data you have collected this far I can take a look at the formulas tomorrow.


Back to top Go down
Bergen Brawlers
 
 
avatar

Posts : 4923

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 6:35 pm

I still have to add safeties and cornerbacks but below is a link for the data for the RWTC drafts:

https://docs.google.com/spreadsheets/d/12H8xCqsxU33YnQafNG0fnQGMb2yPX-vqXQndFTcMQI8/edit?usp=sharing

https://drive.google.com/file/d/0B5GctxiM4XxHdnRQUEJYck9ObkU/view?usp=sharing




Last edited by Bergen Brawlers on Thu Jun 15, 2017 9:40 am; edited 1 time in total
Back to top Go down
California Nightmare
 
 
avatar

Posts : 2379

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 6:39 pm

Can anyone put this in GED level format please.


Back to top Go down
Maplewood Doppelgangers
 
 
avatar

Posts : 2004

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 7:09 pm

California Nightmare wrote:
Can anyone put this in GED level format please.
For real. All those numbers... stars


Back to top Go down
Philadelphia Pigskins
 
 
avatar

Posts : 6467

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 7:15 pm

Maplewood Doppelgangers wrote:
California Nightmare wrote:
Can anyone put this in GED level format please.
For real. All those numbers... stars



Back to top Go down
Philadelphia Pigskins
 
 
avatar

Posts : 6467

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 7:28 pm

I wonder how it would come out if we approached it the way lineup requirements are structured. As in having the DT, DE, LB, CB, S, and then combining the player pool to see how the flexes play out.

Or maybe someone already did that and I don't understand. My classroom is about a million degrees, so my brain might be melting


Back to top Go down
Norfolk Bombers
 
 
avatar

Posts : 5644

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 7:36 pm

Bill calculated a average and standard deviation (SD) of each ranked position. SD talks the the amount of variation hidden in that mean value. For example-

11,12,9,and 8 have a mean of 10. 15,16, 4, and 5 also have mean of 10. The first group has a smaller variation and would also have a smaller SD. Make sense?

From there I calculated a confidence interval(CI) to help predict where the true mean of population of each position was. This assumes that our sample is a representative sample (which it very well might not be). It also seems week do to a low N(number of values in our sample). The larger the N we get the more confidence we have in our prediction. I would prefer to have a much larger N. The CI is a range of a low value and a high value. I used a Z-score (dont ask) that represents 95% confidence for a two tailed test (low and high). If there s overlap between CI's than one cannot say with confidence that their real value is different. If there is no overlap than one can say they are different. Make sense?


Back to top Go down
Kane County Goon Squad
 
 
avatar

Posts : 7472

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 7:42 pm

Maplewood Doppelgangers wrote:
California Nightmare wrote:
Can anyone put this in GED level format please.
For real. All those numbers... stars

fo reelz

dizzy


Back to top Go down
Riverside Rottweilers
 
 
avatar

Posts : 290

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 7:58 pm



These are the correlations for each data set. The closer to -1, the better the pre-draft rankings were at predicting fantasy output for that season. As a whole, seems like the rankings were pretty decent (correlations of roughly -0.7 for all positions).

I then did correlations for only the top 10 players, as I was worried that outliers in amongst the later draft picks could be screwing the correlation. Surprised that the correlations actually dropped (some even went positive) for this data, which could either be attributed to bad pre-draft rankings or more likely parity at the top of each position.


Back to top Go down
Riverside Rottweilers
 
 
avatar

Posts : 290

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 8:06 pm

Norfolk Bombers wrote:
Bill calculated a average and standard deviation (SD) of each ranked position.  SD talks the the amount of variation hidden in that mean value.  For example-

11,12,9,and 8 have a mean of 10.  15,16, 4, and 5 also have  mean of 10.  The first group has a smaller variation and would also have a smaller SD.  Make sense?

From there I calculated a confidence interval(CI) to help predict where the true mean of population of each position was.  This assumes that our sample is a representative sample (which it very well might not be).  It also seems week do to a low N(number of values in our sample).  The larger the N we get the more confidence we have in our prediction.  I would prefer to have a much larger N.  The CI is a range of a low value and a high value.  I used a Z-score (dont ask) that represents 95% confidence for a two tailed test (low and high).  If there s overlap between CI's than one cannot say with confidence that their real value is different.  If there is no overlap than one can say they are different.  Make sense?

What were your final confidence intervals? Also how bad were the SDs. Just by eyeballing the data it seems like the variance should be higher for DE than for the other two.


Back to top Go down
Maplewood Doppelgangers
 
 
avatar

Posts : 2004

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 8:12 pm

I'm happy you nerds find enjoyment out of this... I'll stick to history and literature.


Back to top Go down
Haddonfield Slashers
 
 
avatar

Posts : 192

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 8:16 pm



Back to top Go down
Bergen Brawlers
 
 
avatar

Posts : 4923

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 8:21 pm

One thing that would be interesting, and I will try it once I get the data entered from the RWTC, is to plot the best fit curve (or whatever you stats guys call it) to see how the positions plot out. I'm thinking that the LB position plateaus after the first 30 or so guys at a higher level than other positions and that's where starting multiple LBs would be best, in theory.


Back to top Go down
Bergen Brawlers
 
 
avatar

Posts : 4923

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 8:24 pm

Now, with that being said, going off draft position introduces a ton of variables, such as injuries, bad drafting, drafting for a particular need, etc. But I think that 2 years and 2 conferences and melding a curve to the data should provide data that has some useful benefit -- in part because the unknown variables can occur at any point in the draft.

The one data point that I would take out is the drafting of Ian Williams in both conferences last year because he was on IR when drafted. You could say the same about guys like Watt and Levy but that appears to be more of the natural randomness of the game.


Back to top Go down
Titletown Tyrants
 
 
avatar

Posts : 9593

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 8:29 pm

I would share my secrets, but I don't want any of you to actually do well in IDP for once.


Back to top Go down
Riverside Rottweilers
 
 
avatar

Posts : 290

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 9:03 pm

Bergen Brawlers wrote:
One thing that would be interesting, and I will try it once I get the data entered from the RWTC, is to plot the best fit curve (or whatever you stats guys call it) to see how the positions plot out.  I'm thinking that the LB position plateaus after the first 30 or so guys at a higher level than other positions and that's where starting multiple LBs would be best, in theory.  

To analyze the depth of each position (and hopefully determine an optimal starting lineup), I think it might be better to plot curves on the sorted data of all players (using each player's total fantasy points for the year). This will take out the confounding variables of draft position, team needs, etc. so you can simply look at where the plateaus are for each position.

The curves are a good way to visualize value vs. depth at each position and possibly see where the upper tier for each position ends, although variance from year to year will not be accounted for unless you use data from multiple years and remove the outliers.


Back to top Go down
Norfolk Bombers
 
 
avatar

Posts : 5644

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 9:28 pm

Riverside Rottweilers wrote:
Norfolk Bombers wrote:
Bill calculated a average and standard deviation (SD) of each ranked position.  SD talks the the amount of variation hidden in that mean value.  For example-

11,12,9,and 8 have a mean of 10.  15,16, 4, and 5 also have  mean of 10.  The first group has a smaller variation and would also have a smaller SD.  Make sense?

From there I calculated a confidence interval(CI) to help predict where the true mean of population of each position was.  This assumes that our sample is a representative sample (which it very well might not be).  It also seems week do to a low N(number of values in our sample).  The larger the N we get the more confidence we have in our prediction.  I would prefer to have a much larger N.  The CI is a range of a low value and a high value.  I used a Z-score (dont ask) that represents 95% confidence for a two tailed test (low and high).  If there s overlap between CI's than one cannot say with confidence that their real value is different.  If there is no overlap than one can say they are different.  Make sense?

What were your final confidence intervals? Also how bad were the SDs. Just by eyeballing the data it seems like the variance should be higher for DE than for the other two.

I used the SDs calculated by Bill. My CIs are in my post at 6:04pm in this thread.



Back to top Go down
Norfolk Bombers
 
 
avatar

Posts : 5644

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 9:29 pm

Haddonfield Slashers wrote:

Not formulating strategy, just having fun looking at numbers.


cheers





Back to top Go down
Norfolk Bombers
 
 
avatar

Posts : 5644

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 9:30 pm

Maplewood Doppelgangers wrote:
I'm happy you nerds find enjoyment out of this... I'll stick to history and literature.

I love history too, but Lit is fucking gross.....


sick sick sick



Back to top Go down
Philadelphia Pigskins
 
 
avatar

Posts : 6467

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 9:32 pm



Back to top Go down
Maplewood Doppelgangers
 
 
avatar

Posts : 2004

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 9:46 pm

Norfolk Bombers wrote:
Maplewood Doppelgangers wrote:
I'm happy you nerds find enjoyment out of this... I'll stick to history and literature.

I love history too, but Lit is fucking gross.....


sick  sick  sick

I'd read all day if I could!


Back to top Go down
Kane County Goon Squad
 
 
avatar

Posts : 7472

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 10:00 pm

Philadelphia Pigskins wrote:

Laughing Laughing Laughing


Back to top Go down
Tenafly Vipers
 
 
avatar

Posts : 352

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 10:27 pm

When y'all do all this work, do you enter the numbers by hand, or do you have some sort of script that scrapes the data for you?


Back to top Go down
Maplewood Doppelgangers
 
 
avatar

Posts : 2004

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 10:29 pm

Tenafly Vipers wrote:
When y'all do all this work, do you enter the numbers by hand, or do you have some sort of script that scrapes the data for you?
I was wondering the same thing. Excel is my nightmare.


Back to top Go down
Bergen Brawlers
 
 
avatar

Posts : 4923

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 10:44 pm

Tenafly Vipers wrote:
When y'all do all this work, do you enter the numbers by hand, or do you have some sort of script that scrapes the data for you?

Unfortunately no script for me. But I do have two monitors, which makes it a lot easier.


Back to top Go down
Denver Demons
 
 
avatar

Posts : 520

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 11:13 pm

Wew. Y'all took this further than I expected it to go in one day. I looked at it briefly at work, copied the format for a startup I have going on right now, then went to a baseball game. Now we have CI's and z-scores? I love it! It's been years since I bothered with statistics but it always got me a little excited. grin

I can't wait to play with this tomorrow when I have my computer in front of me. Maybe I'll remember enough to contribute here...maybe.

Excel is absolutely amazing once you screw around with it enough to figure it out.


Back to top Go down
England Dragons
 
 
avatar

Posts : 8509

PostSubject: Re: Positional Consistency/PPG Data   Wed Jun 14, 2017 11:53 pm

This is a great read - well done most of you!

laughing


Back to top Go down
Charm City Justice
 
 
avatar

Posts : 1304

PostSubject: Re: Positional Consistency/PPG Data   Thu Jun 15, 2017 6:47 am

Tenafly Vipers wrote:
When y'all do all this work, do you enter the numbers by hand, or do you have some sort of script that scrapes the data for you?

I copied and pasted off of MFL after I sorted. It took some editing after doing that to get rid of unnecessary columns, etc. but once I got into a pattern, getting the data into the spreadsheet wasn't too bad. The SD and average columns all come from formulas that are easily spread among different cells so that part isn't hard at all.

Also, we have officially reached the breaking point on my statistics knowledge in some of these posts. I pretty much understand what Dave did and appreciate him taking my data points to the next step. If it goes any further I will probably become lost though. I will try to look at Russ' data and see what I can make of it later today.

As for strategy, numbers can only take you but so far. I don't think anything we revealed today was actually groundbreaking, nor do I think that it will change anyone's strategy. It really just started out as me being really bored during a slow time at work and trying to provide some fuel to the debate in the other thread about optimal starting lineup configuration.

Also, yes I am a nerd, I do not deny it.


Back to top Go down
Rocky Mountain Oysters
 
 
avatar

Posts : 2880

PostSubject: Re: Positional Consistency/PPG Data   Thu Jun 15, 2017 8:32 am



Back to top Go down
Online
Sponsored content




PostSubject: Re: Positional Consistency/PPG Data   

Back to top Go down
 
Positional Consistency/PPG Data
View previous topic View next topic Back to top 
Page 1 of 2Go to page : 1, 2  Next

Permissions in this forum:You cannot reply to topics in this forum
Roster Wars :: The Roster Wars Clubhouse :: Main RW Lobby-
Jump to: