A little math for 2006 elections:

[Internut]

Using the odds for individual Senate races from The Political Oddsmaker, (http://www.campaignline.com/oddsmaker/index.cfm?navid=12) plugging in the odds into a little simulator I wrote and calculating the probabilities, this is what it comes up with:

                       Senate Seats
Democratic               |    Democratic 
Gain        Probability  |    Loss        Probability  
-------------------------|---------------------------
  9+            0.005%   |      9+           0.07%
  8+            0.03%    |      8+           0.28%
  7+            0.13%    |      7+           0.96%
  6+            0.47%    |      6+           2.72%
  5+            1.48%    |      5+           6.57%
  4+            3.92%    |      4+          13.63%
  3+            8.87%    |      3+          24.56%
  2+           17.36%    |      2+          38.95%
  1+           29.73%    |      1+          55%
                         |
  0            15.27%    |      0           15.27%

Disclamer: the odds for the individual races are taken from the above Web site. Since the Web site does not have the odds for the Delaware Senate race, it is excluded from the calculations. I am sure the odds for the individual races, and, thus, the above probabilities, will change as we get closer to the elections. The Political Oddsmaker claims "99.4% accuracy rate in 2004 elections" - however they come up with that.

[usregimechange]

1. I would be happy if we could add a few and should not 2008 have

a LOT more R seats up for grabs since the past two cycles had more D up?

[snowbear]

2. All those numbers make my head hurt...

What does it all mean.. that nothing will change? :crazy:

[RUMPLEMINTZ]

3. It looks like according

to that site the rethugs are likely to pick up at least one seat. And the odds are better that they will pick up 3 or 4 seats than it is for us. What a bunch of bullshit!

[snowbear]

4. Yeah.. I agree.. Internut your numbers are bullshit....

Rumplemintz said I could tell you so.. ;)

[carpetbagger]

5. About campaignline odds...

They aren't terribly accurate, skewing overly towards the center. For instance, on election day, Kerry got 9:1 odds on winning Massachusetts, and Bush got 4:1 odds on winning Texas. The odds on both were probably more like 100:1 or greater.

Notice how he gives odds of 2:1 for GOP control of the Senate after the 2006 elections.

I couldn't find any bookie sites with the 2006 odds, but I'd say we will have to run the board in order to take the Senate in 2006. It's happened before.

[Awsi Dooger]

11. In general his odds suck but I think Internut's table is fairly accurate

Just off the top of my head my subjective estimations would look very similar to Internut's numbers. We're not in great shape in 2006. That's why I hope we emphasize gov races to build a state by state foundation. I'm afraid the residue of 9/11 will impact us negatively in federal races for several cycles to come.

Carperbagger is correct, Ron Faucheux's odds skew absurdly low in lopsided races. In fact, my first post on DU in September 2002 was titled, "Ignore those laughable odds" and was a reply to a thread listing Faucheux's numbers for that midterm. I've wagered on politics since '96 and Faucheux's odds don't even approximate the actual betting line. Just because he hit's 99.4%, or whatever, doesn't mean he's a competent oddsmaker. Basically it means incumbents don't lose. If you make a candidate a 4/5 favorite and he wins by 25 points then you are incompetent, but Faucheux will credit himself for another success, calling the correct winner.

As carpetbagger emphasized, Faucheux's numbers on out-price races are hysterically innacurate. Let's just say work would be optional if I could wager into his numbers. He'll list someone at 1/2 or 1/4 when no offshore betting outfit will even book the race becuase it probably should be 1/1000 or greater.

On edit: the GOP at 1/2 to retain control of the senate is a perfect example of Faucheux's idiocy, erring too close to the middle as carpetbagger wrote. I know plenty of Las vegas wise guy gamblers who would literally wager every cent they own and beg, borrow, steal to play into that number.

[Internut]

12. Do you have a URL to a Web site that would

lay out odds for 2006 Senatorial elections that you would consider reasonable/accurate?

To address your misgivings: the skews, if they are confined to "sure bet" odds, would not drastically affect the total probabilities. It is the closer races' odds that mostly affect the totals.

[Awsi Dooger]

13. I don't know of any offshore outfit that's booking 2006 now

I bet politics with a few reliable offshore/internet outfits. None of them have anything available for 2006. They got burnt in 2000 and 2002 on statewide races so now they're a bit more conservative in putting numbers up early. I'll check for a website with political odds but that's rare, especially more than a year out.

You are right, the walkover races won't severely impact the net probabilities. That's why I think your numbers look accurate.

Regarding Alaska, I was just there last week and was amazed at the number of Tony Knowles bumperstickers I still saw. I don't know anything about the governor's race there for 2006, but it's such a GOP state I definitely would bet Murkowski at even money or thereabouts right now. He was near even in the polls with Fran Ulmer in 2002 then the race was a 20 point joke. Last year Knowles led every poll the entire year but the lovely annointed daughter Murkowski took an immediate 10,000 vote lead and it never got closer. I realize that's a difficult state to poll due to the geographic nature and only one person for every square mile. Certain areas skew extremely heavily Republican and I'm not sure they show up in representative number in state polling.

[Internut]

15. Give me a couple of sites to watch for the odds

when they start (if you know they usually do Senate races) so I can follow up on this - it will be interesting to see.

[Awsi Dooger]

19. These are the major ones I use for sports betting; politics when available

Ladbrokes is probably the best one for posting political odds a long time out. The others are the three best and most reliable sports betting online/offshore groups, but they don't emphasize political wagering until the brink of a major election. Note: some of these places might require an account before you can look at their odds. But generally you can sign in as a guest and take a peek.

www.ladbrokes.com
(note on ladbrokes. Last year on their pulldown "Quick Menu for Betting" they had plenty of political odds. I just checked and nothing is available)

www.betcris.com

www.pinnaclesports.com

www.thegreek.com

Below, these are websites that I don't deal with regularly but they offered wagering on politics last year and may expand it. The late money flooded in on Kerry due to early exit polls so betting websites generally profited last year:

www.tradesports.com

www.intrade.com

www.bodog.com

You might also search for the Iowa Futures Market. I don't know their url but they are known for booking presidential races, taking individual bets with people essentially setting the odds by wagering against one another. It might be presidential race only, not senate.

[snowbear]

18. Awsi....

We've had just as many Democratic Governors as Rethugs.

Frank Murkowski was a U.S. Senator for 25+ years, so he won the Governor's race over Fran Ulmer only because of sheer name recognition.

The state of Alaska has far more Independents than either R's or D's.

But......... residents will NOT vote for any candidate who is down on oil drilling. :shrug:

It's as easy as that. If you're pro-drilling and you're a Democrat, you can win easily.

You're right.. Lisa Murkowski was behind in 12 straight months worth of polls. Tony Knowles (former 2-term Governor and former 2-term mayor of Anchorage) was HUGELY POPULAR.

In comes Ted Stevens (senior R Republican in the U.S. Senate).. he told Alaskans that if Lisa lost and Tony won, he'd lose every important seat he sat on and that he'd never be able to bring home the pork again.

Ted Stevens is popular because of PORK, PORK, PORK. And when he threatens residents that the party is over if Tony wins (one month before the election) a lot of them bought into it lock, stock and barrel.

As far as the Governor goes.. fagettaboutit.. Frank Murkowski is completely TOAST in this state.

We have a Democratic mayor in Anchorage and he has the highest approval ratings the city has ever seen.. He ROCKS! :hug:

[Awsi Dooger]

20. Great! A popular Democrat from Anchorage sounds ideal

I know Anchorage has about half the population of the state so that's the perfect recipe to overcome the typical partisan slant, a Democrat from Anchorage. I'm emphasizing the gov races next year so ousting Murkowski is a priority. I hope they show the debates in your gov race on CSPAN. Lisa Murkowski damn near made me puke last fall with her forced cutesy remarks about the Red Sox in one debate I saw.

In one respect your state is like mine, Nevada. Las Vegas, like Anchorage, holds a huge percentage of the population here and I'm hoping our extremely popular Democratic governor Oscar Goodman will run for mayor next year, succeeding a two term Republican. Goodman can probably have the job if he wants it but he's a fun loving former mob lawyer who probably won't want the job because it requires relocation to sleepy Carson City in northern Nevada. Goodman won't be able to photograph Playboy Playmates au naturale, as he did a few weeks ago, in Carson City.

[snowbear]

6. Ok... now I know for sure that website is full of SHIT

Internut.. What on earth would even prompt you to post such a ridiculous bullcrap of a website?

It gives Frank Murkowski, our LOUSY Governor (has a 24% approval rating) good odds of winning again next year :rofl:

He couldn't win if he was running for dogcatcher!!

Internut, uh... :think: oh nevermind.. I'll hold that thought... I'll get my wrist slapped if my fingers type out what my mind is thinking right now.. :evilgrin:

[Internut]

7. It gives Frank Murkowski a 51% chance of

winning - which is VERY low for an incumbent governor.

If you can find another web site that makes odds for all the Senate elections, I can run those odds through to get the probabilities of gaining/losing seats.

[snowbear]

9. If I find another site.. trust me.. you'll be the first to know..

but...

a) I don't search for "oddsmakers" websites

b) If I stumbled on one as bogus as that, I sure as hell wouldn't share it here

c) As far as Frank Murkowsi, the wacko site says

.."Frank Murkowski (R) favored for re-election against the field"


Who maintains the numbers for that website, Free Republic? :silly:

[election_2004]

8. What Senate seats is it basing this on?

Is it even taking the Santorum/Casey race into account?

[Internut]

10. Here are the races/odds:

You can check it yourself at

http://campaignline.com/oddsmaker/index.cfm?navid=12

Currently held by Republicans    |   Currently held by Democrats 
---------------------------------|--------------------------------
Arizona                   75%    |   California             80%
Indiana                   88.8%  |   Connecticut            90%
Maine                     87.5%  |   Delaware               ???
Mississippi               83.3%  |   Florida                54.6% 
Missouri                  57.2%  |   Hawaii                 88.8%
Montana                   60%    |   Maryland               54.6%
Nevada                    66.6%  |   Massachusetts          88.8% 
Ohio                      60%    |   Michigan               54.6%
Pennsylvania              52.7%  |   Minnesota              50% 
Rhode Island              51.7%  |   Nebraska               52.6%
Tennessee                 51.7%  |   New Jersey             60%
Texas                     66.6%  |   New Mexico             60% 
Utah                      87.5%  |   New York               75%
Virginia                  75%    |   North Dakota           60%
Wyoming                   88.8%  |   Washington             52.6%               
                                 |   West Virginia          80%
                                 |   Wisconsin              70%

[dsc]

17. I do find some of these odds dubious

I find it pretty ridiculous that Minnesota is considered our least safe seat and that Washington State is tied with Nebraska. They also rank CT safer than MA and HI which I find bogus. I also find some of their rankings on the other side bizarre. Montana is certainly not safer for them than Tennessee and Texas is probably safer than some places like Arizona.

[Awsi Dooger]

21. You are exactly right

Faucheux has no clue on the onesided races. But he errs the same way on both sides so it sort of balances out.

Frankly, many of those races should be 1/20 or higher. Much higher. I guess at this point you have to be a bit conservative due to potential retirements, mortality, etc. But even on the brink of a race Faucheux still lists the numbers far too low.

My state of Nevada, for example, is obscenely low at 66.6%. That translates into 1/2. John Ensign has a MUCH greater chance than 2 in 3. He probably won't even have an A-list opponent since Oscar Goodman and Shelly Berkley are unlikely to challenge him. At that point the Nevada Democrats might have to fall back on something like a celebrity personal injury attorney, which sounds bizarre but that's exactly who opposed Ensign in 2000.

As dsc points out, Texas at 1/2 also appears incredibly low. I realize you might have an open seat but even so I would hate to bet a Democrat at only 2/1 odds in that state. Ron Kirk got dumped by double digits in an open race in 2002.

[jfern]

24. That seems to have problems

Apply some common sense. CA and NY are staying Dem, and UT and IN are staying GOP.

Also, Sanders and not Jeffords is running for Senate in VT.

[EmperorHasNoClothes]

14. Ummmmm.....

shouldn't that table add up to 100%, and not 220.005%?

[Internut]

16. No - because the probability of winning 1+ seats

includes in itself the probablity of winning 2+ seats, 3+ seats, etc.

The probabilities of D winning 1+ seats, R winning 1+ seats and the "draw" together should (and do) add up to 100%.

[drdtroit]

22. I think it can all be simplified:

                           :Total for States with Diebold Machines:

Democratic                                     Republican
      Gain                Probability               Gain                Probability

       9+                      0%                      9+                    100%    

[jfern]

23. Key problem with your calculation

You're assming that the probabilities are independent. They're clearly not. They give a one third chance (33.3%) of the Democrats taking back the Senate, which is clearly better than the 0.47% odds you have for the Democrats gaining 6 or more seats.

[Internut]

26. Please explain -

how exactly are the probabilities not independent? If you're talking about a "groundswell of support", that would be reflected in the individual races' independent odds.

[carpetbagger]

27. Here's a good example from across the pond.

The British House of Commons elections are basically national referendums. You can calculate the change of seats with a surprising degree of accuracy by knowing the national vote numbers. (I think the BBC prediction based on knowing just the national vote numbers in exit polls pegged the overall results within a few seats). Below is the concept illustrated by the BBC Swingometer. Obviously it's three different swingometers, since there are three parties (on election night, they used a virtual tilting table), but below is the labour vs. conservative swingometer...

http://news.bbc.co.uk/1/shared/vote2005/swingometer/html/labcon.stm

Obviously, the US Senate is less dependent on national shifts. However, national overall opinion clearly impacts on these races.

So let's say there are five even races. The random chance of one party winning them is 1/32. However, if you tilt the playing field by 3%, now the chances of all five of these seats falling out in one direction is much better. campaignline itself gives the democrats 2:1 odds on retaking control of the Senate in 2006.

[Internut]

29. I still don't understand why you say that the

probabilities are not independent. If candidate A has X chance of winning and candidate B in another state has Y chance of winning, then the probability of both of them winning is X*Y. One of them winning or losing does not affect the other's chances, since the elections are simultaneous.

[carpetbagger]

30. These are elections, not dice.

Let's assume for a moment that any given set of odds are completely accurate at a given time, and move on to other arguments.

You don't roll election A and roll election B. Elections are based on how many votes each person gets. They are, in effect, the sum of the probablilites of each individual voter going for candidate A, B, or none of the above.

Also, these probabilities change. They are not static. They might change because candidate C or D get in the race, lose the primary, come up with the killer one-liner, or get caught with people other than their spouses. But they also change because of factors common to multiple races.

Let's say that in 2006, economic growth is strong, jobs are being added around 300k a month, the Iraq government is finally showing signs that it's able to govern and police the place, and Howard Dean does a Dean Scream on Meet the Press. Clearly, the chances of republicans winning Senate races in Pennsylvania go up, but they go up in Missouri and Florida as well.

Now let's say that we hit another recession, go nowhere in Iraq, and Bill Frist talks about the need for a federal Christian Inquisition on Meet the Press. This helps democrats in Pennsylvania, Missouri, and Florida.

[Internut]

32. Of course - but that does not mean

that the probabilities in each race are not independent. That just means that there are factors that can raise the probabilities for one side in all races. The races still stay independent of each others.

What you are saying is that the odds are not fixed and can change. I think that is quite obvious.

[carpetbagger]

36. The outcomes of elections are not independent.

Therefore, they cannot be modelled as independent events. It's like saying that your risk of having heart disease and having a stroke are the same as the product of those two events. It's not. If you have heart disease, you're much more likely to have a stroke. Not because one causes the other (a stroke can result from the chain of events that started with a heart attack, but rarely), but because common factors cause both.

[Internut]

37. That's wrong.

If you have common factors that cause both, then the odds of you having a heart attack are higher and the odds of you having a stroke is higher - just like, if the political situation is favorable for Democrats, each candidate's odds go up. The probability of both a heart attack and a stroke happening (if one does not cause the other) is *still* the product of the two (higher) probabilities. This is elementary stuff, don't know why you're arguing about it.

The definition of an independent event is if the one event has no effect on the other. Since the races are done simultaneously, and no person (hopefully) votes in two or more races, then they are by definition independent events.

[carpetbagger]

38. No, it's not wrong. The frame of reference is what's key.

Like I said, elections aren't coinflips. If you have five voters, and three go blue, two red, blue wins. Every time, barring an inaccurate or incomplete count.

The odds, the probabilistic model, is the guy who's sitting there going "I think it's going to be blue, how often am I going to be right". His decision that there's a so-and-so chance of victory by a republican in election A and a so-and-so chance of victory in election B are not independent. If he's wrong in A, he's more likely to be wrong in B in the same way than the opposite way, because the assumptions are likely to be the same race to race. That's what election odds ande projections are all based on.

A simple example is Bush-Kerry. Let's assume there's no big Dieboldian stuff going on. Now, if you assume the national vote is 51-48 Kerry, you'll probably call Iowa and Ohio for Kerry. If you assume it's 51-48 Bush, you might call them the other way. Sure, personalities matter in the Senate, but the events that alter and shape national mood, which in mathematical terms are those things that make any individual voter or groups of voters more or less likely to vote in a certain way, they tilt the table. And when the table tilts unbeknown to the prognosticator, or after the prognostication, the predictions skew in the same direction.

I hoped that clarified what I'm saying.

[Internut]

39. I don't think so

"The odds, the probabilistic model, is the guy who's sitting there going "I think it's going to be blue, how often am I going to be right". His decision that there's a so-and-so chance of victory by a republican in election A and a so-and-so chance of victory in election B are not independent. If he's wrong in A, he's more likely to be wrong in B in the same way than the opposite way, because the assumptions are likely to be the same race to race. That's what election odds ande projections are all based on."

You're correct, if the oddsmaker is wrong because of US-wide political reasons, he would probably be wrong in all of the races. That's the same as if a coinflipper doesn't know that the coin is slightly weighted on one side - he will be wrong in most trials in the same direction. That still does not mean that the coin flips themselves are NOT independent events.

[Awsi Dooger]

40. This discussion reminds me of the election fraud debate

The conspiracy theorists scream, "No way everything can tilt toward the GOP!" while others insist if the opinion poll or exit poll models are flawed then it should err in the same direction and not zig-zag.

I fall in the latter group. On election day I returned from GOTV in early evening and looked at bizarre early exit poll numbers like Kerry +18 In New Hampshire, +10 in Pennsylvania and down only a couple points in North Carolina. Those were obviously garbage based on the partisan trends of the states. Then I checked out the Florida and Ohio exit poll numbers and felt worse than any time since the SCOTUS decision in December 2000. Kerry was "ahead" by only a point or two in those states. While everyone on DU was celebrating I turned off the TV in stunned silence. Since the Florida and Ohio exit polls figured to err in the same direction as the others and many points, I knew Kerry was finished.

[Internut]

42. That skew in exit polls does have a

fairly reasonable (IMO) explanation:

This is from The Guardian (I don't have the original URL, so will copy the relevant portions)
------------------
..
For years, British pollsters have had to deal with the Shy Tory factor, where conservatives are too cowardly to admit their ruthless conservatism to pollsters and so they pretend they are voting for nice New Labour or the Lib Dems. Some pollsters actually build in a Shy Tory factor into their research and add on a few points to the Tories as a matter of course.

...

By the time of the 1992 election, Shy Toryism was endemic: the polls pointed to a clear Neil Kinnock victory, but the Shy Tories crept into the polling booths in their millions, recording the greatest amount of Conservative votes ever.

...

America, though, has always been the home to heart-on-your-sleeve Republicanism, built along the lines of that 1950s Tory feeling - where a pride in patriotism and small government has been something to shout about.

That shouting has gone quiet, under the assault of Michael Moore, mainstream European opinion and the slivers of liberal America that run down the East and West coasts.

...

And, all the while, across the Atlantic, the Shy Republicans were silently flooding into the polling booths.

[Awsi Dooger]

43. It's like I told my dad a few months before the election

"I know we have the passion. We might not have the numbers."

He agreed instantly. Just because Democrats were the ones enthusiastically plastering on bumperstickers and tilting online polls didn't mean it would translate into a majority at the ballot box. I traveled the country that summer and quizzed literally hundreds of total strangers on the election. By consensus they did not have a warm or positive opinion of John Kerry. I reported that on DU many times after I returned.

But I got caught up in my intensity to oust Bush, plus my Excel election model reported a slight edge to Kerry a few days before the election. I thought he had a 55% chance and said so on DU. My mistake was not realizing 9/11 had made such an impact on party identification. I projected a 1.5 to 2 point Democratic edge but according to every report it shockingly broke even at 37-37. A Democrat can't win with a number like that. Restoring our edge in that category is the key to 2008 and beyond.

Here is a link to a Pew Research study in that regard. Since you are obviouslty interested in numbers, Internut, I think you might find it enlightening and helpful. I found the link a few days before last year's election but was too stubborn to fully accept the ominous implications: http://people-press.org/reports/display.php3?PageID=750

[Internut]

44. Very interesting stuff - bookmarked

I guess this goes a long way to explain the 50:50 party identification split in the exit polls. Let's hope the trend reverses itself.

[Awsi Dooger]

33. You are giving Ron Faucheux WAY TOO MUCH credit

That 33.3 chance he gives the Democrats to retake the senate is pure simpleton mush. I would like to wager every planetary cent into that number.

After studying Faucheux's numbers for several years, it was obvious he uses subjectivity and not a mathematical approach like Internut's. His overall number on a president being elected or one side controlling the senate has no logical relationship to his state-by-state numbers that determine the big picture.

Frankly, I would have assailed Faucheux's numbers and approach in many threads over the past 3 years, but whenever I have posted a numbers-oriented or political wagering thread it basically sails into the archives with zero to few replies, especially after jiacinto was banned.

[Internut]

41. delete

replied to wrong one.

[corbett]

25. That's Why Dean's 50-State Campaign Is So Important!

He is doing a great deal to change those numbers and I am convinced that his plan is sound. I am confident that Santorum will lose and Nelson of Florida will win.

http://www.tompaine.com/articles/20050601/the_key_to_impeachment.php

[geckosfeet]

28. These numbers are bogus,, at best.

How does your "little simulator" calculate this propagandist drivel,, oh I mean data.

I looked at the governors race for Massachusetts just for fun. The sitting Governor Romney (R) is given a 52% chance of winning when his opponent is not even known yet! This is like saying yeah he could win or he could lose. Thanks for the insight.

Campaignline claims to be 99.4% accurate and I sure would like to see how they come up with that figure. I suspect that they base that claim on calculations and results made with data available the DAY BEFORE the 2004 elections if not the day of the elections.

To put stock in these numbers at this time is delusional.

[ISUGRADIA]

34. Great Point

"99.4% accurate": they are continuously changing odds as the election season goes on.

I know they base their high % on their odds days before if not the day of the election. It's much easier to pick a winner on the eve of an election than months out. Numbers a yr and a half out mean nothing.

[Awsi Dooger]

35. I think the 99.4% comes from the final call before the election

So that number is misleading. But DUers were treated to the 99.99% probability of a Kerry victory by one prominent poster with his own simulator last year, so this place has no immunity from phony percentages.

Keep it up, Internut. I like analysis like yours. As I posted early in this thread, via pure subjective instinct I think the numbers your simulator came up with are very accurate. This thread reminds me I have yet to update my state-by-state presidential partisanship trends, numbers I quoted here regularly leading up to November 2004. That outcome depleted my political interest and energy for 6 months. Usually I update my chart within a month after the presidential election.

[RUMMYisFROSTED]

31. 95% "recidivism."

That's all one really needs to know.