Support NeoGAF

[donny2112]

First, no, this is not the next NPD Prediction Thread. (May NPD goes from May 6 - June 2 and will be released June 14.) Also, no, we will not be stopping the current results of showing who was closest by units for the month.

This is for discussion/voting on a new, overall points system for prediction results. Whichever points system we decide to go with will have both a total 2007 ranking and a ranking for just the last three months. For reference, a lot of discussion of this was done in the last couple of pages from April's NPD Prediction topic.



System #1:

Point system based on your rank for the month (relative performance).

Rules

1. Bonus Points for finishing in the top 10: #1 - 17, #2 - 12, #3 - 11, #4 - 9, #5 - 7, #6 - 5, #7 - 4, #8 - 3, #9 - 2, #10 - 1

2. #1 gets 150 points, so 167 with the bonus.

3. #2-#75 drop in 2 point increments, so #2 gets 148 (+12) and #75 gets 2. The exception to that is if there are < 150 predictors, in which case only the top half (rounded up) will get points. Those in the top half would get full points, though.




System #2:

Point system based on how close you were to the "truth" (absolute performance)

Rules

1. 25 points for getting Hardware positions correct.
This means that you got the order (1-7) correct regardless of the actual values you put in for each hardware system. You have to have distinct values for each system, though.

2. 20 points * (percentage of correct units for hardware system, anything over 100% difference = 0 points). Maximum points: 140
This means that if you missed the 360 value by 8%, you'd get 0.92*20 = 18.4 points for the 360 prediction. Repeat for each of the other seven main systems.

*3. Possibly some bonus for place of finish in the normal unit ranking for each month.



Don't lean toward one or the other based on how complicated either one is to tally the score for. If I'm doing the tallying, I can do either without a lot of trouble. Also to repeat, this is in addition to the normal unit ranking that is done each month.


Please state which system you think makes the most sense and/or any comments. Thanks!

 

[Tieno]

Wait, there's a point system? No, wait, there's discussion on a point system?

Shit, sales-age indeed. Hardcore. :lol

 

[deadbeef]

I think both relative and absolute performance should be taken into account.

 

[Pope Benedict XVI]

i suppose the good thing about system #2 is that everyone can work out their own score easily. With system #1 wouldn't someone have to make out a table of 75 names for people to check?

edit: oops i didn't read that last part of the OP

 

[donny2112]

I think both relative and absolute performance should be taken into account.

Both definitely have merit. However if we did both for annual and 3-month along with the normal unit ranking, we'd have 5 different rankings to keep track of each month. I think that's overkill.

Spoiler

*ignores the ringing irony in his head about a points system for video game hardware sales predictions also being overkill*

 

[Deku]

Wait, there's a point system? No, wait, there's discussion on a point system?

Shit, sales-age indeed. Hardcore. :lol

Nothing wrong with that. Especially if Pachter continues to publish his predictions beforehand, it becomes an interesting game of beat the analyst.

 

[Orgen]

System #1 for me

 

[Visualante]

This systems puts me off predicting at all.

[/ralph wiggum]

 

[Crypto]

How many bite size Kit Kats can I get for 1000 points?

 

[duk]

are these Wii points? because I got a shit load

 

[deadbeef]

i suppose the good thing about system #2 is that everyone can work out their own score easily. With system #1 wouldn't someone have to make out a table of 75 names for people to check?

edit: oops i didn't read that last part of the OP

Wouldn't it be pretty trivial to decide on a common format for posting your predictions, and then use either a spreadsheet or some C code or something to process the results? Then you could either wait and see where you placed or run the results yourself.

 

[Pope Benedict XVI]

Wouldn't it be pretty trivial to decide on a common format for posting your predictions, and then use either a spreadsheet or some C code or something to process the results? Then you could either wait and see where you placed or run the results yourself.

lol i'm not doing that


But if donny is happy to post a list then it's fine

 

[duderon]

System number 2, yo.

 

[HomShaBom]

points? WTF IS THIS SHIT

 

[Diablos54]

points? WTF IS THIS SHIT

THIS IS SALES AGE!!

And I prefer System 2.

 

[sonycowboy]

You already know, but I prefer system #2 (as it's mine from a year or two ago )

 

[sonycowboy]

Wouldn't it be pretty trivial to decide on a common format for posting your predictions, and then use either a spreadsheet or some C code or something to process the results? Then you could either wait and see where you placed or run the results yourself.

I've tried getting people to post in a standard format. It doesn't work. The best hope would be to have an application support it. It is a little bit of work, but to get the best participation, it's best to let people make their predictions in their own way, and not make it overly difficult on them.

 

[yoopoo]

This is the worst thread ever made, on any forum. You folks are insane over sales.

 

[DSWii60]

System #2 definitely as its based on absolute perfomance and therefore will be the best indicator has to who is the best armchair analyst.

A relative system does not take into account how close you actually got to the real numbers, just how well you did in relation to others. So, for example, if there is a huge gap between positions 1 and 2 and only a small gap between 2 and 3 this will not affect the rankings, whereas I think it should.

Spoiler

I hope that all made sense

 

[sonycowboy]

This is the worst thread ever made, on any forum. You folks are insane over sales.

Then go somewhere else. Or don't read this thread. And certainly, don't bother posting in it if you don't care.

 

[yoopoo]

Then go somewhere else. Or don't read this thread. And certainly, don't bother posting in it if you don't care.

Cry-age.

 

[deadbeef]

A relative system does not take into account how close you actually got to the real numbers, just how well you did in relation to others. So, for example, if there is a huge gap between positions 1 and 2 and only a small gap between 2 and 3 this will not affect the rankings, whereas I think it should.

No but there is certainly something to be said for being able to predict how the various hardware platforms place in relation to one another.

Just drop the number of points for System #1 to something more reasonable and add it to the points from System #2. So Total Points = P(System 1) + P(System 2)

 

[Stumpokapow]

This is the worst thread ever made, on any forum. You folks are insane over sales.

I know! I was expecting Devil May Cry 4 screenshots when I clicked on this thread! What a misleading title!

 

[yoopoo]

I know! I was expecting Devil May Cry 4 screenshots when I clicked on this thread! What a misleading title!

Really? Wow you're stupid.

 

[sonycowboy]

Cry-age.

I'm just stating the obvious and it comes up at some level every single month. If I see a Pokemon thread, I know what's going to be in there and I don't go crazy if people are Pokemon freaks.

Why some people are surprised when they come into a sales thread that people are actively and earnestly interested in sales is beyond me. I do understand that folks can have a casual interest in seeing what sales are, but there's a decent number of folks @ NeoGAF who like to talk about sales, ad-naseum.

People should stop being surprised, annoyed, dismayed, disgusted, whatever over this.

 

[Burai]

 

[donny2112]

http://i190.photobucket.com/albums/z73/Burai_uk/Achievement.jpg

:lol That's great.

 

[Parl]

Edit: Irrelevant read...

 

[donny2112]

Well, a percentage difference is more logical than actual difference when it comes to fair scoring, IMO.

That's basically System #2. Using your first example, person A would get 77.68 points for those predictions, and person B would get 66.54 points.

 

[Parl]

That's basically System #2. Using your first example, person A would get 77.68 points for those predictions, and person B would get 66.54 points.

Yeah, and counting the systems individually (your system #2) is actually better because you don't get penalised for getting one system way off. I agree on the ordering, too. As somebody could have inflated figures, but get the market shares quite well. But I don't think it should be as black and white as "25 points for order right, but 0 if you get ANY of the order wrong".

Maybe 10 points for all right and two mixed up or something? (I don't think you wanna go into calculating individual market shares for each console for each prediction and compare those against actual market shares for each console in the actual numbers and have a separate formula for that).

I definitely go with system #2. System #1 relies on the skills/luck of other GAFfers to determine your score, and system #1 gives you a score on your skills/luck, but maintains the ability to compare yourself to other GAFfers. AND, it's a much fairer system, IMHO.

 

[Odysseus]

this is some hardcore nerdery going on right here

i mean, it was cool just seeing who was closest month-to-month going by absolute differences

but if you guys want to go nextel cup with it, i'm eager to see who makes the chase at the end of the season. make it so that only the top ten or twelve or whatever qualify to make predictions for november and december.

 

[Tieno]

Nothing wrong with that. Especially if Pachter continues to publish his predictions beforehand, it becomes an interesting game of beat the analyst.

I don't mind really, I was just a bit surprised and find it funny. Carry on.

 

[bud]

this is some hardcore nerdery going on right here

understatement of the year.

 

[JJConrad]

I kind of like #2, but I'd remove its rule 1 and modify 3 with several oportunties for small bonuses. Some are relevent, some are fun. For example...

5 point bonus for proper ranking order
5 points for being the closest per platform
3 points for making overall top 10
3 points for making overall top 3
3 points for being overall #1
3 points for accurately predicting PS3>GBA or 360>PS2 or 360>Wii
1 point for predicting all required platforms
1 point for predicting better than Pachter
1 point for beating last month's winner
1 point for not getting banned in the official NPD thread
1 point per title for predicting the games over 100k
-1 point for per title for incorrect games over 100k
10 points for being JJConrad

 

[Cheesemeister]

Nah, just sum the absolute differences between the prediction and actual for each system. Lowest diff wins.

 

[bjork]

Wow. This is *almost* as comical as that thread about how this forum has tiers of posters, and people got rated all through it. Almost.

 

[mentalfloss]

System 2

 

[Parl]

Nah, just sum the absolute differences between the prediction and actual for each system. Lowest diff wins.

Disregard proportions entirely?

 

[starship]

I prefer System 2 as well.

 

[Zynx]

Nah, just sum the absolute differences between the prediction and actual for each system. Lowest diff wins.

Yes, I think this is best. It rewards being accurate with predictions, and it doesn't matter what others have predicted.

Disregard proportions entirely?

Yes - with smaller actual amounts (say, GBA or PS3), being off by the same absolute amount (say, 1k) is a larger percentage error, sure. But you could consider the prediction for that platform as being less important since the platform you're predicting is a smaller part of the overall market.




I don't like either ranking by comparing with others, or ranking by getting the hardware 'order' correct.

Comparing with others - your rankings get 'expanded' if a lot of people predict around the same number.

Let's say example A has true numbers of 147k, and people predicted 145k 144k 143k 100k. The 143k predictor is in 3rd place despite being very close.
Example B has true numbers of 147k again, but people predicted 145k 110k 109k 100k. In this case the 3rd place person is far off, but still gets the same score as the previous 3rd place.

That's why I like systems that don't take into account what others have predicted.

Also, 'ranking' I think is less meaningful when some systems are very close together. If you rank PS3 ahead of GBA, but they differ only by a little, I think guessing the ballpark which both platforms are close to is more meaningful than guessing which one is ahead.



Oh, as carryover your q from the previous thread...what I propose to avoid the need for hand copy-pasting into excel is requiring everyone to follow the same format when posting their predictions. It doesn't need to be complicated. Then a program could automatically extract predictions. (Actually it'd end up being a very, very, very simple version of what Cheesemeister does for Media-Create.)

As simple as [system] prediction (k) would do for a format, for example:

[Wii] 430k
[DS] 450k
[PS2] 190k
[Xbox 360] 180k
[GBA] 80k
[PS3] 90k
[Xbox] 10k

 

[zsidane]

System #2 for me

 

[Cheesemeister]

Oh, as carryover your q from the previous thread...what I propose to avoid the need for hand copy-pasting into excel is requiring everyone to follow the same format when posting their predictions. It doesn't need to be complicated. Then a program could automatically extract predictions. (Actually it'd end up being a very, very, very simple version of what Cheesemeister does for Media-Create.)

That'd be a cinch to program.

 

[Armitage]

If you get 1000 points you should be able to trade it in for a Schrute Buck.

 

[titiklabingapat]

Wow. This is *almost* as comical as that thread about how this forum has tiers of posters, and people got rated all through it. Almost.

Link?

 

[Earthstrike]

System #2 sounds better, but what about the possibility of a conglomerate system?

If we're still doing monthly deviation sums and ranking people, people can get bonus points for how they place in that ranking system. These points shouldn't have the ability to dominate the points distributed by system #2, but they should play a small role.

 

[Fredescu]

I like System 2. Thanks for the effort you and others are putting into this by the way, despite the insults from those who don't care.

 

[Flakster99]

Nah, just sum the absolute differences between the prediction and actual for each system. Lowest diff wins.

My choice.

Though, if I had to choose a system, #2 for me.

 

[ComputerNerd]

Wouldn't it be easier for somebody to just create a simple website to record the stats?

Enter your user name, and the #'s for each system. Simple. Disable it when NPD's are about to come out.

Then you can calculate the results MUCH easier.

I think it would be easier than using a parser. It'd be faster, and more accurate.

 

[bjork]

Link?

might've been on the old board, I don't see it under a search for "tier"

 

[Bildi]

I'll go with System 2 for the reasons stated by Sonycowboy in the other thread. It's a good idea. I admit I don't really care for point (1) of System 2 since someone could easily miss out on that in spite of being the person who hit closest to the mark overall, but that's not a big deal.

Nice work donny.