7/24/2004 KERRY: 98.4% WIN PROBABILITY with 331 EV and 53.16% of vote

[TruthIsAll]

IT'S NOT CLOSE!
									
Projections and EV Simulation Models created by
TruthIsAll									
National Polling data: Pollingreport.com									
State Polling data: Electoral-Vote.com 									
									
									
COMBINED NATIONAL AND STATE PROJECTION									
Kerry:	53.43%								
Bush:	46.57%								
Spread	6.87%								
									
NATIONAL PROJECTION MODEL									
National polling data from pollingreport.com
includes:									
IBD, ABC, NWK, ARG, NBC, CBS, PEW, LAT, ZOGBY, TIME									
									
Kerry	53.71%					
Bush	46.29%					
Spread	7.42%					
						
STATE ELECTORAL VOTE SIMULATION MODEL						
						
Most Likely Vote%;  Electoral votes						
Kerry 	53.16%	331				
Bush	46.84%	207				
Spread	6.31%	124				
						
Kerry wins	984	of 1000 simulation trial runs, 				
or a	98.4%	probability of winning the election.				
						
						
ELECTORAL VOTE SENSITIVITY ANALYSIS						
			
Data is downloaded from electoral-vote.com to Excel for the
forecasting model.			
One thousand  trials are run in each simulation.			
			
Kerry Forecast Statistics:			
Simulation I: Conservative Case 			
Assume	60%	of undecided/other votes are allocated to Kerry.	
Receives	52.1%	of the vote.	
Wins	91.1%	of the trials 	
Average	310	electoral votes.	
Maximum 	416	electoral votes.	
			
			
Simulation II: Most Likely Case 			
Assume	70%	of undecided/other votes are allocated to Kerry.	
Receives	53.2%	of the vote.	
Wins	98.4%	of the trials 	
Average	331	electoral votes.	
Maximum 	423	electoral votes.	
			
			
Simulation III: Optimistic Case 			
Assume	80%	of undecided/other votes are allocated to Kerry.	
Receives	54.2%	of the vote.	
Wins	99.9%	of the trials 	
Average	352	electoral votes.	
Maximum 	434	electoral votes.	
			
			


LAST THREE PRESIDENTIAL ELECTIONS			
(in millions of votes)												
Dem	138.75	52.6%										
Rep	125.03	47.4%										

												

BUSH JOB APPROVAL												
The data source is PollingReport.com    												
												
2004	Avg	Nwk	Fox	CNN	Pew	Harris	CBS	ABC	Time	NBC	AP	Zogby
Jan.	54.50	50	58	60	56	na	50	58	54	54	56	49
Feb.	49.67	48	48	51	48	51	50	50	54	na	47	na
Mar.	48.75	48	48	49	46	na	51	50	na	50	48	na
Apr.	48.80	49	50	52	48	48	46	51	49	na	48	47
May	45.20	42	48	47	44	na	41	47	46	47	48	42
June	47.00	na	48	49	48	50	42	47	na	45	48	46
July	48.00	48	47	49	46	na	45	na	50	48	50	49

												
												
												
												
NATIONAL POLL PROJECTION METHODOLOGY												
												
The data source is PollingReport.com    												
Monthly polling data from: 												
IBD, ABC, NWK, ARG, NBC, CBS, PEW, LAT, ZOGBY,
TIME												
												
The projection is based on the latest monthly average of the
ten polls												
adjusted by allocating  70% of undecided/other voters to
Kerry.												
												
												
`						
	10 Poll Average	Projection	
	Kerry	Bush 	Diff	Kerry	Bush 	Diff
Jan 	42.5	50.2	-11.5	46.5	53.5	-7.0
Feb	48.0	45.4	2.6	52.6	47.4	5.2
Mar	48.1	44.4	3.8	53.4	46.6	6.8
Apr	47.4	44.9	2.5	52.8	47.2	5.6
May	47.1	44.2	2.9	53.2	46.8	6.4
June	47.1	45.0	2.1	52.6	47.4	5.3
July	48.6	44.1	4.5	53.7	46.3	7.4
						
						
Kerry National Poll Projection Formula:						
10-Poll Avg 	48.6					
+ allocation	5.1					
= Projection:	53.7	% of total vote.				
													
				Kerry:  Undecided/other allocation									
	Latest Polls		60%	60%	70%	70%	80%	80%					
	Kerry	Bush 	Proj %	P (win)	Proj %	P (win)	Proj %	P (win)					
													
CNN	49.0	47.0	51.1	64.3%	51.8	72.6%	52.2	76.8%					
AP	45.0	49.0	50.6	57.3%	49.2	39.5%	49.8	47.3%					
FOX	45.0	44.0	52.4	78.8%	52.7	81.6%	53.8	89.7%					
													
	46.3	46.7	51.4	67.4%	51.2	66.0%	51.9	74.0%					
													
													
IBD	44.0	41.0	52.6	80.7%	54.5	93.3%	56.0	97.7%					
ABC	46.0	46.0	48.5	30.9%	51.6	70.3%	52.4	78.8%					
NWK	51.0	45.0	54.8	94.7%	53.8	89.7%	54.2	91.9%					
ARG	49.0	45.0	53.0	84.1%	53.2	85.7%	53.8	89.7%					
NBC	54.0	43.0	58.4	99.7%	56.1	97.9%	56.4	98.4%
CBS	49.0	44.0	54.3	92.4%	53.9	90.3%	54.6	93.7%
PEW	46.0	44.0	50.2	52.7%	53.0	84.1%	54.0	90.9%
LAT	48.0	46.0	51.2	65.5%	52.2	76.8%	52.8	82.5%
ZOGBY	46.0	44.0	52.0	74.8%	53.0	84.1%	54.0	90.9%
TIME	50.0	45.0	52.0	74.8%	53.5	87.8%	54.0	90.9%
								
AVG	48.3	44.3	52.7	91.1%	53.5	95.9%	54.2	98.3%
								
								
								
								
								
								
								
								
								
ELECTORAL VOTE SIMULATION FORECAST METHODOLOGY								

Kerry's vote % is weighted by each state's percentage of the
total vote in the last 3 elections..								

Kerry's projected % for each state is based on the latest
state poll, adjusted								
by the allocation of undecided/other voters.								

The probability of Kerry winning a state is determined by the
current poll spread: Kerry - Bush. 								
The spread is comapred to the MoE.								
The greater the spread, the higher the probability that Kerry
will win the state.								

Most Likely Case								
Assume	70%	Undecided/other to Kerry						
Win	Prob.	Pct	EV					
Kerry	98.4%	53.2%	331					
Bush	1.6%	46.8%	207										

State Electoral Vote Simulation: Historical Voting, Kerry
Projection and Win Probability 													
Note: Dem% is Democratic % of the two-party vote in the last
three pres. elections													
													
Hist%	Dem%	Kerry%	Kerry%	EV Trials runs 1-10 of 1000									
Kerry	Vote	Proj.	Prob.	1	2	3	4	5	6	7	8	9	10
Proj.	52.60%	53.16%	Win	351	372	278	309	305	393	275	390	326	363
													
AL	44.8	45.0	10.6		9	9							
AK	37.6	40.7	1.0										
AZ	48.8	53.9	83.5	10	10		10	10	10	10	10	10	10
AR	55.2	50.6	56.0	6	6		6	6	6		6	6	
CA	57.4	58.1	97.9	55	55		55	55	55	55	55	55	55
													
CO	48.8	49.3	43.1				9		9	9	9	9	9
CT	57.7	62.6	99.9	7	7	7	7	7	7	7	7	7	7
DE	56.8	57.1	96.2	3	3	3	3	3	3		3	3	3
DC	90.3	90.4	100.0	3	3	3	3	3	3	3	3	3	3
FL	50.7	51.7	66.5		27				27		27	27	27
													
GA	47.6	45.7	14.1	15	15								
HI	59.0	60.9	99.7	4	4	4	4	4	4	4	4	4	4
ID	35.7	39.0	0.3										
IL	57.9	60.0	99.4	21	21	21	21	21	21	21	21	21	21
IN	44.8	45.9	15.3						11				
													
IA	51.8	51.9	68.3	7		7	7	7			7		7
KS	42.4	41.6	1.8	6									
KY	46.7	45.3	12.0										
LA	49.2	49.0	40.1		9	9		9			9		
ME	57.1	52.3	71.7		4	4	4	4	4		4	4	4
													
MD	57.8	58.6	98.4	10	10	10	10	10	10	10	10	10	10
MA	65.3	66.7	100.0	12	12	12	12	12	12	12	12	12	12
MI	54.7	53.0	77.3	17	17	17		17	17		17		
MN	55.7	53.0	77.3	10		10		10	10	10	10	10	10
MS	44.3	36.3	0.0										
													
MO	52.5	53.0	77.3	11	11	11	11	11	11	11	11		11
MT	44.9	42.8	3.6										
NE	37.5	36.5	0.0										
NV	49.9	53.2	78.8	5	5	5	5	5	5	5	5	5	5
NH	51.7	52.6	74.2	4	4			4	4		4		4
													
NJ	56.5	56.6	95.1	15	15	15	15		15	15	15	15	15
NM	53.0	55.3	90.7	5	5		5	5	5	5		5	5
NY	62.6	65.0	100.0	31	31	31	31	31	31	31	31	31	31
NC	46.6	48.9	39.2	15			15	15	15		15	15	15
ND	40.8	37.2	0.1										
													
OH	50.8	52.6	74.2	20	20	20	20		20		20		20
OK	42.8	39.2	0.3										
OR	53.6	55.6	91.9	7	7	7	7	7	7	7	7	7	7
PA	54.2	57.8	97.4		21	21	21	21	21	21	21	21	21
RI	65.6	64.2	100.0	4	4	4	4	4	4	4	4	4	4
													
SC	44.4	47.5	26.6		8				8		8		8
SD	44.5	44.8	9.7						3				
TN	50.5	50.8	57.9	11		11			11	11	11		11
TX	44.3	42.6	3.2										
UT	33.6	29.7	0.0										
													
VT	59.4	60.1	99.4	3	3	3	3	3	3	3	3	3	3
VA	47.3	48.5	35.4	13		13						13	
WA	55.9	55.1	89.9	11	11	11	11	11	11	11	11	11	11
WV	54.0	47.2	24.2		5							5	
WI	52.7	55.0	89.4	10	10	10	10	10	10	10	10	10	10
WY	38.3	30.8	0.0										
													
Avg	52.60%	53.16%	-	351	372	278	309	305	393	275	390	326	363

[MissouriTeacher]

1. I've seen such projections before...

and while they're certainly encouraging, keep in mind they only reflect the current data.

All they're really capable of telling you is that if the election were held NOW Kerry would win. That's good news of course, but it's a long way to November. There are a lot of things that could happen between now and then. It's perfectly fine to be optimistic, but we shouldn't act as if we've already won.

[Rowdyboy]

2. A bit too optimistic for me...He's giving us Tennessee, Arkansas

and Florida as near certainties. I'm not willing to go quite that far yet. Still, Itsalways interesting to read and helps keep me optimistic.

[TruthIsAll]

5. NO. LOOK AT THE INDIV STATE PROBABILITIES.

All three states are under 70% probability, hardly a near certainty.

AR 56%
FL 66.5%
TN 57.9%

[Rowdyboy]

7. I'm not entirely sure I understand....

Is he saying there is a 57.9% probability that Kerry will take Tennessee?

[TruthIsAll]

8. That's what I'm saying. Look at the latest poll numbers for TN.

They are dead even, which means Kerry is the most likely winner, based on the 70% allocation of other voters to him.

[Rowdyboy]

9. Okay, I follow what you're saying....

I wasn't factoring in the undecideds.

[DerBeppo]

3. 70% undecided to Kerry

this seems to be based off that assumption, doesn't seem to explain why either.

any reason why it's at 70 and not 50 or 62?

[TruthIsAll]

6. I SHOW IT FOR 60,70,80%. historically its 60-80% to the challenger

I took the mid-point 70%.

It is very conservative. Other includes undecided and Nader votes.

[VOX]

4. I like these numbers, but I want to play this one like we're behind...

There's so much to lose, that losing is unthinkable.

[Nexus7]

10. Methodology

I'm not understanding why you have to run a simulation. Why not just take the polling data for each state, add (probability of voting for challenger X % undecided) and add that result over all the states?

A simulation is useful if you have a system with multiple stochastic inputs and a complex or unknown transfer function or mechanism. But here every state is independent since the votes in each state decide only its own result.

[TruthIsAll]

11. Here is why I run a simulation.

There are many states which could go either way. We want to be able to determine what would happen if we ran 1000 combinations of outcomes (electoral votes) knowing the probability of winning each state based on the latest polls.

Lets assume the following election between candidates A and B in a country in which there are only three states.

State 1: 90% prob of a win (10 EV) for candidate A
State 2: 51% prob of a win (8 EV) for A
State 3: 51% prob of a win for (17 EV) for B

If we just add up the states where A leads, then he would always be a winner with 18EV vs. 17EV for B.

But we really want to see what the probability is that A wins the election, because we cannot just assume that he will win States 1 and State 2 and lose state 3. So we use Monte Carlo simulation and run it 1000 times (trial elections). We add up the electoral votes for the states that A wins in each trial. If it is greater than 269, A wins that trial run.

For example, if A wins 520 trials, the probability that A will win the election is 52%. Candidate B would then have a 48% probability.

[Nexus7]

12. Results without simulation

OK, I see why you do this, but ...
Since you start off the simulation for each state with given win and undecided voters' breaking probabilities, the outcome is fixed. You can run 1000 simulations, but the result you get is given by the following.

P(A_win) = P(A_poll) + P(undecided)*P(undecided_breaking_for_A)

If this is > P(B_win), then A wins that state and gets the EV.

Over all states (say N in number), you have only 2^^N possible
outcomes. Using your example, you can calculate the probability of each outcome.

P(A gets 35 EV) = P(A_wins_1)*P(A_wins_2)*P(A_wins_3)
= 0.90*0.51*0.49
P(A gets 18 EV) = P(A_wins_1)*P(A_wins_2)*P(A_loses_3)
= 0.90*0.51*0.51

Then you add up the probabilities of all the outcomes that give A a winning number of EV.

[TruthIsAll]

13. Not quite. I use random numbers in a Monte Carlo simulation.

The outcome is not fixed. It varies for each simulation trial run, based on a random number applied to the Cum. probability function for each state.

Lets take a simple case:

Asume Kerry leads 46-44 in the polls in a given state. Applying the 70% allocation to the other 10%, Kerry is thus projected to take the state 53-47.

Assume that with a 53-47 split, the probability is 95% that Kerry wins the state, assuming a MoE of +/-4%.

We generate a random number RND between 0 and 1. If RND <=.95, Kerry wins the state; otherwise Bush wins. We do the same for each state.

That's it. We total up the electoral votes for all the states which Kerry "wins" in this trial run.

We run 1000 trials. We sum the X trials where Kerry has won more than 269 EV. Then we say Kerry has a X % chance of winning the election.