BRF GJP Trial-2: How many seats will UPA win?

ramana · Post by **ramana** » 08 Apr 2014 01:26

A second poll item could be for the UPA tally. I have no clue about that one. but the ranges can start from < 100 and proceed in steps of 10. So something like:

Modify as needed. I edited this a couple of times to get the scaling right.. the median range is close to Nielsen.

Again give your range of probabilities in your post.
And post new version after each of the 7 rounds.
Thanks for participation.
Ramana

Patni · Post by **Patni** » 08 Apr 2014 01:41

(60%,30%,8%,2%,0%,0%,0%)

RamaY · Post by **RamaY** » 08 Apr 2014 01:50

[60%, 30%, 10%, 0%, 0%, 0%, 0%]

KJo · Post by **KJo** » 08 Apr 2014 01:56

1) Less than <100 - 75%
2) Between 101-110 - 10%
3) Between 111-120 - 5%
4) Between 121-130 - 4%
5) Between 131-140 - 3%
6) Between 141-150 - 2
7) Greater than > 150 - 1%

Vayutuvan · Post by **Vayutuvan** » 08 Apr 2014 02:19

[0%, 0%, 0%, 5%, 15%, 20%, 60%]

Saral · Post by **Saral** » 08 Apr 2014 02:54

[5%, 25%, 25%, 25%, 15%, 5%, 0%]

kish · Post by **kish** » 08 Apr 2014 03:07

1) Less than <100 - 35%
2) Between 101-110 - 30%
3) Between 111-120 - 25%
4) Between 121-130 - 7%
5) Between 131-140 - 3%
6) Between 141-150 - 0%
7) Greater than > 150 - 0%

krisna · Post by **krisna** » 08 Apr 2014 03:10

less than 100 ~50%
less than 150 >100%

Kanson · Post by **Kanson** » 08 Apr 2014 08:42

This is I want Cong I to have only 20 seats max., to seed the Decimation of Nehru-Sonia dynasty.

Klaus · Post by **Klaus** » 08 Apr 2014 09:40

<100: 70%
101-110: 15%
111-120: 10%
121-130: 5%
131-140: 0%
141-150: 0%
>150: 0%

Manu · Post by **Manu** » 08 Apr 2014 11:04

Below in CONGRESS PARTY only:
1) Less than <100 - 15%
2) Between 101-110 - 40%
3) Between 111-120 - 20%
4) Between 121-130 - 10%
5) Between 131-140 - 10%
6) Between 141-150 - 5%
7) Greater than > 150 - 0%

UPA will be close to 120-130.
I am calculating that Left Front, NC, and JD(U) will be wiped out in these elections, along with AAP.
MNS, JD (S), RJD, DMK, NCP, and SP will do very badly. Maya, Mamta and Jaya (in that order) will count, but not as much as is hoped by mainstream media.

Lilo · Post by **Lilo** » 08 Apr 2014 11:40

UPA (not satisfied as I don't seem to have a good grip on Congi numbers)

1) Less than <100 - 10%
2) Between 101-110 - 25%
3) Between 111-120 - 30%
4) Between 121-130 - 15%
5) Between 131-140 - 15%
6) Between 141-150 - 3%
7) Greater than > 150 - 2%

panduranghari · Post by **panduranghari** » 08 Apr 2014 12:32

90% between 100-110
10% above or below that number.

jamwal · Post by **jamwal** » 08 Apr 2014 13:12

1) Less than <100 - 0%
2) Between 101-110 - 10%
3) Between 111-120 - 10%
4) Between 121-130 - 15%
5) Between 131-140 - 20%
6) Between 141-150 - 20%
7) Greater than > 150 - 25%

kapilrdave · Post by **kapilrdave** » 08 Apr 2014 15:15

(80%,15%,5%,0%,0%,0%,0%)

merlin · Post by **merlin** » 08 Apr 2014 16:29

1) Less than <100 - 5%
2) Between 101-110 - 35%
3) Between 111-120 - 50%
4) Between 121-130 - 10%
5) Between 131-140 - 0%
6) Between 141-150 - 0%
7) Greater than > 150 - 0%

Denis · Post by **Denis** » 08 Apr 2014 16:43

UPA should get 95-120. 130 at Max ... by a long stretch. Congress should be in 70-80 range

saadhak · Post by **saadhak** » 08 Apr 2014 20:32

For UPA:
1) Less than <100 - 10%
2) Between 101-110 - 15%
3) Between 111-120 - 15%
4) Between 121-130 - 15%
5) Between 131-140 - 15%
6) Between 141-150 - 20%
7) Greater than > 150 - 10%

RajeshA · Post by **RajeshA** » 08 Apr 2014 21:14

Congress would be getting around 50. UPA tally would be 15 more.

muraliravi · Post by **muraliravi** » 08 Apr 2014 22:14

Congress on its own will get 115-120 seats. Allies (Kerala Allies-2, NCP-10, RJD-11, Others - 4) will get around 25-30 seats.

So UPA will get 140-150 seats.

Saral · Post by **Saral** » 08 Apr 2014 22:27

muraliravi, rajeshA: If you had a 100 rupees to bet, would you put all of it in one small range 141-150? And none for the others? Are you sure?? If you can, please provide the probability estimates for the provided ranges (see Sadhak above for example) so that we can use your inputs. Clearly your (muraliravi) number for 141-150 could be high (say 35%) but the neighboring ranges (> 150 and 131-140) would also be non-zero. Just to give an idea. You don't need to use my example numbers. Just make sure they add to 100.

Denis · Post by **Denis** » 08 Apr 2014 23:00

For UPA:
1) Less than <100 - 15%
2) Between 101-110 - 25%
3) Between 111-120 - 50%
4) Between 121-130 - 05%
5) Between 131-140 - 0%
6) Between 141-150 - 0%
7) Greater than > 150 - 0%

muraliravi · Post by **muraliravi** » 09 Apr 2014 19:28

Saral wrote:muraliravi, rajeshA: If you had a 100 rupees to bet, would you put all of it in one small range 141-150? And none for the others? Are you sure?? If you can, please provide the probability estimates for the provided ranges (see Sadhak above for example) so that we can use your inputs. Clearly your (muraliravi) number for 141-150 could be high (say 35%) but the neighboring ranges (> 150 and 131-140) would also be non-zero. Just to give an idea. You don't need to use my example numbers. Just make sure they add to 100.

Sir, no offense, but what is the point in predicting like that, putting 15 or 20% in all slots and then claiming that you are right whatever the result is, because you had bets in all ranges. I am basing my numbers on hard realities and stick with that prediction. But if you do want a broader range, i can give one. UPA will get a minimum of 130 and a max of 160 seats.

chaanakya · Post by **chaanakya** » 09 Apr 2014 20:10

Well UPA would be in two digits , more likely 70-90 range. AP is wash out. If it passes three digit mark I would be surprised.

ramana · Post by **ramana** » 09 Apr 2014 20:30

muraliravi, there is a difference in GJP methodology and a simple poll. In a simple poll one puts probability of 1.0 in one bucket. GJP allows a range of probabilities and adjusting with time and information. We are trying to pilot the methodology with these two timebound trials. I was thinkign last nigh that thinking in a range of probabilities is tough but once one gets used to it becomes second nature. Essentially its hedging like allocating your retirement funds in an account.

SaiK · Post by **SaiK** » 09 Apr 2014 23:54

Should'nt we get this voting done by only UPA supporters within BRF?

meaning, is this data biased?

ramana · Post by **ramana** » 10 Apr 2014 01:53

No. Its a sense of how folks can see.

Are you going to post your estimates or just drive by?

Vayutuvan · Post by **Vayutuvan** » 10 Apr 2014 02:49

Are you going to post your estimates or just drive by?

I have modified estimate now:

List of list lists.The first in the list is the brf userid, seond and third are lists of lists. First is the estimate number and the second is a list of the estimates. This is amenable for python processing as these are Py lists (or lisp).

[matrimc,[1,[0%, 2%, 3%, 5%, 20%, 40%, 30%]],[0,[0%, 0%, 0%, 5%, 15%, 20%, 60%]]]

SaiK · Post by **SaiK** » 10 Apr 2014 03:31

you mean 0th is the first list element in your list of lists? but, the elements in lowest level of your list node does not have a label to understand snake ji. i have to ass-u-me 30% and 60% are the <100 for UPA, and your estimate is 30%? or you reverse the ordered list?

--

going by the order given by the majority pattern, this is mine:

klaus++

ramana · Post by **ramana** » 10 Apr 2014 03:58

SaiK wrote:you mean 0th is the first list element in your list of lists? but, the elements in lowest level of your list node does not have a label to understand snake ji. i have to ass-u-me 30% and 60% are the <100 for UPA, and your estimate is 30%? or you reverse the ordered list?

--

going by the order given by the majority pattern, this is mine:

klaus++

Saik, We need a decoder!

muraliravi · Post by **muraliravi** » 10 Apr 2014 06:48

ramana wrote:muraliravi, there is a difference in GJP methodology and a simple poll. In a simple poll one puts probability of 1.0 in one bucket. GJP allows a range of probabilities and adjusting with time and information. We are trying to pilot the methodology with these two timebound trials. I was thinkign last nigh that thinking in a range of probabilities is tough but once one gets used to it becomes second nature. Essentially its hedging like allocating your retirement funds in an account.

Got it Sir, I'll put all my bets in that portfolio or at best hedge it very little, because I know returns are going to be highest there. I believe in this concept, at the end of the day, there is one spectrum that will yield returns and the rest will yield nothing. That I believe is how I should be approaching this problem since I am quite confident of how congress will perform in these elections. Whatever the washout, I just dont see congress getting below 110 seats.

So I will say
130-140 seats for UPA at 10%
140-150 at 80%
>150 seats at 10%.

disha · Post by **disha** » 10 Apr 2014 08:17

Okay getting the hang of it now,

1) Less than <100 10%
2) Between 101-110 30%
3) Between 111-120 40% <= This is what I think UPA will end up in
4) Between 121-130 20%
5) Between 131-140 10% <= outlier. Since UPA is relying only in Mah, Ker, Karn, AP (YSRC) and NE.
6) Between 141-150
7) Greater than > 150

ramana · Post by **ramana** » 11 Apr 2014 01:45

Locking for data extraction. Will repone 7:00pm PST.

ramana · Post by **ramana** » 11 Apr 2014 07:45

X-Post...
Week 1 UPA

Table 1 is sorted in order of mean estimates (ESTIMATE) from each of the 21 distributions (rows). I use the center of each range as the weighting factor for each frequency (so 105.5 for the range 101 to 110). I used 85 and 165 as weights for the end ranges.

Table 2 is sorted by CONFIDENCE inferred from each distribution. A perfectly uniform distribution indicates zero confidence (extreme hedging). At the other extreme, a confidence of 100 is assigned to people who put everything in 1 bin (zero hedging). Calculation: Each column in a row has an entropy term -p*log(p). Sum these up for each row and take the exponential of the entropy (sum of terms within row). Subtract from 7, divide by 6, and multiply by 100 to map the score to the range 0 to 100.

ESTIMATE has mean=108, median=105, and SD=20. CONFIDENCE has mean=63, median=68, and SD=28. The correlation between deviation of ESTIMATE from the mean (of 108) and the CONFIDENCE score is modest at 0.26. If we weigh each ESTIMATE by the CONFIDENCE, then the weighted estimate is 104.

Table 1

Code: Select all

          JINGO      UPA85 UPA105 UPA115 UPA125 UPA135 UPA145 UPA165    ESTIMATE    CONFIDENCE
        
        Kanson      100      0      0      0      0      0      0          85           100
       RajeshA      100      0      0      0      0      0      0          85           100
     chaanakya      100      0      0      0      0      0      0          85           100
    kapilrdave       80     15      5      0      0      0      0          90            86
         Klaus       70     15     10      5      0      0      0          93            75
       KJoishy       75     10      5      4      3      2      1          94            73
         Patni       60     30      8      2      0      0      0          94            74
         RamaY       60     30     10      0      0      0      0          94            76
          kish       35     30     25      7      3      0      0         103            51
        krisna       50     10     10     10     10     10      0         105            42
 panduranghari        5     90      5      0      0      0      0         105            92
         Denis       15     25     50      5      5      0      0         110            57
          Manu       15     40     20     10     10      5      0         111            35
         disha       10     30     40     20      0      0      0         111            57
        merlin        5     35     50     10      0      0      0         111            67
          Lilo       10     25     30     15     15      3      2         116            26
         Saral        5     25     25     25     15      5      0         118            32
       saadhak       10     15     15     15     15     20     10         126             3
        jamwal        0     10     10     15     20     20     25         138            22
    muraliravi        0      0      0      0     10     80     10         146            85
       matrimc        0      0      0      5     15     20     60         155            68

Table 2

Code: Select all

          JINGO      UPA85 UPA105 UPA115 UPA125 UPA135 UPA145 UPA165    ESTIMATE    CONFIDENCE
          
       saadhak       10     15     15     15     15     20     10         126             3
        jamwal        0     10     10     15     20     20     25         138            22
          Lilo       10     25     30     15     15      3      2         116            26
         Saral        5     25     25     25     15      5      0         118            32
          Manu       15     40     20     10     10      5      0         111            35
        krisna       50     10     10     10     10     10      0         105            42
          kish       35     30     25      7      3      0      0         103            51
         Denis       15     25     50      5      5      0      0         110            57
         disha       10     30     40     20      0      0      0         111            57
        merlin        5     35     50     10      0      0      0         111            67
       matrimc        0      0      0      5     15     20     60         155            68
       KJoishy       75     10      5      4      3      2      1          94            73
         Patni       60     30      8      2      0      0      0          94            74
         Klaus       70     15     10      5      0      0      0          93            75
         RamaY       60     30     10      0      0      0      0          94            76
    muraliravi        0      0      0      0     10     80     10         146            85
    kapilrdave       80     15      5      0      0      0      0          90            86
 panduranghari        5     90      5      0      0      0      0         105            92
        Kanson      100      0      0      0      0      0      0          85           100
       RajeshA      100      0      0      0      0      0      0          85           100
     chaanakya      100      0      0      0      0      0      0          85           100

Yagnasri · Post by **Yagnasri** » 11 Apr 2014 09:18

Average seats mafia won in the last 5 elections except 2009 are 130-140. I think it may fall below that this time. 100-110 this time considering the decimation in AP and 10 years of loot record.

Denis · Post by **Denis** » 11 Apr 2014 18:38

Week 2 GJP Trial

Denis: 30 60 10 0 0 0

SaiK · Post by **SaiK** » 11 Apr 2014 20:47

<100: 80%
101-110: 10%
111-120: 5%
121-130: 4%
131-140: 3%
141-150: 2%
>150: 1%

Hari Seldon · Post by **Hari Seldon** » 11 Apr 2014 22:56

Week 2 GJP Trial

Hari: 0 10 60 30 0 0

saip · Post by **saip** » 11 Apr 2014 23:44

(50%,30%,15%,5%,0%,0%,0%)

chaanakya · Post by **chaanakya** » 11 Apr 2014 23:50

After 10th April

100%,0,0,0,0,0..

No way Congis are getting beyond two digit figure.