adak

Your post shows me this:
Code:
Attachments are not available: Download requirements not met


Which doesn't help.

What I need to see is, the spreadsheet with 2011 data, and then a projected spreadsheet (same equations), with twice as much production - and see how it looks.

What you're trying to do is to quantify a number of points increase, to be equal to a number of percent growth. 

So how many points is equal to a 1% growth increase, and does it seem reasonable?

I'm skittish about this kind of math, because points will surely be increasing tremendously from year to year, in the bigger teams. 

But post your spreadsheet data so it can be seen, and let's take a look at it.

 
This is his reply to me sending him to those posts  

Well, Adak, 1% increase is 1% increase. If the previous year's production is 100, you need to do 101 for 1%. Not to question your intent, but is it really difficult to judge what "1% increase" means?
Given this incredibly eluding explanation, how can you say that it's easier for bigger teams to grow faster? If the previous year's production is large, it means you need more points to get the same 1% growth as a smaller team. For example, say another team has a production of 1000, they need to do 1010 to get 1% growth. 101 minus 100 = 1. 1010 minus 1000 = 10. 10 is larger than 1. 
 
Sorry that you couldn't see the tables before, but don't make disingenuous comments just to be critical.
 
I'm posting up the table for #502 again:

 
As you see, the smaller teams have in fact much larger growth. ChimpPowerUp's growth is more than three times EVGA's. So, no, Growth does not favor larger teams. In fact, I can argue just the opposite.
 
You asked for the same table with hypothetical double points in CC2011. I'll humor you:

 
As you see, the result is the same: growth ranking, and the overall ranking is the same.
 

In this post adak appears to be completely oblivious to the type of format we have been discussing.  Granted, this may be because things have not been clearly explained to him.  If things have not been laid out crystal clearly that needs to be done asap.

adak

What I need to see is, the spreadsheet with 2011 data, and then a projected spreadsheet (same equations), with twice as much production - and see how it looks.

 
No you don't, you simply need to take the time to understand what is being proposed.
 

adak

What you're trying to do is to quantify a number of points increase, to be equal to a number of percent growth. 

 
No we're not.  Percent growth is based purely on percent growth.  (pretty mind-blowing, huh?).  The number of points is only relevant so much as it relates to the individual team's production last year.

adak

So how many points is equal to a 1% growth increase, and does it seem reasonable?

 

A 1% increase in points equals a 1% increase in growth.  That seems reasonable to me.

 

adak

I'm skittish about this kind of math, because points will surely be increasing tremendously from year to year, in the bigger teams. 

 
While this is true, there is absolutely no reason to believe their relative point increase or large teams is not in line with the relative point increase of small teams.  That is the metric we are using.
 

adak

But post your spreadsheet data so it can be seen, and let's take a look at it.

 
Not worth throwing one together if he doesn't understand what is going on.

I think (from what I've read over at OCN) Adak doesn't intend to register on our Forum which is why he is having problems seeing certain types of uploaded/added content.  Copying these comments and spreadsheet data over into the Captains Forum (assuming he is in there) may be the only way to get a response back to him.

I'm gonna go ahead and put another explanation for the three categories (they're performance measures, NOT tiers):
 
Points = Number of millions of points under CC name
Conversion = Points under CC name divided by Total Team points
Growth = (CC name points this year minus CC name points last year) divided by (CC name points last year)
 
I didn't invent these, but I believe they make most sense. "Points" favors larger teams, but "Conversion" favors smaller teams. "Growth" has no obvious favor; theoretically, it is easier to grow when you're smaller, but I see the argument that when you're larger, you have more means to grow. 
 
I have an idea for another performance measure, but I'm afraid to put the calculations for it. It may be seen as "too complicated", so I'll try to explain it in plain English. I call it "Stamina". It measures how the CC name points hold up during the competition. For example, if the CC name daily points during the competition keep increasing, Stamina will be the highest. If there are decreases along the way, Stamina measure will drop. If you look at the "Points per update" charts in (http://www.hardwarecanucks.com/chimp_challenge/2011/), I believe you'll see what it's trying to capture.
 
Does this "Stamina" measure make sense? Do you think it's a valid measure? Would like to see how it would have worked last year? (For this I need the daily production values, which are the "points per update" values from CC2011. Can anyone provide me those? Thanks!)
 

Gryphon, one of my members suggested a similar criteria. I did see two flaws:

1 - What's to stop a team holding back their points on the first day, and then slowly bringing their Folders online? Not necessarily something that will happen, but if a team would otherwise be mid-table in all categories, and was small enough - future teams - to co-ordinate it, the gain in that category could far outweigh any loss in the others.

2 - There are various ways of implementing it. Considerimg where we are, if we're aiming for a May CC, I dunno if we'd get agreement - especially when we have 3 tn test out as is. If we're moving tm later in the year though, then we can look at some simulations.

zodac


Gryphon, one of my members suggested a similar criteria. I did see two flaws:

1 - What's to stop a team holding back their points on the first day, and then slowly bringing their Folders online? Not necessarily something that will happen, but if a team would otherwise be mid-table in all categories, and was small enough - future teams - to co-ordinate it, the gain in that category could far outweigh any loss in the others.

2 - There are various ways of implementing it. Considerimg where we are, if we're aiming for a May CC, I dunno if we'd get agreement - especially when we have 3 tn test out as is. If we're moving tm later in the year though, then we can look at some simulations.

 
zodac, thanks for your response.
I know that a team can try to manipulate the Stamina measure by holding back their points on the first day, but I thought about a solution to deter from such action. Think of a very tight cap on "percent increase", like 5%. Meaning, if a team increases their daily production by more than 5%, then all they get is 5%. No caps on decreases. This way, no team will have an incentive to manipulate the score, by holding back, because they can do very good anyway by doing their best all the way. Don't forget, this is just one of the performance measures (categories), and it takes to be good in all categories to win the whole thing.
 
The way I intend to implement this is using a geometric mean (happy, Punchy? ), just like finding compound interest. So, each team will have a "compound change in points during the CC".
 

Yep, I think that 2011 CC Points Per Update chart shows what happens when the Team starts waning.  I think for us, it was hard to maintain momentum once it settled-in that no matter what we did, we weren't gonna have a chance to win.  We being "human-beans", I think that drop-off in production can/will happen to Teams when the moment sets-in, unless there is some over-riding motivation to push-on.  IIRC, the Team "gung-ho" word (to keep us going) in the contest was "well let's at least show 'em what we can do in overall Points". 
 
So, if the contest is well structured in design (fair as possible to all Teams), then I like something that rewards "Stamina".  I feel that most Teams want to stick things out to the end in most contests as long as the contest remains healthy.  None of us needs motivation to Fold, but we do need motivation to Fold harder than normal, push temps, break Rigs, turn our homes into ovens, etc. 
 
I'm not strong in math like some of you, so thanks Gryph for the Layman's description.  How does Stamina differ from Conversion in simple terms?  I was thinking that if Conversion fell off toward the end of the contest, that also feels like loss of Stamina to me.

zodac


Gryphon, one of my members suggested a similar criteria. I did see two flaws:

1 - What's to stop a team holding back their points on the first day, and then slowly bringing their Folders online? Not necessarily something that will happen, but if a team would otherwise be mid-table in all categories, and was small enough - future teams - to co-ordinate it, the gain in that category could far outweigh any loss in the others.

2 - There are various ways of implementing it. Considerimg where we are, if we're aiming for a May CC, I dunno if we'd get agreement - especially when we have 3 tn test out as is. If we're moving tm later in the year though, then we can look at some simulations.

Just a thought on #1... I always think of what I do not like to see. The opposite effect and what we see every year... Sandbaggers timing drops, or hording completed WU's by unplugging from the net... Then the contest start... and boom... HUGE numbers out of the gate. 
 
Just a different way for folks to do what they do... 

Separate note... 
 
We do need to know if this is going to happen sooner than later (CC). We have a lot of new folks that we need to help guide to the way this works and prime their name with EVGApes passkey.

texinga


Yep, I think that 2011 CC Points Per Update chart shows what happens when the Team starts waning.  I think for us, it was hard to maintain momentum once it settled-in that no matter what we did, we weren't gonna have a chance to win.  We being "human-beans", I think that drop-off in production can/will happen to Teams when the moment sets-in, unless there is some over-riding motivation to push-on.  IIRC, the Team "gung-ho" word (to keep us going) in the contest was "well let's at least show 'em what we can do in overall Points". 

So, if the contest is well structured in design (fair as possible to all Teams), then I like something that rewards "Stamina".  I feel that most Teams want to stick things out to the end in most contests as long as the contest remains healthy.  None of us needs motivation to Fold, but we do need motivation to Fold harder than normal, push temps, break Rigs, turn our homes into ovens, etc. 

I'm not strong in math like some of you, so thanks Gryph for the Layman's description.  How does Stamina differ from Conversion in simple terms?  I was thinking that if Conversion fell off toward the end of the contest, that also feels like loss of Stamina to me.

 
Well, they are different. Let me try to explain. 
 
There are two ways the  conversion (the CC-name production relative to the total team production) can drop during the competition, otherwise it's going to stay somewhat constant: 
1) If people change their names from the "CC name" to their individual names during the competition,
2) If the "non-CC-name" team production increases more than "CC name" production.
 
Only the first one will affect the Stamina measure, but only if the CC-name points is not compensated by increased production. Similarly, the first case will also affect the "Points" measure, as well as the "Growth" measure. In short, while there is some interaction between these measures, they are not measuring the same thing.
 
What Stamina is measuring is the dynamics of the daily production of the CC name, which is not relative to anything but CC name points. It's about how the daily production changes during the competition.
 
Conversion is a cumulative measure of CC name points relative to Total Team points. It puts together the whole CC name production (regardless of how it changed during the competition) and divides it by the Total Team production in the same timeframe.
 
 

Afterburner

zodac


Gryphon, one of my members suggested a similar criteria. I did see two flaws:

1 - What's to stop a team holding back their points on the first day, and then slowly bringing their Folders online? Not necessarily something that will happen, but if a team would otherwise be mid-table in all categories, and was small enough - future teams - to co-ordinate it, the gain in that category could far outweigh any loss in the others.

2 - There are various ways of implementing it. Considerimg where we are, if we're aiming for a May CC, I dunno if we'd get agreement - especially when we have 3 tn test out as is. If we're moving tm later in the year though, then we can look at some simulations.

Just a thought on #1... I always think of what I do not like to see. The opposite effect and what we see every year... Sandbaggers timing drops, or hording completed WU's by unplugging from the net... Then the contest start... and boom... HUGE numbers out of the gate. 

Just a different way for folks to do what they do... 

 
Hi AB. The 5% cap that I intend to put would take care of unnatural spikes in points due to timing, etc. With that 5% cap for increases, there won't be an incentive to cheat the Stamina measure. At the end, what it is intended to capture is "how well the team can hold it up during the competition", and NOT "how fast they can accelerate"...
 

Thanks Gryph, I was wondering what the time-spans were for calculating Conversion and Stamina...makes sense now.  Thanks for the explanations.

I have reservations about the stamina measure.  Given the inherent variability and random nature of folding, I can't think of a statistically relevant way to measure anything useful over such a small sample size.  Arbitrary elements such as a 5% cap are undesirable. 

Afterburner


Separate note... 

We do need to know if this is going to happen sooner than later (CC). We have a lot of new folks that we need to help guide to the way this works and prime their name with EVGApes passkey.

???
I think what you might mean is to prime their passkey with the EVGApes name, but that's only needed if people want to watch their own stats while folding under EVGApes name.  Otherwise you can just switch to the shared passkey shortly before the contest and not worry about priming.
 
Your first two points are still important and time-critical.

troy8d


I have reservations about the stamina measure.  Given the inherent variability and random nature of folding, I can't think of a statistically relevant way to measure anything useful over such a small sample size.  Arbitrary elements such as a 5% cap are undesirable. 

 
I understand your concern regarding natural variability. However, that variability will apply to all teams, so they're vulnerable the same way. Therefore, stamina measure will be fairly calculated, even though there will be some noise for each team.
 
I'm not happy with an arbitrary cap either. However, the idea is to deter any manipulation. We can of course discuss where the cap should be; 5% was just an initial thought, but (Edit) a cap like that it would work very well as a disincentive. The intention of this Stamina measure is again: to capture "how well the team can hold it up during the competition", and NOT "how fast they can accelerate".
 
I can experiment with different caps like 5%, 10%, 25% and see how they fare.
 

OK, I have an idea: A cap of zero! That is, any increases in daily production will have zero effect, which means, it will not decrease your Stamina. This way, Stamina will be measuring exactly "how well can the team hold it up during the competition". Plus, the nice thing about zero cap is that it is not arbitrary; it serves a purpose, saying "only decreasing points will affect your Stamina".
 
I've gathered the daily production points from CC2011, and ran the Stamina on that data, using zero cap. 
 
First, the Stamina calculations, ranked:

 
As you see, the two large teams are in the middle of the pack. Therefore, it's not favoring neither large nor small teams. 
 
Then, overall final scores with four measures, "Points", "Conversion", "Growth", and "Stamina":

 
 

In my opinion it's still too easily affected. What if by chance we have lots of 6903/4s drop on Day 1? We'd do much worse the next day. See what I'm saying?

texinga


I think (from what I've read over at OCN) Adak doesn't intend to register on our Forum which is why he is having problems seeing certain types of uploaded/added content.  Copying these comments and spreadsheet data over into the Captains Forum (assuming he is in there) may be the only way to get a response back to him.

This is exactly true, thats mainly how Ive been transferring answers to him is copy and pasting and downloading pics as attachments.

adak

OK, I like some of the spreadsheet, but this is a question:

TSC! and T32 have VERY similar stats in both 2010, and 2011. We're neck and neck in points. But when you look at growth rate, TSC! is magically jumped up by 34 points!

What is the equation he's using for the first growth column, before it's corrected. (Yes, I see that after the correction, we're back to being very close again, but why are we so different in the left side growth column figure?

 
He gave this reply to me sending him the spreadsheet from post 502. I just posted the other replies that you guys have done.

adak

OK, I like some of the spreadsheet, but this is a question:

TSC! and T32 have VERY similar stats in both 2010, and 2011. We're neck and neck in points. But when you look at growth rate, TSC! is magically jumped up by 34 points!

What is the equation he's using for the first growth column, before it's corrected. (Yes, I see that after the correction, we're back to being very close again, but why are we so different in the left side growth column figure?

Huh?  Not really neck and neck.  Looking at the spreadsheet, (and I'm assuming you divide CC2011 points by 10 to get ppd),
TSC: 2.1 -> 4.57 = 117% increase year-to-year
T32: 2.3 -> 4.27 = 86% increase

Im glad you guys are here to help. Im good with numbers but not this good.

adak

OK, I like some of the spreadsheet, but this is a question:

TSC! and T32 have VERY similar stats in both 2010, and 2011. We're neck and neck in points. But when you look at growth rate, TSC! is magically jumped up by 34 points!

What is the equation he's using for the first growth column, before it's corrected. (Yes, I see that after the correction, we're back to being very close again, but why are we so different in the left side growth column figure?

Adak, there is no magic with those calculations. The equation for the growth column is (for the zillionth time): (Points in 2011 minus Points in 2010) divided by (Points in 2010). That's the regular percent change formula, nothing invented. Just pay attention that the data for 2010 was received as PPD, so to find the points, I had to multiply it by 10 (10 days in CC 2011).
 

csm725OCN


In my opinion it's still too easily affected. What if by chance we have lots of 6903/4s drop on Day 1? We'd do much worse the next day. See what I'm saying?

 
CSM, it is true that the Stamina measure is susceptible to some noise. But again, all teams are susceptible to the same noise and therefore the measure is unbiased. That is what matters for a competition. 

While I agree that a "Stamina" category would be handy, I think there are things that need to be adjusted to it before it can be used.
 
My opinion - posted in the CC forum - is that we keep the three already listed (points, growth, conversion) for this year. This means we keep the CC name too. Next year, we drop the CC name, and aim to add another category or two, which would include the Stamina equation.
 
Again, if we're planning for May, I think we need to be finalising things, rather than trying to add more.

Sure. I'm not pushing anything, just sharing and trying to explain what I have in mind.
devdog51, please let me know if you need my help in clarifying things related to my proposal(s).
 

Alright chums lets do this...LEEEEEEROOOOOOOOYYYYY!

Punchy

Afterburner


Separate note... 

We do need to know if this is going to happen sooner than later (CC). We have a lot of new folks that we need to help guide to the way this works and prime their name with EVGApes passkey.

???
I think what you might mean is to prime their passkey with the EVGApes name, but that's only needed if people want to watch their own stats while folding under EVGApes name.  Otherwise you can just switch to the shared passkey shortly before the contest and not worry about priming.

Your first two points are still important and time-critical.

Thanks Captain Turnip (Really I mean that, even if I am still having fun with your name )

Love where this is right now folks.. Pretty exciting actually 

theGryphon


Sure. I'm not pushing anything, just sharing and trying to explain what I have in mind.
devdog51, please let me know if you need my help in clarifying things related to my proposal(s).

Will do and thanks to you and Punchy for all the help so far.