-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Official April 2007 NPD Prediction Thread
- Thread starter donny2112
- Start date
Souldriver
Member
:lolMerovingian said:
Corwood Rep
Member
So are we doing the numbers in this thread or a new one?
P.S. OMGOMGOMGOMG FROTHING DEMAND INCREASES
P.S. OMGOMGOMGOMG FROTHING DEMAND INCREASES
Merovingian
Banned
Parl said:
Parl was right about getting 14 points from that scale. :lol
April 2007 NPD Prediction Results - 7 main consoles only
1. BuzzJive - 108K
2. Tf53 - 109K
3. speculawyer - 118K
4. Dimmuxx - 121K
5. apujanata - 127K
6. DayShallCome - 130K
7. Mallika - 135K
8. argon - 153K
8. Giddeon - 153K
10. donny2112 - 154K
Congratulations, BuzzJive!
Bonus Results - Wii, PS3, and 360 only
1. sonycowboy - 17K
2. chibcicylist - 24K
2. corez - 24K
2. Rancid Mildew - 24K
2. jarrod - 24K
Pachter did much better this month ranking at number 25 out of 160.
andthebeatgoeson
Junior Member
donny2112 said:
I'm counting 11. He just barely missed with that second 360 note of less than 175K in sales.
Edit: Oh man, I see, that's just awful. We might need to start a how low can the 360 go pic.
donny2112 said:April 2007 NPD Prediction Results - 7 main consoles only
1. BuzzJive - 108K
2. Tf53 - 109K
3. speculawyer - 118K
4. Dimmuxx - 121K
5. apujanata - 127K
6. DayShallCome - 130K
7. Mallika - 135K
8. argon - 153K
8. Giddeon - 153K
10. donny2112 - 154K
Congratulations, BuzzJive!
Bonus Results - Wii, PS3, and 360 only
1. sonycowboy - 17K
2. chibcicylist - 24K
2. corez - 24K
2. Rancid Mildew - 24K
2. jarrod - 24K
Pachter did much better this month ranking at number 25 out of 160.
I am happy that I got #5.
BTW, I am going post later the overall result of march Top10 and April Top10, in which I will list the April result of March Top10, and March result of April Top 10, so we can get a leaderboard of the all Top10 predictor.
Sample :
Name : March 07, April 07 : Total
Xisiqomelir : 133,000, 442,000 : 575,000
Dice Man : 158,000, N/A : 158,000 *
LOCK : 186,000, 840,000 : 1,026,000
Heidir : 198,000, 231,000 : 429,000
Bildi : 202,000, 214,000 : 416,000
speculawyer : N/A, 118,000 : 118,000 *
apujanata : 238,000 , 127,000 : 365,000
donny2112 : 334,500, 154,000 : 488,500
* means the person have some missing month(s) of prediction, which mean he could not be ranked (because of incomplete data).
Code:
Complete List of Top 10 Predictor (have both March and April)
User ID Mar-07 Apr-07 Total Months# Total Rank
apujanata 238,000 127,000 365,000 2 365,000 1
buzzjive 277,000 108,000 385,000 2 385,000 2
Bildi 202,000 214,000 416,000 2 416,000 3
Heidir 198,000 231,000 429,000 2 429,000 4
Tf53 346,000 109,000 455,000 2 455,000 5
DSWii60 197,000 280,000 477,000 2 477,000 6
donny2112 334,500 154,000 488,500 2 488,500 7
argon 342,000 153,000 495,000 2 495,000 8
Dayshallcome 391,000 130,000 521,000 2 521,000 9
Miniboss1232 197,000 355,000 552,000 2 552,000 10
TheKurgan 188,000 384,000 572,000 2 572,000 11
Xisiqomelir 133,000 442,000 575,000 2 575,000 12
LOCK 186,000 840,000 1,026,000 2 1,026,000 13
Incomplete List (don't have both March and April)
User ID Mar-07 Apr-07 Total Months#
Dice Man 158,000 N/A 158,000 1
the thoroughbred 166,000 N/A 166,000 1
jet1911 177,000 N/A 177,000 1
speculawyer N/A 118,000 118,000 1
Dimmuxx N/A 121,000 121,000 1
Mallika N/A 135,000 135,000 1
Giddeon N/A 153,000 153,000 1Anyone I missed ?
Grampa Simpson
Banned
Am I to be found anywhere in your compiled list of who placed where?apujanata said:
dirtmonkey37
flinging feces ---->
I've been contemplating these numbers -- did the 360's figures in NA alone outperform the world wide PS3 numbers for April?
PS3:
NA: 82K
JPN: 50K (I believe)
EU: ?
360:
NA: 174K
JPN: 10K (I believe)
EU: ?
PS3:
NA: 82K
JPN: 50K (I believe)
EU: ?
360:
NA: 174K
JPN: 10K (I believe)
EU: ?
bmf said:
Nope. The list I compile is only for the ones who have managed to get into the Top 10 for any month (currently only for March and April). If you wanted to get your rank in April, ask donny2112. He might be willing to do that for you.
BTW, I noticed you didn't do any prediction for March 2007.
Grampa Simpson
Banned
Yeah. I've never really been much for the prediction threads, although I've been more interested in them in the last couple of months.apujanata said:
DayShallCome
Member
Didn't Donny also do data for February predictions?
Grampa Simpson
Banned
I can't decide if you're funny or oblivious.Windu said:
RPS37 said:
link
bmf said:
You're #50 for April. 254K off.
DayShallCome said:
And January.
If we're going to do cumulative prediction rankings, it might be better to do it via "% diff" instead of "unit diff." The reason is that for November and December, #10 might be 700K off. That weights the predictions heavily toward those two months, whereas it should be equal amongst all the months.
For a single month, "unit diff" is fine as everyone is ranked against the same magnitude of numbers.
There are 21 users who have predicted every month this year (Jan - Apr). Here are their rankings.
Total units:
1. argon - 1,071,000
2. sonycowboy - 1,095,000
3. donny2112 - 1,099,500
4. Clever Pun - 1,176,000
5. BuzzJive - 1,195,000
6. GhaleonEB - 1,202,000
7. JoshuaJSlone - 1,270,000
8. duderon - 1,306,000
9. Orgen - 1,309,000
10. Ganondorf>Link - 1,326,000
11. Kurosaki Ichigo - 1,447,000
12. rage1973 - 1,467,000
13. starship - 1,517,000
14. hadareud - 1,614,000
15. djtiesto - 1,631,000
16. bycha - 1,732,000
17. Lock - 1,748,000
18. Heidir - 1,774,000
19. Earthstrike - 1,992,000
20. PhoenixDark - 2,034,000
21. beermonkey@tehbias - 2,054,060
By percentages:
1. sonycowboy - 448.50
2. BuzzJive - 466.44
3. argon - 469.54
4. donny2112 - 483.13
5. Clever Pun - 487.95
6. GhaleonEB - 546.73
7. duderon - 556.16
8. Orgen - 558.79
9. JoshuaJSlone - 569.03
10. rage1973 - 591.32
11. Ganondorf>Link - 593.87
12. Kurosaki Ichigo - 608.76
13. starship - 628.74
14. Heidir - 681.15
15. Lock - 706.02
16. hadareud - 731.98
17. bycha - 732.42
18. djtiesto - 744.12
19. Earthstrike - 825.36
20. beermonkey@tehbias - 860.79
21. PhoenixDark - 952.06
Here's another option. This uses a modified version of the NASCAR Point System and is not limited to people who actually predicted all four months.
#1 gets 183 points.
#2-#10 get 5 points less per position, so #2 gets 178, #3 gets 173, etc.
#11-#25 get 4 points less per position, so #11 gets 134, #12 gets 130, etc.
#26-#50 get 3 points less per position, so #26 gets 75, #27 gets 72, ..., #50 gets 3.
#51-whatever get zero points.
Basically, this removes the deadly penalty of missing a month (now it's as if you finished #51 or worse), and ranks you more on how you did compared to your peers that month than in an absolute sense. I like a points system best of the three, but it doesn't have to be this particular points setup.
Here are the top 21 through April using that method.
1. argon - 474
2. donny2112 - 454
3. BuzzJive - 392
4. Lock - 379
5. sonycowboy - 365
6. Clever Pun - 346
7. GhaleonEB - 329
8. DayShallCome - 321
9. JoshuaJSlone - 321
10. Orgen - 271
11. apujanata - 269
12. JJConrad - 261
13. Ganondorf>Link - 257
14. bycha - 248
15. duderon - 233
16. starship - 216
17. Kurosaki Ichigo - 207
18. rage1973 - 201
19. Bildi - 198
20. Heidir - 197
21. jet1911 - 192
To use apujanata as an example...
He didn't predict anything in January or February, which completely eliminates him from the first two rankings. However with the points setup, his #22 March rank and #5 April rank put him at #11 overall.
To use DayShallCome as an example...
He didn't predict anything in January, which completely eliminates him from the first two rankings. However with the point setup, his #5 February rank and #6 April rank (he finished #104 in March) put him at #8 overall.
Comments? Suggestions?
Total units:
1. argon - 1,071,000
2. sonycowboy - 1,095,000
3. donny2112 - 1,099,500
4. Clever Pun - 1,176,000
5. BuzzJive - 1,195,000
6. GhaleonEB - 1,202,000
7. JoshuaJSlone - 1,270,000
8. duderon - 1,306,000
9. Orgen - 1,309,000
10. Ganondorf>Link - 1,326,000
11. Kurosaki Ichigo - 1,447,000
12. rage1973 - 1,467,000
13. starship - 1,517,000
14. hadareud - 1,614,000
15. djtiesto - 1,631,000
16. bycha - 1,732,000
17. Lock - 1,748,000
18. Heidir - 1,774,000
19. Earthstrike - 1,992,000
20. PhoenixDark - 2,034,000
21. beermonkey@tehbias - 2,054,060
By percentages:
1. sonycowboy - 448.50
2. BuzzJive - 466.44
3. argon - 469.54
4. donny2112 - 483.13
5. Clever Pun - 487.95
6. GhaleonEB - 546.73
7. duderon - 556.16
8. Orgen - 558.79
9. JoshuaJSlone - 569.03
10. rage1973 - 591.32
11. Ganondorf>Link - 593.87
12. Kurosaki Ichigo - 608.76
13. starship - 628.74
14. Heidir - 681.15
15. Lock - 706.02
16. hadareud - 731.98
17. bycha - 732.42
18. djtiesto - 744.12
19. Earthstrike - 825.36
20. beermonkey@tehbias - 860.79
21. PhoenixDark - 952.06
Here's another option. This uses a modified version of the NASCAR Point System and is not limited to people who actually predicted all four months.
No, I'm not into NASCAR, but I've heard of their point system before.
#1 gets 183 points.
#2-#10 get 5 points less per position, so #2 gets 178, #3 gets 173, etc.
#11-#25 get 4 points less per position, so #11 gets 134, #12 gets 130, etc.
#26-#50 get 3 points less per position, so #26 gets 75, #27 gets 72, ..., #50 gets 3.
#51-whatever get zero points.
Basically, this removes the deadly penalty of missing a month (now it's as if you finished #51 or worse), and ranks you more on how you did compared to your peers that month than in an absolute sense. I like a points system best of the three, but it doesn't have to be this particular points setup.
Here are the top 21 through April using that method.
1. argon - 474
2. donny2112 - 454
3. BuzzJive - 392
4. Lock - 379
5. sonycowboy - 365
6. Clever Pun - 346
7. GhaleonEB - 329
8. DayShallCome - 321
9. JoshuaJSlone - 321
10. Orgen - 271
11. apujanata - 269
12. JJConrad - 261
13. Ganondorf>Link - 257
14. bycha - 248
15. duderon - 233
16. starship - 216
17. Kurosaki Ichigo - 207
18. rage1973 - 201
19. Bildi - 198
20. Heidir - 197
21. jet1911 - 192
To use apujanata as an example...
He didn't predict anything in January or February, which completely eliminates him from the first two rankings. However with the points setup, his #22 March rank and #5 April rank put him at #11 overall.
To use DayShallCome as an example...
He didn't predict anything in January, which completely eliminates him from the first two rankings. However with the point setup, his #5 February rank and #6 April rank (he finished #104 in March) put him at #8 overall.
Comments? Suggestions?
DayShallCome
Member
Though I'm obviously biased (see me as example B), I think the points system makes the most sense.
As you said, it makes it so that missing a month doesn't completely ruin a user (and if a n00b wanted to start predicting now, the points system would allow him to jump right into the competition).
As you said, it makes it so that missing a month doesn't completely ruin a user (and if a n00b wanted to start predicting now, the points system would allow him to jump right into the competition).
Points to consider for an official scoring system:
1) Somebody might be crap to begin with, but quickly get better and be a great predictor, but if their previous poor predictions are still counted, that would make it unfair on them as they would score low, even though they're now great at predicting. You're only as good as your latest prediction (however, one can be lucky too).
2) It's a different ballgame when predicting supply constained consoles, even if the console is in stock in a few places.
3) The percentage by which somebody mispredicted is more relevant than the actually amount, however, listing people in scores by percentages doesn't seem as satisfying to look at. Also, it's probably better if you get screwed over more for mispredicting the bigger systems.
-----
I think the official predictions chart should total the inaccuracies for the previous 3 months, in actual numbers (not percentages), and for those who have missed months, it should be extrapolated to include the extra month, then a 10% penalty should be added to their total for each month off (to combat those who just did one prediction and got lucky).
For instance, if they predicted the previous 2 months, but not the one before it, you add 10% to their total extrapolated total. If they were off 150k last month, 140k 2 months ago, you add 145k, then add another 10%. That makes 479k.
And if they only predicted one month, you times it by 3 then add 10% twice. So if they predicted 130k, their total becomes 472k.
Also, listing winners for individual months would be very appropriate too.
1) Somebody might be crap to begin with, but quickly get better and be a great predictor, but if their previous poor predictions are still counted, that would make it unfair on them as they would score low, even though they're now great at predicting. You're only as good as your latest prediction (however, one can be lucky too).
2) It's a different ballgame when predicting supply constained consoles, even if the console is in stock in a few places.
3) The percentage by which somebody mispredicted is more relevant than the actually amount, however, listing people in scores by percentages doesn't seem as satisfying to look at. Also, it's probably better if you get screwed over more for mispredicting the bigger systems.
-----
I think the official predictions chart should total the inaccuracies for the previous 3 months, in actual numbers (not percentages), and for those who have missed months, it should be extrapolated to include the extra month, then a 10% penalty should be added to their total for each month off (to combat those who just did one prediction and got lucky).
For instance, if they predicted the previous 2 months, but not the one before it, you add 10% to their total extrapolated total. If they were off 150k last month, 140k 2 months ago, you add 145k, then add another 10%. That makes 479k.
And if they only predicted one month, you times it by 3 then add 10% twice. So if they predicted 130k, their total becomes 472k.
Also, listing winners for individual months would be very appropriate too.
Parl said:
That's certainly doable.
Parl said:
No excuses.
If they correctly predict the supply-constrained months and correctly predict the non-supply-constrained months, just how awesome a predictor would they be? :lol Parl said:
I don't believe it is reasonable to extrapolate how much they would be off for a particular month. In keeping with how I've handled missing data in the past, it would be more appropriate to assume that since they didn't predict for that month, they must've thought no systems would have sold that month. For March, that would've put them 1.7 million (700%) in the hole. That's why I just excluded them altogether from the first two rankings.
Parl said:
Do you mean something new or just don't stop doing the rankings that are already done?
Just so it's clear, I'm not talking about stopping the rankings that are currently done. apujanata had brought forward the idea of keeping an overall ranking in addition to the current method. The current discussion is just about what way makes the most sense to do such an overall ranking.
Clever Pun
Member
wow, i've done consistently well so far this year! I didn't even realize I had predicted each month
I like the points system best. Speaking as an unbiased person (and I can because I rank worse in the points system than in apujanata's system) I like the points system because people are joining in all the time and they will therefore get some enjoyment out of the rankings just like people who have been doing it longer.donny2112 said:
Starting the count again every January seems good. Thanks for all the number crunching donny & apujanata.
Bildi said:I like the points system best. Speaking as an unbiased person (and I can because I rank worse in the points system than in apujanata's system) I like the points system because people are joining in all the time and they will therefore get some enjoyment out of the rankings just like people who have been doing it longer.
Starting the count again every January seems good. Thanks for all the number crunching donny & apujanata.
I agree with the points system for the same reason.
My suggestion :
Points system for the whole year (yearly leader board)
Points system for the last three months or the last six months (which will change every month)
The purpose of yearly leaderboard is to encourage sales enthusiast to keep on predicting each and every month, and the point system can encourage one to keep on doing it, even though they might have done badly in the previous 3 months. The point system helps minimize the effect of the badly done past prediction, and will make for a more heated competition (compared to total difference and percentage difference).
I did not know this, since I only started to do prediction in March 07 (my prediction in 05 and 06 sucks like a vacuum cleaner)DayShallCome said:
donny2112, you haven't post the whole list for this month.
And you forgot to mention this :
when there are more than one person who predict the exact same amount, the rank that will be used will be the highest available.
So, from this :
18. arne - 238K
19. Restricted_Area - 238K
20. Densha Otoko - 238K
21. Nuclear Muffin - 238K
22. apujanata - 238K
23. Freki - 239K
to this :
18. arne - 238K
18. Restricted_Area - 238K
18. Densha Otoko - 238K
18. Nuclear Muffin - 238K
18. apujanata - 238K
23. Freki - 239K
And you forgot to mention this :
when there are more than one person who predict the exact same amount, the rank that will be used will be the highest available.
So, from this :
18. arne - 238K
19. Restricted_Area - 238K
20. Densha Otoko - 238K
21. Nuclear Muffin - 238K
22. apujanata - 238K
23. Freki - 239K
to this :
18. arne - 238K
18. Restricted_Area - 238K
18. Densha Otoko - 238K
18. Nuclear Muffin - 238K
18. apujanata - 238K
23. Freki - 239K
Fuzzy said:And to actually answer the question, the second Thursday after the close of the month seen in that post.
Oh, yeah that. Um, thanks Fuzzy! :lol
apujanata said:
Sounds good to me.
One addition to the points system I was considering is a bonus for finishing first. Maybe an additional 15 or 30 points if you finish first that month?
That would mean that there would be a 1) monthly ranking based on units, 2) 3 month ranking based on points, and 3) yearly ranking based on points posted each month.
Sound about right?
apujanata said:
I usually only post the top 5/10 for the month. Last month was the only time I've posted the whole chart. I don't know that I'll always post the full results, but here's this month's, anyways.
apujanata said:
I do that manually when I post each month's unit results (i.e. the top 5/10). The method I'm using to figure out the points system automatically takes that into account when assigning points, though.
DayShallCome
Member
Sounds perfect. Thanks for all your hard work Donny, it makes NPD even more fun.
DayShallCome said:
Here are the point totals for the year and the last three months. I went ahead and gave the top finisher each month a 15-point bonus.
NPD Prediction Points Leaders (2007)
1. argon - 474
2. donny2112 - 454
3. BuzzJive - 407
4. sonycowboy - 380
5. Lock - 379
6. Clever Pun - 346
7. GhaleonEB - 329
8. DayShallCome - 321
8. JoshuaJSlone - 321
10. Orgen - 271
NPD Prediction Points Leaders (3 months)
1. DayShallCome - 321
2. argon - 306
2. donny2112 - 306
4. apujanata - 269
4. LOCK - 269
6. BuzzJive - 249
7. Clever Pun - 216
8. Bildi - 198
8. Xisiqomelir - 198
8. pilonv1 - 198
donny2112 said:
Here are the point totals for the year and the last three months. I went ahead and gave the top finisher each month a 15-point bonus.
NPD Prediction Points Leaders (2007)
1. argon - 474
2. donny2112 - 454
3. BuzzJive - 407
4. sonycowboy - 380
5. Lock - 379
6. Clever Pun - 346
7. GhaleonEB - 329
8. DayShallCome - 321
8. JoshuaJSlone - 321
10. Orgen - 271
NPD Prediction Points Leaders (3 months)
1. DayShallCome - 321
2. argon - 306
2. donny2112 - 306
4. apujanata - 269
4. LOCK - 269
6. BuzzJive - 249
7. Clever Pun - 216
8. Bildi - 198
8. Xisiqomelir - 198
8. pilonv1 - 198
Is Pachter ranked?
JoshuaJSlone
Member
With a points system, the specific amount you were off by each month ceases to matter; just how well you did compared to others. If we were to do something similar with home console sales (first gets 3, second gets 2, third gets 1, rest get 0) then from November-April things would look like...starship said:
PS2 - 15
Wii - 13
X360 - 8
While by units the order would be PS2, X360, Wii.
Yeah I know, but the difference in rankings is so big that I can't see this as the only reason.JoshuaJSlone said:
Two suggestions:
1. Constant difference between spots at any level.
2. The last person gets the lowest point, for example zero, and the #1 gets the highest point, for example 100. I think if for example 130 users participate in the prediction, its not fair that form #130 to #50 all get zero points.