2004 Projected Presidential Electoral Votes as of 6/21/2004

TradeSports.com ^ | Monday, June 21, 2004 | Momaw Nadon

[Momaw Nadon]

State	% Chance of Bush Winning	Bush Electoral Votes	Kerry Electoral Votes
Alabama	98.8	9	0
Alaska	92.6	3	0
Arizona	69.9	10	0
Arkansas	73.9	6	0
California	14.2	0	55
Colorado	76.1	9	0
Connecticut	12.4	0	7
Delaware	22.8	0	3
District of Columbia	1.0	0	3
Florida	63.9	27	0
Georgia	91.1	15	0
Hawaii	7.1	0	4
Idaho	96.9	4	0
Illinois	14.9	0	21
Indiana	94.9	11	0
Iowa	41.9	0	7
Kansas	95.9	6	0
Kentucky	87.1	8	0
Louisiana	84.9	9	0
Maine	18.1	0	4
Maryland	11.1	0	10
Massachusetts	1.2	0	12
Michigan	36.7	0	17
Minnesota	32.9	0	10
Mississippi	94.1	6	0
Missouri	60.1	11	0
Montana	94.9	3	0
Nebraska	95.1	5	0
Nevada	64.8	5	0
New Hampshire	53.0	4	0
New Jersey	18.1	0	15
New Mexico	53.1	5	0
New York	10.0	0	31
North Carolina	76.1	15	0
North Dakota	94.1	3	0
Ohio	59.1	20	0
Oklahoma	94.1	7	0
Oregon	45.1	0	7
Pennsylvania	46.2	0	21
Rhode Island	4.1	0	4
South Carolina	92.9	8	0
South Dakota	97.8	3	0
Tennessee	84.1	11	0
Texas	98.8	34	0
Utah	98.8	5	0
Vermont	7.9	0	3
Virginia	81.9	13	0
Washington	23.1	0	11
West Virginia	57.9	5	0
Wisconsin	43.9	0	10
Wyoming	98.8	3	0
Totals		283	255

[Momaw Nadon]

I went to TradeSports.com which is a futures market in which people bet on the outcomes of various things.

According to the current trading prices of the futures contracts, an estimate can be found of what traders are betting will be the outcome of 2004 Presidential Election.

If the traders are correct, President Bush would receive 283 Electoral Votes and John Kerry would receive 255 Electoral Votes.

If the weighted probabilities of President Bush winning in the states are added up, and then divided by 538, and multiplied by 100, then President Bush should get 283.96 Electoral Votes.

270 Electoral Votes are needed to win the Presidency.

Opinions and commentary are welcome.

[Momaw Nadon]

Previous projections:

2004 Projected Presidential Electoral Votes as of 6/14/2004
2004 Projected Presidential Electoral Votes as of 6/7/2004
2004 Projected Presidential Electoral Votes as of 5/31/2004
2004 Projected Presidential Electoral Votes as of 5/24/2004
2004 Projected Presidential Electoral Votes as of 5/17/2004
2004 Projected Presidential Electoral Votes as of 5/10/2004
2004 Projected Presidential Electoral Votes as of 5/3/2004
2004 Projected Presidential Electoral Votes as of 4/26/2004
2004 Projected Presidential Electoral Votes as of 4/19/2004
2004 Projected Presidential Electoral Votes as of 4/12/2004
2004 Projected Presidential Electoral Votes as of 4/5/2004
2004 Projected Presidential Electoral Votes as of 3/29/2004
2004 Projected Presidential Electoral Votes as of 3/22/2004
2004 Projected Presidential Electoral Votes as of 3/15/2004
2004 Projected Presidential Electoral Votes as of 3/8/2004
2004 Projected Presidential Electoral Votes as of 3/1/2004
2004 Projected Presidential Electoral Votes as of 2/23/2004
2004 Projected Presidential Electoral Votes as of 2/16/2004
2004 Projected Presidential Electoral Votes as of 2/9/2004
2004 Projected Presidential Electoral Votes as of 2/2/2004
2004 Projected Presidential Electoral Votes as of 1/26/2004
2004 Projected Presidential Electoral Votes as of 1/21/2004

[Momaw Nadon]

[Momaw Nadon]

[Momaw Nadon]

If you want on (or off) of the weekly TradeSports.com Projected Presidential Electoral Vote ping list, please let me know via FREEPmail.

[freedom1st]

I think I'll put more credence in this poll than Rasmussen. Money talks and bull jive walks.

[CyberAnt]

Plus .. we need to remember that the Rasmussen poll is DAILY .. and will flucuate all over the place.

[Remember_Salamis]

put me on the tradesports list.

[supercat]

One major problem with this sort of calculation is that it presumes the 50 state outcomes are independent when they clearly aren't; many events which would affect any state outcome are likely to affect many.

Consider, for example, a "contest" with three states (Freedonia, Pottsylvania, and Wackyland) and two candidates (Alex and Bob). Alex has a 66.7% chance of winning in Fredonia, and a 66.7% chance of winning in Pottsylvania. He only has a 33.3% chance of winning in Wackyland. What is the probability of his winning at least two states?

If the events were independent, the probability would be 16/27 [8/27 of Alex winning the first two and not the third; 2/27 of his winning the first and third making and not the second; 2/27 of the winning second and third and not the first; 4/27 of winning all three].

If the events are not independent, though, the probability can range anywhere from 1/3 to 2/3. To see why, imagine that a die will be rolled at midnight on June 22. Consider the following two scenarios:

1. Freedonia votes for Alex if the die roll is 1-4. Pottsylvania votes for Alex if the die roll is 3-6. Wackyland votes for Alex if the roll is 3-4. In this case, the only way Alex wins is with a roll of 3-4, probability 1/3.
2. Freedonia votes for Alex if the die roll is 1-4. Pottsylvania votes for Alex if the die roll is 2-5. Wackyland votes for Alex if the roll is 1 or 5. In this case, Alex will win if the die roll is anything other than six, probability 5/6.

As you can see, innterdependence of events can make a huge difference in the probabilities of combinations of events occurring.

[Rome2000]

Wackyland

Thats down the road from Stoopidsville.

[supercat]

Thats down the road from Stoopidsville.

Sorry, I was having trouble thinking up names for the fictitious states. The essential nature of the probabilties remains, though.

[mkj6080]

If you take the probabilites multiple by the electoral votes one gets:

Expected Bush Electoral Votes: 284
Expected Kerry Elec Votes: 254

This is analgous to saying if I flip a coin and win $10000 if heads or $0 if tails then the expected value of the gamble is:

50%*$10000 + 50%*$0 = $5000

[supercat]

This is analgous to saying if I flip a coin and win $10000 if heads or $0 if tails then the expected value of the gamble is: 50%*$10000 + 50%*$0 = $5000

Yes, but what matters is not the "expected" number of electoral votes, but rather the probability of winning a majority of the electoral votes.

Consider the following scenario:

• Candidate A has a 90% probability of winning 51% of the electoral votes, and a 10% probability of winning zero.
• Candidate B has a 10% probability of winnig 100% of the electoral votes and a 90% probability of winning 49%.

The "expected" number of votes for candidate A is only 45.9%; candidate B "expects" to receive 54.1. That would tend to suggest that candidate B would have the advantage. The actual probability of candidate A winning, however, is 90%--an overwhelming advantage.

[Political Junkie Too]

But why isn't that accounted for in a 50-node symmetrical decision tree? All possible combinations would be accounted for, with their probabilities of occuring taken into account.

I still can't get it in my head why each state's electoral vote winner isn't considered an independent event. It's not like the voters of one state wait to see how another state voted before voting themselves. Are you suggesting that there is some event that ties the outcomes of separate states together into one dependent outcome?

-PJ

[supercat]

I still can't get it in my head why each state's electoral vote winner isn't considered an independent event. It's not like the voters of one state wait to see how another state voted before voting themselves. Are you suggesting that there is some event that ties the outcomes of separate states together into one dependent outcome?

Precisely. The presidential race outcome in each state will be affected by some factors unique to that state, but in most cases it will be more strongly affected by factors which will affect many states.

For example, if the economy continues to do well, this will improve Bush's chances for reelection in almost every state. If it suddenly starts doing poorly, this will likely hurt his chances. if we catch Osama Bin Laden, this will improve Bush's chances. If it were (hypothetically) discovered that Bush personally intervened to prevent Atta et al. from being deported prior to 9/11, that would hurt his chances.

Although there is a certain amount of statistical randomness in election results, and the 'swing' states' results may be due as much to chance as to outside factors, almost any event which would cause Bush or Kerry to lose a 'safe' state would also ensure that he lost all or nearly all of the swing states as well.

[codercpc]

Please add me to the ping list.

Thanks

[Political Junkie Too]

For example, if the economy continues to do well, this will improve Bush's chances for reelection in almost every state. If it suddenly starts doing poorly, this will likely hurt his chances. if we catch Osama Bin Laden, this will improve Bush's chances...

Then what are the state probabilities depicting, if not the aggregate effect of all these events on the voters' perceptions on a state-by-state basis?

Why wouldn't one assume that what makes a state lean strongly or weakly one way or the other is just what you say, but just not that Texas voted one way because Massachusetts voted the other, or vice versa?

When I think of dependent events, I assume dependency in the uncertain events. In this case, the uncertain event is which way a state's electoral vote will go, not how voters in one state feel about the economy or the war, versus voters in another state. I haven't bought into the idea that the individual state results are not independent events.

-PJ

[supercat]

When I think of dependent events, I assume dependency in the uncertain events. In this case, the uncertain event is which way a state's electoral vote will go, not how voters in one state feel about the economy or the war, versus voters in another state. I haven't bought into the idea that the individual state results are not independent events.

Let me put it to you this way: what would have to happen for Bush to lose, e.g. Texas?

If such an event were to occur, would you still expect Bush to have a 59% chance of winning Ohio?

The only way Bush is going to lose a state like Texas is if something happens that is very bad for him. Alnmost anything that would be bad enough to erase his lead in Texas would be bad enough to sink him in anything even remotely resembling swing states.

[supercat]

Another thing to consider: suppose a candidate had a 77.7% chance of winning each state other than his opponent's home state. For someone to capture all 49 such states would be a once-in-a-million-years event. And yet despite the fact that Reagan's odds would probably have no measured that high, he did precisely that.

[Political Junkie Too]

Let me put it to you this way: what would have to happen for Bush to lose, e.g. Texas? If such an event were to occur, would you still expect Bush to have a 59% chance of winning Ohio?

I understand what you're trying to say, however, that is not what is being modeled here.

If events were to cause Bush to lose Texas, I would not expect Ohio to remain at 59% probability, because the same event would cause voters in Ohio to reassess their positions, too. But, given the set of events as we see them today, the perception of those events distills into a 98.8% chance of winning Texas and a 59.1% chance of winning Ohio. The outcome in Ohio is not based on the outcome in Texas (no "So goes Maine, so goes the nation?"), which is my understanding of how dependent events works. Can you say something like, "If Massachusetts votes Kerry, then Connecticut will vote Kerry, too?"

-PJ