spacer
netfoos fooball statistics
spacer
spacer

Is the Generate Draw really random??

Discussion, suggestions and help regarding NetFoos Tournament Software, Online Statistics and Local Foosball Forums.

Postby stevieb on Wed Feb 03, 2010 10:53 pm

Just out of curiousity - we are experiencing something at our DYPs that seems to be a little bit odd.

On average we get anywhere from 5-7 teams - so out of the 10-14 people I would guess that approximately 10 people a week are here everyweek.

What we are seeing is - the same teams get drawn together seemingly too often for it to a coincedence or truly random.

Not sure if anyone else is experiencing this - but it really is quite often.

any input on this?

steve bell.
stevieb
 
Posts: 10
Joined: Sat Jan 05, 2008 8:26 pm

Postby Mark on Thu Feb 04, 2010 3:29 pm

Hi Steve,

I just posted an admin module that will allow you to view how many times players have partnered up in events in relation to each other. Click on the "Admin Modules" link in the NetFoos Admin, then click on "Partner_Summary_1_0". From there just follow the on screen instructions.

Also, I think Grant was going to post some data about this as well in the next day or so.

If you have any questions about the module please let me know.

Thanks,
Mark
User avatar
Mark
Site Admin
 
Posts: 759
Joined: Wed Jun 01, 2005 11:56 am
Location: Boca Raton, Florida

Postby stevieb on Thu Feb 04, 2010 6:45 pm

Thanks Mark. That admin module is pretty cool. I just spent some time analyzing some of our DYPs.

I guess the thing I am struggling with is - for example - when I looked at my stats, for our DYPs with a minimum participation of 20 weeks - I have someone that comes up at 18% and many others that are 8% or lower.

I am certainly no statistician - and I could logically see how this could go either way. But - the question is - ... is this really random? I am sure that true randomness is a factor of sample size - and I have no idea what a good sample size is for this would be - but if these draws were done "old school" with a chart and slips of paper with names on them - is it likely that I would draw the same person 18% of the time? Another one of our players draws someone 20% of the time. It almost seems weighted - because - wouldn't the law of averages say that over time, I should draw everyone at roughly the same rate??

There are 2 other players at our weeklys that, according to the admin module, have played in the event with me 20 more times each than this 1 person with 18% and they are both rated around 8%.

I guess my point is -- I don't want to come across as complaining or bitching -and I am not asking for a fix or an enhancement to this wonderful and FREE tool that is a great help to running our weekly DYPs. It is just that after a while, some of our players start to recognize a pattern and it is a difficult thing to explain to some of our lower ranked players or newbies that the computer draws the chart "completely random" yet for 4 out of the past 5 weeks, the two best players in the room were drawn together - and go on to win the event.

Thanks again.

Hopefully I will see you @ the DYP when I am in South Florida next month.

Steve
stevieb
 
Posts: 10
Joined: Sat Jan 05, 2008 8:26 pm

Postby Mark on Thu Feb 04, 2010 8:38 pm

Hi Steve,

No worries about the questions and I appreciate the input on the subject. As I mentioned above, Grant will be putting some details together on this as this question does come up from time to time. But, the basic answer is that the system is designed so that each time the chart is drawn up it is random using the server's built in random functions.

An interesting note is that I briefly looked at our local charts here. Since our tournament directors have found it easier to run the weekly tournaments on paper then enter them afterwards all our charts were drawn up manually. After briefly looking through the data here are some player extremes samples...

7/37 (19%)
1/46 (2%)

One other player had some extremes like...

13/71 (18%)
1/24 (4%)

One other was like...

5/25 (20%)
1/34 (3%)

I'm guessing that with a small number of tournaments it just isn't enough to get an "even distribution" yet. If this went on for 100's then 1000's of events I'm thinking they would start tightening up the distribution percentages.

Grant should be posting something in the next day or two as well that expands on this.

Thanks again,
Mark
User avatar
Mark
Site Admin
 
Posts: 759
Joined: Wed Jun 01, 2005 11:56 am
Location: Boca Raton, Florida

Postby hughsfoos on Thu Feb 04, 2010 9:56 pm

Hey guys,

Mark, that was the correct answer there.

I've been dealing with complaints from players for MANY years about crazy draws with or without using netfoos. The biggest complainers have usually been the ones with limited attendance OR the ones having what I call a "bounce back". This is where someone has had some great draws for a while, then the draws swing the other way to a stretch of terrible draws.

This has also played out with "live draws" with chips, paper lots, etc. I've seen players draw the same partner 4-5 times in a row with live draws. I've also seen the lucky player that only shows up once a month or so, but draws one of the best players there.

With over 14 years of running tourneys, I've noticed that most ideas, suggestions, and complaints come from those players who are most concerned about making sure it's geared for themselves. Therefore, I usually use the rule that if it's not the best thing for EVERYBODY it's probably just something self serving.

Locally, we've taken the approach of using netfoos OR chips to do the draw. Sometimes it's because we're playing a split-draw (A/B draw) that isn't supported by netfoos. Sometimes it's just to mix things up. It is NEVER because some player didn't like netfoos or some other means of drawing it up.

I used the admin module provided by Mark to check the partners I've had since we started using netfoos and learned that, in dyp's, I drew the players that played the most more often than the ones who played less. In the long run, you'll see that it will come pretty close to even. In the short run, it'll look like it's not random.

Remember, even with a random draw, there WILL be aberrations that don't APPEAR random.

Also, always remember to look at the underlying reason for the complaints or suggestions.

Oh, if things REALLY ARE out of balance where YOU play, just make sure there's ALWAYS a witness to the draw. There actually are tournament directors that will try to cheat the system.

Once again, good job Mark.


Hugh
hughsfoos
 
Posts: 523
Joined: Wed Jul 25, 2007 3:49 pm
Location: University Billiard Club

Postby deadbarfoos on Fri Feb 05, 2010 7:46 am

when you do draws ,you have a signup list (is it in the same order of names or a different order of names?) this has some bearing on how you end up with the same person again

1st week
1 Joe
2 Mary
3 Sean
4 Fred

2nd week
1 Sean
2 Fred
3 Mary
4 Joe

example
1st week you have Joe/Sean (thats a 1 and a 3)
2nd week you have the names in a different order and end up drawing a 4 and 1
thats Joe and Sean again!

we manual draw ours with plastic number chips and input the chart afterwards
i got the same person 3 times for the 3 weeks (in a months time) i know the signup list was in different order each time ,but i figure this happens ,its luck
so far i've gotten different people like it normally should happen ,some other 2 people started drawing each other like 4 times!
deadbarfoos
 
Posts: 775
Joined: Tue Sep 05, 2006 9:29 am
Location: Dallas,TX

Postby foosfighter on Fri Feb 05, 2010 1:47 pm

7/37 (19%)
1/46 (2%)

One other player had some extremes like...

13/71 (18%)
1/24 (4%)

One other was like...

5/25 (20%)
1/34 (3%)
Last edited by foosfighter on Thu Apr 08, 2010 12:45 pm, edited 1 time in total.
foosfighter
 
Posts: 279
Joined: Tue Sep 13, 2005 2:06 pm

Postby Mark on Fri Feb 05, 2010 4:19 pm

I believe I have it programmed in to only count the events in which both players participated. I'll double check that to make sure though.

Mark
User avatar
Mark
Site Admin
 
Posts: 759
Joined: Wed Jun 01, 2005 11:56 am
Location: Boca Raton, Florida

Postby nebraskafoosball on Mon Feb 15, 2010 2:19 am

I did a min of 100 events with the module and here's what I got...


Times Partered / Times Played in Same Event / Percentage Partnered
8 / 120 / 7%
10 / 114 / 9%
5 / 114 / 4%
11 / 133 / 8%
10 / 106 / 9%
7 / 118 / 6%
6 / 100 / 6%
8 / 161 / 5%
5 / 123 / 4%



Looks fairly even to me.
User avatar
nebraskafoosball
 
Posts: 35
Joined: Mon Jan 22, 2007 1:02 am

Postby damnlucky on Mon Feb 15, 2010 8:04 pm

The nature of being "random" is random. Stuff happens. Patterns repeat, more often in small samples. We've been using numbered chips, drawn by unaffiliated bystanders, for six years of local events that I've overseen. We get that "recurring pairs" phenomenon more times than what I would ever expect. But nobody can see through the Crown Royal bag that holds the chips. The chips are identical in texture, so touch would offer no clues about which one is which. The draws are always done by non-participants, often total strangers. Sign-ups are done in the order that the players arrive in, which is relatively consistent, but without any actual impact. Maybe it's Foosball Magic... with or without the generator.
damnlucky
 
Posts: 2
Joined: Mon Feb 15, 2010 7:15 pm


Return to NetFoos Services

cron

 

copyright 2005-2009 netfoos.com • all rights reserved