Message Topic

ken_admin (1459 posts)  8/29 10:04 AM

Along with our previous standings, you can also view a new "Advanced Standings".  These standings will use the probability that you'll beat another team to determine what percent of 1 point you receive.  For example, previously, if you had higher points than a team, you would receive 1 point.  Now, we calculate the probability you'll beat that team so say 70%.  In this case, you receive 0.7 points, and the other team receives 0.3 points.

For now, this is available under Analysis Monster and Trade Monster.  In the next week or two, this will be added to the Draft Tracker after we get some feedback.  Also, we consider this "beta" since tweaks will likely be made over the next couple of weeks.

When you check "NEW: Show Advanced Standings (beta)", you will see a second set of columns for the Advanced numbers.  This applies to Roto Standings and H2H Standings.

If you have any questions or feedback, please post those here.

davewhale (14 posts)  8/29 10:27 AM

This is such a great idea, thanks!

zass17  (5 posts)  8/29 10:56 AM

do you use a normal distribution assumption?

ken_admin (1459 posts)  8/29 11:00 AM

Yes, we're using normal distributions along with stdev to determine the probabilities.  This is calculated for each player so your rosters will affect the probability range.

wyl11wyl (1 posts)  8/31 7:15 PM

None of the category is normal distrbuted though... 

qc0114  (2 posts)  9/28 9:46 PM

seems the new draft tracker vs the old one will give me differnt ranking in the same player? or just i misread

ken_admin (1459 posts)  9/28 10:14 PM

If you let me know which league, I can check.

imann871  (14 posts)  9/29 11:56 AM

Yeah, looks like rankings differ from each other..?

ken_admin (1459 posts)  9/29 12:04 PM

Do you have an example league I can check?

awaaz116 (1 posts)  10/1 8:28 PM

Will this only factor in to category leagues? How if any does this affect points league projections? 

ken_admin (1459 posts)  10/1 8:35 PM

Yes, it only applies to category leagues standings.

Henry24  (237 posts)  10/6 5:19 PM

This tool works great. A big improvement over the previous draft tracker. 

saras (1 posts)  10/8 2:35 PM

Why are there differences in standings and categorical scarcity on the new and old draft tracker?

ken_admin (1459 posts)  10/8 2:49 PM

I would expect them to be the same.  My only thought is maybe the value settings are different?

lilbtaskforce1  (2 posts)  10/10 4:04 PM

Is there / will there be a target player filter like on the previous draft tracker?

ken_admin (1459 posts)  10/11 8:34 AM

We still hope to add a version of this.  Would something simple work for now where you highlight players you're targetting, and we color them different?  In other words, not include 'analyze including targets' and a few other features.

lilbtaskforce1  (2 posts)  10/11 12:27 PM

Yep that would be perfect, thanks!

Bell8865 (1 posts)  10/13 11:37 AM

Can this be used for Roto?  Or is it more for Head to Head?

ken_admin (1459 posts)  10/13 11:42 AM

it's just as useful for both.  The Roto Standings will also reflect the probability of beating each team in each category.

dderenio (17 posts)  10/14 11:26 AM

+1 for the target player in the draft tracker

ken_admin (1459 posts)  10/14 12:11 PM

I would expect them to be the same with the same settings.  Which league is this?

ATaylor109  (1 posts)  10/21 12:53 PM

Does this work for auction?

ken_admin (1459 posts)  10/21 1:41 PM

Yes, this should be the same for auction.

#70264  (7 posts)  11/17 9:02 PM

Hey @ken_admin, couple questions for you:

1. Can you share some more detail on how the distributions are calculated for each player? For example - what sample are you using for a given player?  Is it just all games of the current season, are you using past seaons etc.  
2. Also I'm curious if the normal distribution assumption was validated.  It does seem reasonable that every player would vary normally around their averages but could imagine scenarios where this might not be the case - for examples players that average low blocks. 

Thanks!

ken_admin (1459 posts)  11/18 11:08 AM

For samples, we're using the past two season and not including players with too few minutes.  Those players get a default. 

For the probabilities, we calculate a probability based on the # of stdev from the average for the player.  I believe this is a reasonable estimate.