MyDD :: The Great Predictive Value of Elections Futures Markets (Not)

The Great Predictive Value of Elections Futures Markets (Not)

by Chris Bowers, Sat Sep 30, 2006 at 12:08:19 AM EST

MyDD is a place where being geeky about numbers is encouraged. I blame myself for that. Over the past week, I have noticed a huge upswing in something even geekier than my usual standard: commenters and diarists insisting that election futures trading is an accurate predictor of election outcomes. I say hogwash, and I will never watch that stuff. Quite frankly, I don't think it is anything else than arch-capitalist, libertarian, market deification. I have also noted a tendency in these markets to be wrong about almost everything.

This was a good call, on November 4th, 2002:

On the eve on the Nov. 5 Congressional elections, traders in the Iowa Electronic Markets (IEM) continue to see a Democratic-controlled Senate and a Republican-controlled House of Representatives as the most likely outcome in Congress.

Nailed that one. I really enjoyed Democrats retaining the Senate in the 2002 elections. That was the highlight of those elections, if you ask me. A little more than a year later, they made another good call:

Howard Dean retains his lead in the Iowa Electronic Markets (IEM) going into the Iowa Caucuses, but his price has slipped about 20 cents, losing market share to other Democratic contenders.

In the IEM's Democratic Nomination Market, futures contracts for Dean were selling for about 50 cents Friday, down from a high of 76 cents on Dec. 9. In recent weeks, Wesley Clark has picked up market share; his contracts are trading at about 20 cents, up about nine cents on Dec. 9. John Kerry contracts are selling at 13 cents and "rest of field" contracts (covering John Edwards, Dennis Kucinich and others) are at 10 cents.

That was some awesome predictive value there. There days before the Iowa caucuses, the markets had Dean way out in front, and Clark ahead of Kerry. This was amusing ten days later:

Release: Jan. 26, 2004
Edwards Contract Created In Iowa Electronic Markets

Yup. It took them until the eve of the New Hampshire primary to even create a contract for Edwards. More good predictive value. All of this was almost as good as their predictive value in the 2000 election:

Traders in the Iowa Electronic Markets (IEM) still see a very close presidential election, but give George W. Bush a slight edge in the final days of the campaign.

The IEM, the University of Iowa's real-money market where traders buy and sell political futures, show Bush with a narrow lead at 52 cents with Al Gore close behind at 48 cents in the IEM vote share market. Vote share prices can be seen as predicting the percentage of the vote each candidate will receive.

In the "winner-take-all" market, IEM traders favor Bush at 63 cents over Gore at 37 cents. These prices predict each candidate's chances of winning the greatest share of the popular vote rather than the size of the victory.

I'm sorry--I don't remember Bush winning the popular vote. That was an awesome prediction by the electronic markets.

The markets did predict a narrow Bush victory in 2004, but even considering that they finally got something right, there is a common thread to be found in all of their predictions. Bush winning the popular vote in 2000, Democrats controlling the Senate in 2002, Dean winning the nomination before Iowa (with Clark in second and Edwards nowhere to be found), and Bush winning narrowly in 2004--in every case, this is what the conventional wisdom thought was going to happen. In the end, the markets didn't represent anything except conventional wisdom. Since conventional wisdom was wrong in 2000, in 2002, and in the 2004 Democratic primary, the markets were also wrong. Since conventional wisdom was right in the 2004 general election, the markets were right.

If you want conventional wisdom, read the electronic markets and/or David Broder. In fact, I would actually recommend just reading David Broder. Reading his columns will only cause you to become enraged--playing the markets could very well cause you to lose money. MyDD might not be much better--it is entirely possible that at any given moment, I am just reflecting the CW of the progressive netroots in my predictions, which currently say that Dems win a narrow majority in the House, and Republicans maintain a narrow majority in the Senate. That probably is the most commonly held position on the outcome of the elections in the progressive netroots and blogosphere right now. If, instead, you are looking for something that will tell you exactly what will happen--well, I don't think you will ever find it. As we found out yet again today in FL-16, politics is ultimately an unpredictable area of human endeavor. If anyone can predict exactly how 100,000,000 million people will decide to act between now and November 7th, then that person probably isn't human anyway. We all do the best we can to try and figure out what will happen, but now one knows for sure. In that environment, the only appropriate way to act is to work as hard as you can to cause your desired outcome to manifest itself in real life.

Tags: Election forecasts (all tags)

Permalink :: 105 Comments :: Post a Comment

Re: Elections Futures Markets (none / 0)

Reading this post was extremely satisfactory to me! I have noticed the same thing about these markets. All they do is reflect conventional wisdom, which often times proves to be wrong.

by eskimo on Sat Sep 30, 2006 at 12:19:34 AM EST

Re: Elections Futures Markets (none / 0)

I have been out of town for two days and just returned to find that my Limit Order on the Iowa Electronic Market for contracts for "Republicans Lose House 2006" had been filled. I sent the IEM my $100 and I have been buying those contracts as cheaply as possible, between 34 cents and 44 cents, each one payng a dollar if we take the House from the Republicans. I do not care a bit about the predictive value of this market. What it predicts is the wishful thinking of a few hundred self selected people willing to play with small amounts of money.

As it turns out, for the last few months I can infer that those playing this market were harboring wishful thoughts about the Republicans maintaining control of the house. That seems like a good bet to me: putting up 40 cents to get a dollar in return if, as most of us on this site probably agree, we believe we have a better than even chance of taking back the House.

Instead of this being a predictive market, I participate because it is a fun way to take some money away from others who believe in the Republicans.

by Arthurkc on Sat Sep 30, 2006 at 01:17:32 PM EST
[ Parent ]

Re: Elections Futures Markets (none / 0)

if you're going to do this, you may as well do it in the real markets with real money.

In other words, skip the Iowa futures and go straight to Chicago and buy options on real stocks right before the election. Figure once it becomes clear that the Democrats are taking over and will start investigating stuff, Halliburton puts should go through the roof (N.B. puts go up when the stock goes down) when the stockholders start shitting themselves at the prospect.

Then again, markets are known for being irrational, so YMMV.

by wrog on Sat Sep 30, 2006 at 04:30:24 PM EST
[ Parent ]

No one said markets are edicts (3.00 / 1)

I think my comment touched a nerve.

I have heard this critique before about markets but it seems to come from the perspective that those advocating markets are saying that whatever the market is indicating is the way it will happen, so let it be written, so let it be done. This is not true nor is it what is being implied.

Futures markets cannot guarantee a foretelling of the outcome. In any extremely close election, ala Senate 2002, Bush v Gore 2000 the market is going to lay odds one way or the other and any perceived slight percentage advantage such as a 1 point victory is going to end up giving decent odds for that person or party to win when you are just a day or so before the actual event.

The 2004 Democratic primary going into Iowa was a shocker to almost everyone. Markets are not magical. They can't foresee things that are entirely unknowable. They can only accumulate the knowledge and information (and opinions) of everyone in the market. And the point is that tends to get you closer than one person making a prediction based on whatever the person is predicting on. Markets are no where near 100% accurate, they are just more reflective of real odds. Odds is the key. Unless the odds are 100% then there is a chance it won't go the way the odds are leaning.

Pointing out a few close calls that didn't come down exactly on the odds side is like saying calculating odds in poker is pointless because you were a 2-1 favorite and lost two in a row. That happens, quite a lot, but when you know the real odds it gives you a higher chance of being right. 70% chance is still wrong 1 in every 3 times. I have operated in electronic political markets since 2001 and have managed to make 8 times my initial deposit in that time frame.

So I know how to tell when the odds are wrong and to make money off of those who don't get them right. If you would give me 90% odds right now Chris, I would bet every penny I have in the markets on that bet. I may turn out to be wrong. But I would be right 50% of the time and get a 9-1 payoff when I am. Thats an easy bet for anyone who is used to laying bets.

Continually ignoring these odds (not with respect to action but with respect to expectations) is what leads to big disappointments in the long run.

by lynx on Sat Sep 30, 2006 at 12:31:53 AM EST

Re: No one said markets are edicts (3.00 / 2)

They are not odds. They are CW of the people partiicipating.

by Chris Bowers on Sat Sep 30, 2006 at 01:50:36 AM EST
[ Parent ]

Re: No one said markets are edicts (3.00 / 1)

WOW!

I don't even know how to respond to this. How do you think odds get set? Are odds in vegas odds? They get set by the collective whole of who wants to bet how much on which outcome. Odds makers come up with what they "think" the odds should be and if they are "wrong" too many bets get placed on one side and then the adjust the odds to bring in bets on the other side to even out there exposure. It creates a market. Which sets the odds. Thats what markets do, they set odds.

by lynx on Sat Sep 30, 2006 at 02:01:59 AM EST
[ Parent ]

Very true and I'll provide an example (3.00 / 1)

One sportsbook tried to get cute in the early '90s. It was called Sport of Kings, located across from the Las Vegas Hilton on the current site of a nightclub called The Beach.

They brought in a sharp sportsbook manager and tried to pretend they could get away with fixed odds. In other words, they set the number on sports and horses and arrogantly took the action, confident in their number and not moving it.

Well, let's just say it was a paradise. Bettors left alone the numbers that were accurate and relentlessly pounced on the mistakes. Sport of Kings slowly cut their limits, then went out of business. It was open less than a year and has served as a laughing stock ever since then, for local gamblers on how the knowledge of many will destroy the knowledge of one or a few, no matter how astute.

by jagakid on Sat Sep 30, 2006 at 02:10:56 AM EST
[ Parent ]

Re: No one said markets are edicts (3.00 / 2)

Odds can only be usefully determined by having a defined range of outcomes. In this case, the "odds" are not being determined in that manner. Instead, they are determined by popular vote.

Popular perception of the possible range of outcomes does not alter the range of outcomes. If 56% of the country thinks that a coin will be heads, it does not make the odds of the coin flip coming up heads 56%. The odds are still 50%.

You are working from a gambler's perspective. From my perspective, odds in this case should only be determined from a defined set of quantiative data--polls, cash, economic condition, partisan voting indexes, Democratic performances, etc, and the way previous election results varied in relation to the pre-eleciton data in these categories.

The markets are nothing more than the way people partiicapting in the markets perceive the eleciton. As with the coin flip, the range of possible outcomes in the eleciton is not altered by the way people perceive the outcome.

Those are not odds. Those are gambler's lines. They are the way people betting on the eleciton perceive the election. Theya re the CW of people partiiciapting int he market.

by Chris Bowers on Sat Sep 30, 2006 at 02:25:47 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

popular perception would not be 56% on a coin flip so thats a rediculous straw man.

You think you can come up with your own way of setting them or that someone can and can get them perfect or accurately reflect their real value, such as in a coin flip. You are wrong. Nobody can get them perfect. Not even the market. But the market gets them closest.

You just don't understand odds. Everyone in this thread has tried to explain that and why your 90% prediction is so far off from the true odds as to be rediculous. But you are un-moved. So be it.

by lynx on Sat Sep 30, 2006 at 02:35:57 AM EST
[ Parent ]

Re: No one said markets are edicts (3.00 / 1)

You are wrong. People do not choose evenly between heads and tails, just as they do not choose evenly between a, b, c or d on a multiple choice test. Just look at our own results on MyDD form a poll with five identical chocies. Identical results received compeletly different vote totals.

The key is that it was a blind test. People did not know before ahdn which votes were getting more. You are right that a market system of heads and tails would hold steady at 50%, but that is only because peopel can see how other people are acting. If the bets were placed without knowing how other people were acting, I am quite confident that heads would trounce tails.

My point is that not only are these markets perception on how the elections will end, but also perceptions on which bets are relatively better deals to the perceived outcome at any given moment. For exampe, you might place a bet on Demcorats at 45%, but not at 40%, because you are weighing your own odds relative to the collective odds. At that point, the system is already corrupted, ebcause instead of trying to figure out the result of the eleciton as clsoely as possible, people are trying to play the system instead.

Which again, shows that this is all about perception rather than about trying to accurately appraise the eleciton situation as best as possible. It will also gravitate toward a general group-think, becaus everyone playing is basing their results not only on the eleciton results, but also on what other people are thinking.

This is not a useful predictive system. My personal bet is that if the election were held tomorrow, in nine out of ten universes, Demcorats would win the House. However, if I were playing the market, I would certianyl sell my shares in a Democratic win far before they reached a value of 90%. That is because at that point my goal would be to make money, rather than to predict the outcome. I wold start working on the groupthink model, rather than on the data. My buying and selling would be based ont he thoughts of others, rather than my own analysis.

by Chris Bowers on Sat Sep 30, 2006 at 02:48:44 AM EST
[ Parent ]

Re: No one said markets are edicts (3.00 / 1)

But in a market people can see how others are acting. So I don't understand the point about betting without seeing how others are acting because markets don't do that.

Even if these are based on perceptions you don't seem to be following the accuracy that gets created out of the diversity of perceptions and opinions that enter the market.

So by your admission you clearly believe that 9 times out of 10 the democrats would win the house. Every market disagrees and I would be happy to take you up on your personal bet as I said before if there were a way to do so.

If you are so certain of this, have you deposited any money with any of these markets that have the odds so wrong? It should be a very easy way to make money.

As a side note back in 2002 there were a number of posts here by Jerome about how the chances that the Democrats would not take back both houses were very close to zero and yet those odds were trading at about 5% in Iowa Electronic Markets. He suggested he and others add money to the market and make a killing. I traded in that market at the time. After his post, in the next couple days the odds of Dems taking back both houses shot up from 5% to about 15-20% based on that artificial buying. I sold into that bump like crazy until I had no more money left to sell. In a few weeks the markets returned to their rightful place and I bought them back and made a bunch of money. The markets know what they are doing.

If the odds truly were 90% then you would be wrong to sell your shares at 80% or 85% or even 89%. Once again, betraying a lack of understanding of how odds work. And if you did sell them someone would gladly snatch them up and the price would still tend towards its rightful place of 90 even after you sold (unless of course you had so much supply as to overwhelm the market but that is a different scenario which is unlikly in these types of markets).

Clearly we just don't agree on this. Research is on the side of the markets on this. We will just have to disagree.

I do hope I find you on the other side of my trades though. :)

by lynx on Sat Sep 30, 2006 at 03:08:59 AM EST
[ Parent ]

Re: No one said markets are edicts (2.50 / 2)

"I would be happy to take you up on your personal bet as I said before if there were a way to do so."

That is the whole point of our disagreement. Your goal is to make money. Of course markets are better for that. My goal is follow elections. Polls are better for that.

I am not here to place bets on elections and win money from those bets. If that is your purpose, so be it. I'm in this to follow campaigns--and win them. Any campaign that would follow markets instead of polls would be insane.

If you want to make money on bets on elections, then fine, follow the markets. But that isn't my purpose here. I want to find out what is happening in the election, and use that information to help progressives as much as possible. This isn't a game to me. I am not here to make money. I am here to change the contry. And markets don't provide me with the information I need to do that. Any campaign that is trying to enact change would agree with me. Anyone out to just make money would probably agree with you.

by Chris Bowers on Sat Sep 30, 2006 at 03:18:52 AM EST
[ Parent ]

Chris, you just don't get it (none / 0)

If you aren't man enough to bet real money on your predications, then you must be wrong. Period.

<snark>

I completely agree with you. People who play games, especially games involving real risk (gambling) are constantly looking for a way to justify their obsession with the game. They call it a "market" so it seems legit. No one calls betting on a poker game a "market" yet within the context of the game, they're doing the same thing - placing bets on what they think is going to happen and what cards they have and or have seen.

If I bought "shares" of the stock "Democrats win the house in 06" at 50cents (or percent or whatever) and the price rose to 80 cents and I thought maybe it would go to 90 cents but maybe it might drop...I sell it at 80 and take my "Profit" of 30 cents. If I bought shares at 80 cents cause I thought the Dem chances were 90 and I was right I would have made 10 cents. I would take a 30 cent gain for sure rather than risk it for and extra 10 cents. If that's normal psychology, then the markets would show the Deems losing steam at 80 cents and the other bettors might react to the down tick or slowdown by re-assessing the likelihood of 90cents being reached and selling off their Dem stock to preserve whatever existing profit they have.

To suggest that that kind of market action is predictive how people will actually vote is beyond silly. Its desperate gambling addicts refusing to admit their version of poker or sports betting is gambling on elections.

by David in Burbank on Sat Sep 30, 2006 at 01:16:44 PM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

Some people are actually interested in elections as a means to further the progressive movement, rather than to make money.

by adamterando on Sat Sep 30, 2006 at 09:41:46 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

"You just don't understand odds."

I think I do understand odds. I just think we disagre on how they should be calculated for elecitons. I think they should be calculated form polls. You think they should be calculated from markets. I'll take my method over yours any day of the week for determining the outcome of elections. If I wanted to make money, I'd follow your path. But I am here to follow elections, not to make money.

by Chris Bowers on Sat Sep 30, 2006 at 03:13:28 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

Fair enough. If your results lined up with other poll based prediction sites I would not find them so out of line with the markets.

Any thoughts on that?

by lynx on Sat Sep 30, 2006 at 03:17:16 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

Well, ONE interesting point of attack on predictive markets is, say, the gross discrepencies between tradesports and iowa em in the house gop win.

by lokiloki on Sat Sep 30, 2006 at 03:19:12 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

Actually, I am under several other sites like that. Cook, Sabato and Rothenberg all favor Dems more than I do. CQ politics and Real Clear Politics are in the other direction.

I am not the most optimistic predictor for Democrats out there. Not by a longshot. I still have Republicans controlling the Senate, for example. Several predictors give Dems better than a 50% chance.

by Chris Bowers on Sat Sep 30, 2006 at 03:21:48 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

Thats not what I have heard. I have heard Cook is 14-17, Rothenberg I heard is 15-20. I don't know the real values cause I don't see those sites. I am just looking at poll agrigator sites such as electoral-vote.com. And its not so much the numbers although 18-24 seems to be on the high end that I have seen (and thats not to say it can't be 30, cause it can). But the 90% just doesn't compute.

I think you may not understand how to translate the ranges into odds. I think the odds are pulled right out of your gut.

As an example, a few weeks back you had the range at 15-23 and you put the odds at 95%. I specifically remember that one because it made no sense to me. Now you have an even better range and lower odds at 90%.

The 90% odds number just doesn't compute.

by lynx on Sat Sep 30, 2006 at 03:27:42 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

If you are just upset if that number, then I think you have wasted a lot of bytes. I just throw that on at the end of the forecast. I don't really care about it. Probability doesn't matter when you are trying to win an eleciton, because the eleciton only happens once, and the rewards and penalities are the same no matter what the odds were going in. If Demcorats have a 99% chance to win control, and they win control, the prize is the same than if they had 1%. The odds of victory don't matter much because it only happens once, and there is no larger payoff.

The blood, sweat and tears go into placing each district into tiers. For example, the only campaigns that go in "lean Dem" show recent poll results with Democrats ahead outside the MoE, and cash competitive. In other words, those are the campaigns where right now I see the odds of Democratic victory at or over 95%.

Each tier is based on a certian level of victory probability for Democrats. It starts at 95%+,a nd goes down from there. then, I calculate the projected numebr of seats won in each tier, and that is that.

Adn Charlie Cook thinks it is going to be a lot more than 14-17. Go ahead and ask him personally. Also, when Rothenberg had it 15-20, I had it lower (12-19, I beleive). That was two or three weeks ago. I update more regularly than he does.

by Chris Bowers on Sat Sep 30, 2006 at 03:42:09 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

Fair enough,

I suppose I have a betters mentality where all that matters is the odds and it matters a great deal even though its only a one time event.

So the odds do come from you gut as I suspected so with your prediction perhaps a statistical model would put the odds at betwen 60-70% or something like that. Thats still on the high end of what the markets are saying but no wheres near the difference of the 90%.

I know you put a lot of effort into your race by race breakdown and I appreciate that.

I do think your percentage odds might get people a little overly optimistic. I mean 95% odds a few weeks back, I was thinking why even have the election. :)

by lynx on Sat Sep 30, 2006 at 03:50:20 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

And there is one more thought on that--The Incumbent Rule. My samples for congressional elctions are a lot better htan the ones I have for PResidential elecitons. I think where the polls stand rihgt now, in individual races, which probably show something like a 55-60% chance of a Demcoratic House win, will change dramatically when the undecideds break ovr the next few weeks. If Dems are aleady ahead, and undecideds should break at least 15% in their favor between now and election day (and possibly 50% in their favor), then the odds of a Democratic House victory increase dramaticly, and fast.

by Chris Bowers on Sat Sep 30, 2006 at 03:27:03 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

Well you may be right. If your 90% is correct, your crystal ball is a lot better than everyone elses.

by lynx on Sat Sep 30, 2006 at 03:29:19 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

I was going to make a point along those lines. Electoral-vote.com is based strictly on what the polls say RIGHT NOW. And we all know that polls are only accurate as snapshots of public opinion at a moment in time. For many of us who believe this is a wave election, and that Democratic momentum has not yet hit its apex, it's quite reasonable to predict that we'll win races where polls 5 weeks out show us barely behind.

That site still has FL-16 as a solidly Republican seat, for example. It will take at least a few days for a poll to come out showing otherwise. That doesn't make us wrong when we consider FL-16 to be an almost certain Democratic pickup.

by Sandwich Repairman on Sat Sep 30, 2006 at 08:34:35 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

It's like what Kevin Neilan (sp?) said on Saturday night live,

"A new poll finds that if the election were held today, 80% of voters would be really surprised."

by adamterando on Sat Sep 30, 2006 at 09:45:35 AM EST
[ Parent ]

dude, I'm a scientist and mathematician (3.00 / 1)

bachelors in physics and mathematics. Masters in mathematics and a Ph.D. in neurobiology.

I work with statistics you didn't even dream existed. And I can assure you that the probability of an occurrence has nothing do with what markets assess it to be.

Nothing at all.

by Teaser on Sat Sep 30, 2006 at 07:55:39 AM EST
[ Parent ]

Re: dude, I'm a scientist and mathematician (none / 0)

Exactly!

Now, if we were having ELECTIONS for congress every day or evey week, then you could probably forecast very well what the outcome would be on election day. Unless it's just white noise, but there would be a lot of dependency in the data, so I doubt that. So you could actually use the previous results from the elections every two years and come up with some model to estimate odds.

Also, I thought The Market (blessed art thou, O Great Market) wasn't about gambling?

by adamterando on Sat Sep 30, 2006 at 09:50:26 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

exactly. The odds reflect the preconceived notions of the betters. A person like you, should be able to do quite well betting.

Wouldn't it be nice to have a mathematical model to predict elections, like what the hurricane forecasters have? I couldn't find one. They're probably unethical, like counting cards.

Dare to be free.

by misscee on Sat Sep 30, 2006 at 08:24:18 AM EST
[ Parent ]

Re: No one said markets are edicts (none / 0)

No, it'd be rather easy to come up with one. You'd just have to base it on prior elections, and then figure out how to add the noise.

It'd probably be way off though. Then again, it'd probably be about as accurate as hurricane models, because as you get closer to election day (assuming you include polls in your model) your cone of uncertainty might shrink.

by adamterando on Sat Sep 30, 2006 at 09:53:13 AM EST
[ Parent ]

Re: No one said markets are edicts (2.00 / 1)

Wagering doesn't have a feedback loop to prices in a future markets.

Your argument makes about as much sense as saying the line in Vegas is an accurate predictor of what the scores in football will be. But even in football wagering there is more of a feedback loop to future wagers than there is in election markets.

by RickD on Sat Sep 30, 2006 at 06:09:04 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (2.00 / 1)

Just because the market says something is the most likely, does not at all mean the market thinks it is guaranteed to happen. The price of the contracts reflects roughly the percent chance of it happening. For instance (to make up numbers, I've not checked), suppose in 2004 the final market numbers were Kerry 52 Bush 48. I would call this within a few percent of the correct market price for that contract. It is of absolutely no consequence that such prices would "predict" a Kerry win, because they do not make a confident prediction of such a win.

I've done lots of statistical measurements in physics experiments. Interpreting the prices as the probabilities, one can look at the cases you cite and imagine analagous statistical probabilities and outcomes in an experiment. The cases you cite would raise absolutely no eyebrows about the validity of the probability assignments. They wouldn't necessarily confirm them, either. I'm just saying that in the world of rare, one time statistical events, this collection of outcomes is not very surprising even if the underlying probabilities were correctly estimated.

Or, put more succinctly: if there's a problem with the markets, you haven't demonstrated it.

In fairness, I should note that most of what's been written in support of these markets is similarly anecdotal, and similarly not soundly based in statistical analysis of aggregate events, but instead based on "The markets had event X priced at 51 cents and it happened!"

by Professor Foland on Sat Sep 30, 2006 at 12:42:55 AM EST

Re: The Great Predictive Value of Elections Future (none / 0)

"In fairness, I should note that most of what's been written in support of these markets is similarly anecdotal, and similarly not soundly based in statistical analysis"

spit take

That's about all we need to know, isn't it?

by RickD on Sat Sep 30, 2006 at 06:10:47 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Right, so markets are decent generators of statistical probabilities with which one could make money.

Too bad, we're not here to gamble on elections, we're here to further a progressive movment. And that's why the Iowa Markets are lame. They help nobody but the gamblers.

by adamterando on Sat Sep 30, 2006 at 09:56:43 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (3.00 / 1)

And since we're on the topic, I'll make my own comment about the IEM.

For both the Senate and the House, of late it has been possible to get a price of roughly 10 cents on contracts for the Republicans picking up net seats.

In the House, many careful people believe it possible that the GOP will be shut out of any seat gains at all. There is literally 0% chance that if the election were held tomorrow that they would net seats in the House. There might be a 1% chance in the Senate (if you actually believe we're in trouble in New Jersey.)

One of two things is true. Either the people who are buying these contracts are "noise traders" (which honestly seems very unlikely to me), or they are banking on the fact that the election is not tomorrow. Perhaps they think there's a 10% chance of a game-changing October Surprise.

Just something to chew on.

by Professor Foland on Sat Sep 30, 2006 at 12:49:17 AM EST

Re: The Great Predictive Value of Elections Future (none / 0)

Great point. And I am exploiting that. I am currently betting everything I can against that happening specifically in the Senate because the odds of a gain in the senate has been running higher. Those odds are clearly wrong. The odds are as you say, somewhere around 1%. And so I will gladly exploit those odds.

I don't know why the odds are that wrong on that one. The problem is with such a cheap contract at 10 cents someone can actually hold up the price and its tough to push them out because it would take so many sellers into them to do it. At the odds for the other contracts no one is able to corner the market enough to pull that off and so those prices fall in line quickly when they get out of line. The Gain contracts will fall in line before election as well but I suspect they will never trade as low as they should simply because once a contract gets extremely cheap, someone is always buying it or proping it up.

by lynx on Sat Sep 30, 2006 at 01:05:59 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

So I can't help myself--being not a finance person and wondering this. Is there a way to think of these contracts as options and calculate an implied volatility?

(This is now highly OT bordering on technical so sorry Chris :) )

by Professor Foland on Sat Sep 30, 2006 at 01:16:45 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

I kind of do think of them as options with an expiration date but I don't know much about calculating implied volatility from it. I only know how to calculate the risk and see when the odds are wrong.

by lynx on Sat Sep 30, 2006 at 01:18:39 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Sure there is... you can even calculate the rate of the rate of volatilty change (whats that, the theta)? ahh memories of Options Pricing and Volatility (simply the best book on options pricing available:

http://www.amazon.com/Option-Volatility- Pricing-Strategies-Techniques/dp/1557384 86X/sr=8-1/qid=1159596813/ref=pd_bbs_1/1 02-2659384-7952905?ie=UTF8&s=books

by lokiloki on Sat Sep 30, 2006 at 02:14:04 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Well, it's basically Brownian Motion ain't it?

by adamterando on Sat Sep 30, 2006 at 09:59:00 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

" Perhaps they think there's a 10% chance of a game-changing October Surprise."

That is fine if they think that. It is their right ot do so. But just because they think it doesn't make them right.

The physics experiements you deal with are testing an actual world. These markets are testing people's opinion.

I just don't see testing people's opinion as relevant tot he determining the outcome of an election. I would if the people participating represented an accurate dmeographic cross-sample of the electorate. However, we have no information that they do, and I seriously doubt that they do anyway.

I think that polls are more acccurate than markets when it comes to determining the outcome of an election. Right now, the polls show a very high Democratic probability of victory--much higher than the markets show.

by Chris Bowers on Sat Sep 30, 2006 at 03:01:26 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

A Random Walk Down Tradesports?

by lokiloki on Sat Sep 30, 2006 at 03:04:18 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

the markets don't have to be a cross section of the electorate because they are not betting as they would vote.

electoral-vote.com is based solely on polls and it has the Repubs winning a 220-215 seat majority today. I has hoped around a few seats in the last few days but it has been from even to a slight Repub edge. Exactly what the markets have.

Honestly, you analysis is the only one I have seen that shows a very high probability of Democrats winning the majority. As a better in the markets I follow many sites that make analysis based on the polls. Some show it even to a slight edge either way but no one shows a high edge to except for your analysis.

by lynx on Sat Sep 30, 2006 at 03:14:28 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

"I just don't see testing people's opinion as relevant tot he determining the outcome of an election. "

Ahh, the joys of quoting someone out of context. In all seriousnes Chris, given the numerous issues with polls I don't understand the antipathy to the market. Yes it is self selecting. Yes it just represents the opinions of the (tens of?) thousands of people participating. But don't the polls also just measure opinions? Now, I understand that one is based on hearsay 'I think that they will think' rather than direct opinion. But in my experience, groups of people are actually pretty perceptive. Can't you figure out the way most of the folks you meet would vote?

The fact that the markets exist as a way to make money is irrelevant to me. If anything the market is the aggregation of a whole lot of people like you literally putting their livelihood on the line to understand where the races are really heading. Now should the markets drive policy and campaigns? No, but I don't think polling should either except at the most basic level.

Finally, you may be right that the polls are correct and the markets wrong. In that case it is an opportunity to make money for folks. These are young beasts. As more people pay attention the lag between new information and market correction will lessen. Eventually it is reasonable to believe that the market will generally have an accurate snapshot. But it will still be based on folks like you doing the research and analysis.

The market is a mirror.

by PghArch on Sat Sep 30, 2006 at 01:06:52 PM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Electoral-vote.com has this comment up today about polls. Just another reason why polls are so subjective and markets can allow thousands of people to take into account all kinds of factors. Polls definately have limitations.

--------------------------

The PublicMind poll at Fairleigh Dickinson University in New Jersey ran a wonderful controlled experiment in July that illustrates the problem. They asked 1/3 of the respondents the following (simplifed) questions in this order:

1. Do you believe the country is going in the right direction or the wrong direction?

Would you rate the job the President is doing as good, fair, or poor?
Do you approve of the way President Bush is handling his job?
Is the military effort in Iraq going very well, fairly well, not too well, or not well at all?
Did the U.S. do the right thing going to war in Iraq or was it a mistake?

Then they asked them whether they were going to vote for the Democrat, Bob Menendez, or the Republican, Tom Kean, Jr., for senator.

A second 1/3 was asked questions 1-3 before the question on the Senate race and the final 1/3 was asked about the Senate first, then the five questions above.

The results are significant. In groups 2 and 3, Menendez and Kean were virtually tied. But in group 1, Menendez held an 8% lead, In other words, just bringing up the subject of Iraq, causes a NJ state senator (Kean) who had nothing to do with the Iraq policy, to tank. While this result suggests a strategy for the Democrats (talk about Iraq all the time) and one for the Republicans (hide the fact that you are a Republican), it clearly illustrates our point that methodological issues often dominate in polling. To really understand what a poll means, you have to look at the exact questions asked, the order they were asked in, and fair number of other issues.

by lynx on Mon Oct 02, 2006 at 10:00:55 AM EST
[ Parent ]

Have you looked at NewsFutures.com? (none / 0)

They only tinker with play money, and have a range of contracts well beyond politics.

Still, if I vaguely recall right, they were pretty solid on Bush up until the first exit polls.

Just thought I'd offer another futures market to look at.

Also, there's TradeSport, which plays with real money. Can't recall their numbers on Bush-Kerry.

by jcjcjc on Sat Sep 30, 2006 at 01:10:39 AM EST

Tradesports... (none / 0)

...is the good one.

by MNPundit on Sat Sep 30, 2006 at 02:08:03 AM EST
[ Parent ]

I'm cheap (3.00 / 1)

What can I say? I'm not spending any money on any futures market. I'm a holding stocks kinda guy.

Also, I've seen a few studies that having some skin in the game tends to overweight a player's view of a market.

For example, could you imagine if we used a politicos-only structure for a news futures market?

Strangely, the wish to profit also overweights a market. Look no further than tech stocks, oil and housing over the last few years to see how hard it is for players to get out at bad points when they have skin in the game. They cling to the hope for more profits and often get burned when the bubble explodes.

So, I think you have to evaluate what's keeping money in a contract before judging the contract too far.

I can't imagine a scenario where you'd get an honest read from a political futures contract.

by jcjcjc on Sat Sep 30, 2006 at 03:02:24 AM EST
[ Parent ]

Re: I'm cheap (none / 0)

I agree. I think that if you want to test the probability of the outcome of an election, it is better to conduct a survey of of how people plan on voting, as long as that survey is representative of the people who will actually vote. That will offer you the proper range of probability on the outcome of an election.

Pollsters predicted that Kerry would win Iowa, markets did not. Polslters predicted that Republicans would take the Senate in 2002, markets did not. There is sometimes wisdom in crowds, but often a survey knows a crowd better than the crowd does.

by Chris Bowers on Sat Sep 30, 2006 at 03:07:58 AM EST
[ Parent ]

Re: I'm cheap (none / 0)

I'd offer that it is possible for contracts to work, but the group of people interested in this sort of contract is simply too self-selecting to do the job.

Who is going to buy into a political futures contrat except a person with deep interest?

Remember, the old predicition in numbers theory stems from a lack of investment in the outcome.

The classic example is that a group of people can give a fair estimate of the number jelly beans in a jar.

Now, if you change that to hundred dollar bills and offer to give the jar to the person who guesses the closest, I'd be surprised if the estimates weren't skewed by a degree of wish fulfillment.

Similar effects have to come into play when players buy into these markets.

Likewise, I'd also be surprised if in such a volatile contract as a political future (ask Foley) if there isn't a tendancy to just go with the flow for lack of a better prevailing wisdom to guide buyers.

Lastly, I'd offer that purely adversarial contracts with endings may be flawed.

The stock market works because stocks are not purely adversarial, and there is always a tomorrow.

Perhaps the political futures contract should be left open, with people perpetually buying in. So, when one election ends, the contract merely rolls over to the next election.

There probably is a feaible way to make it work, but at this stage news futures are in somewhere around version 0.5.

by jcjcjc on Sat Sep 30, 2006 at 11:19:57 PM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Nice annecdotes Chris, but solid, academic research has shown that these markets are pretty damn good.

http://www.biz.uiowa.edu/faculty/jberg/p apers/ResultsIEM_2003Mar.pdf

http://www.atypon-link.com/AEAP/doi/abs/ 10.1257/0895330041371321

Might not be able to access the second paper unless you have access to a university network.

It's likely that the Iowa Caucases produce highly deviant results because they're simply harder to predict, as are all presidential nominating contests.

by dantheman on Sat Sep 30, 2006 at 01:25:11 AM EST

Re: The Great Predictive Value of Elections Future (none / 0)

Good find on that research paper (the first link), I highly recommend the hard data graphs on pages 11 and 12. These include presidential elections 88-00, and the result for these elections is an avg. absolute error of 1.37%.

The more knowledgeable players that participate in the market, the more valuable the result. This is why, in all the data above, the less watched the event, the higher absolute error. Very few financial types would bother getting good data about democratic caucuses. I think there's about a 1-2% conservative bias in these markets because most of the moneyed traders are conservative and biased (I briefly worked in the NYC financial industry), so it shouldn't be surprising that the dem caucus got little attention. The Intrade site lists their contract as "GOP Control" - what framing!

I certainly agree that the best way to know what's going on is to conduct a careful study of the data and make an estimate based on the data, as Chris has done -- the "market" is no panacea, as he points out.

There are biased idiots participating in the market, but it's highly likely that there are also people trading with insider information, with the latter likely to put more money on the line than the former. I have read Chris's forecast, and it all looks logical to me, but it's all based on public information. Insider information spreads like wildfire, and a method that enabled peopel to use this information might actually come up with better results than your dilligent forecast.

end the occupation of Iraq

by aip on Sat Sep 30, 2006 at 04:41:56 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Read Wisdom of the Crowds.

by lokiloki on Sat Sep 30, 2006 at 01:35:09 AM EST

Re: The Great Predictive Value of Elections Future (3.00 / 1)

Since markets essentially reflect the collective view of the people participating in them, I think Chris is right that they are just some amalgamation of CW and little more.

You're just aggregating the opinions of many traders. Without knowing more about the traders themselves, as in how poll savy or politically knowledgable they are, one must assume they fall on a spectrum from prescient to foolhardy. We can be generous and assume they are more heavily weighted towards people who are more knowledgable, but I'm not actually sure that's true. Greed will attract morons, and morons don't necessarily think they're morons.

The average of that, will be less accurate than an expert poll watcher using sound statistical models to evaluate polling methodology and soundness.

Lest we forget, polling companies have a very strong interest in getting things right. Even the partisan ones can't be caught dead wrong too often lest their reputations go in the toilet.

The value then, of watching these markets is that it is an averaging of opinions versus choosing any one opinion, say like MyDD or Cook. An individual is more likely to make a larger error than an aggregate.

But the aggregate is still more likely to be wrong overall in any one instance.

I'd say the market will never be as wrong as a bad analyst, but good analysts will be right more often than them.

by scientician on Sat Sep 30, 2006 at 01:43:19 AM EST

Re: The Great Predictive Value of Elections Future (none / 0)

Sorry, but you don't seem to understand much about prediction.

Markets that contain foolhardy and prescient people are what makes a markets predictive power so potent. It is the DIVERSITY of opinions that makes such markets so accurate. Experts rarely are right (in fact, as the book Expert Political Judgment points out, they are MORE OFTEN wrong than amateurs). But a market stacked with rightwingers, leftwingers, idiots, geniuses, etc etc, will amalgamate all hidden knowledge (NOT just CW).

Read the intro of Wisdom of the Crowds on Amazon.com. It highlights Francis Galton's surprise when he found that a collection of townspeople could guess the weight of a pig more accurately, collectively, than 99% of individual participants.

by lokiloki on Sat Sep 30, 2006 at 01:55:34 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Read Galton here:

http://www.amazon.com/gp/reader/03857217 06/ref=sib_dp_pt/102-2659384-7952905#rea der-link

(the "excerpt")

by lokiloki on Sat Sep 30, 2006 at 01:57:17 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (3.00 / 1)

Any system of definitions that insists that experts are wrong more often than amateurs really needs to rethink its basic logic. I would think that somebody wrong more often than an amateur would hardly qualify for the label "expert" as a matter of tautology. Certainly somebody who is "rarely right" is hardly an expert in any meaningful sense.

Wait a minute - are you an "expert" at this particular question? Or are you an amateur? `

by RickD on Sat Sep 30, 2006 at 06:24:35 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

That may sound like it would be right but the research shows that markets do a better job than so called experts.

See the links in dantheman's post above.

Sitting around as "experts" and deciding what we think is the way things work is exactly why markets beat experts. Markets in the whole average out the noise that is introduced when experts try to "think" about how things should work.

The problem is that it doesn't seem like there should be wisdom in markets. They are just a collective averaging of opinions informed and uninformed. So we can say we think that doesn't make sense and dismiss it. Research and evidence shows otherwise. Certainly markets can get it wrong, they just do so less often than experts do.

Rely on what you think should be true if you like, I have money on the line so I rely on the research that shows that markets do a fairly accurate job.

by lynx on Sat Sep 30, 2006 at 01:56:27 AM EST
[ Parent ]

Gambling was easier prior to the markets (3.00 / 1)

Very astute comments in this thread. What exactly is demonstrated in Chris's long rant, that a 52% likelihood or a 70% likelihood can produce the opposite result? OH MY GOD!! I need to open the window and absorb that. Next thing you know he'll be telling me a 1 point favorite and a 7 point favorite in football have been known to lose, even in the biggest games. Seems like my alma mater USC forgot to win as a 7 point choice, which is roughly 70%, in the Rose Bowl last year.

I have been wagering on politics since '96, the year I entered a 16-man wagering pool. In those days it was so simple it was a farce. There was no such thing as the markets or Tradesports and all I did was basic analysis, an early version of the Partisan Index and some sampling of how undecideds figured to break. None of my competitors had any comprehension of true odds and I walloped them in '96 and '98.

It got a helluva lot more difficult in 2000. Simply due to the explosion of the internet and the more accurate odds all over the place. As commenters before me in this thread have already emphasized, it all boils down to properly understanding odds and therefore likelihood. Chris focuses on a handful of events, while the correct attitude is to form a proper understanding of percentage over a massive sample. As a newcomer to Las Vegas I pretended I didn't need to know that in regard to sports. I remember one example, where I loved Ivan Lendl to defeat Boris Becker in the US Open. But the odds were much higher than I anticipated. One of the sharpest guys in town asked if I was still going to wager on Lendl. "Of course," I said arrogantly, "he's going to win, so what do I care what the odds are?" He laughed, and correctly so. Lendl lost but that's beside the point. I was ignorant of basic probability and therefore I had no chance in the long run. Very shortly thereafter, I did. It is all the difference. You take the best of it by a few percent as often as possible and it's remarkable how the the results and the luck spill in your favor.

Same thing in politics. No way Chris would be using 90% in regard to the House takeover if he had knowledge of true probability. There is nothing in his analysis or current form that even threatens 90%. It is a guess, and a very poor one. The result doesn't matter and I don't think Chris realizes that. Even if Democrats gain 25 seats, saying they have a 90% probability of taking the House is rank foolishness.

As lynx correctly points out, the cumulative knowledge of everyone in the market will inevitably trump one man's opinion in the long run, no matter who it is. That's why any huge sample of pointspread results hits almost exactly 50/50, in terms of how many favorites and how many underdogs cover. You will have individual favorites lose by 50 or win by 50, and one year the tilt might be 58-42 one way or the other, but drag it out long enough and it hits almost exactly 50/50. That demonstrates the basic odds are correct, and it will play out in politics also.

Playing the markets will indeed cause you to lose money, if you think a 50/50 scenario is 90/10.

by jagakid on Sat Sep 30, 2006 at 02:04:52 AM EST

Re: Gambling was easier prior to the markets (none / 0)

Duh. Of course there is a range of probability. What I reject is the notion that these markets set it for this or any other el$ection.

Why are the actions of the people who partiicpate in this market the defining trend for the range of outcomes in this eleciton? I outright reject the premise that the acitons of the people who partiicpate in this market define the probability of outcomes in this eleciton. I reject the premise that their changing perception changes the range of outcomes, or is reflective of it.

The issue has to do witht he quality of information people playing int he markets receive, and their ability to process that information. If they don't have the right information, and they are not good at processing it, then their conclusions are worthless. We have no way of knowing either category.

by Chris Bowers on Sat Sep 30, 2006 at 02:34:56 AM EST
[ Parent ]

Re: Gambling was easier prior to the markets (none / 0)

They don't set it, they reflect it, as you stated later.

Once again, you proclaim that markets cannot be trust because you proclaim that they cannot. You reject the use of markets even though research shows their value. You are un-moved by any evidence to the contrary.

Very astute. Again, so be it.

by lynx on Sat Sep 30, 2006 at 02:38:37 AM EST
[ Parent ]

Re: Gambling was easier prior to the markets (none / 0)

The day a market makes a prediction that goes against popular opinion and is right, then I will start to consider your position. But I don't think a amrket will ever reflect anything except popular opinion. And popular opinion is just that. It has no special insight or knowledge.

If there was an electronic market in this country on evolution vs. creationism, creationism would win. But that doesn't make it right.

by Chris Bowers on Sat Sep 30, 2006 at 02:54:43 AM EST
[ Parent ]

I'm with Chris (none / 0)

A reactive market is useless. That's what these political futures markets give you.

Go back and watch the spikes and drops in the 2004 Bush contract.

If I recall, across the markets, Bush bottomed three times:

1. After the First Battle of Fallujah, April 2004.

2. After the first debate, when Bush repeatedly forgot to wipe his chin before babbling.

3. When the first exit polls were leaked (you know, the ones that in any other country on this planet would have forced a new election).

In all cases, the markets were merely reacting to the CW. And since those reactions proved useless in predicting the ultimate outcome, the markets themselves are of no great use.

How do you define whether a contract predicted an outcome, unless you freeze trading before the outcome is decided?

In fact, an effect like this does exist in a real world futures market.

Watch how oil closes every Friday afternoon. It's basically a news futures contract on the likelihood the Middle East will explode before Sunday evening trading resumes.

A soft Friday of buying is pretty much a vote by the market that Israel and Iran will manage to keep their heads on this weekend.

Strangely, the predictive value comes from the halt in trading over the weekend.

Otherwise, with an ability to react more dynamically to ongoing events, the market becomes less predictive.

It's a thought. At some point, you have to freeze trading BEFORE the bad things happen.

That way, traders have to make clear decisions with no ability to get out before all hell breaks loose.

by jcjcjc on Sat Sep 30, 2006 at 03:14:55 AM EST
[ Parent ]

Re: I'm with Chris (3.00 / 1)

There is some validity to that thought. My biggest point is that even if you don't think the markets are spot on I don't see how you can't think they are in the ball park.

Do you really think the odds are 90% and the markets think they are 45%. Are the traders in the market that out of touch? If that were true then the market is not just inaccurate. It is not even useless. It is down right a contrary predictor that is very very wrong with lots of very stupid money in it.

That doesn't make any sense.

The truth is if Chris had predicted 60% and the markets were predicting 45% this thread wouldn't even have gotten started. Its the huge disparity that does not compute.

by lynx on Sat Sep 30, 2006 at 03:21:53 AM EST
[ Parent ]

Re: I'm with Chris (none / 0)

The problem is that neither of the two probabilities matters because this is a one-time event. Chris predicts 90% (I'd go with about 65%), but that means he still believes there is a 10% chance he could be wrong. Maybe this is one of those 10 times out of 100 that based on all available information, the opposite result occurs.

Maybe something more useful for the markets to do would be to gamble on several number of seats for Dem pickup. Let's create a probability distribution and THEN see how useful the market is for prediction. What's their prediction on Dems picking up 0-5 seats? 90%? 95%?. How about 10-15? 15-20? 40-50?

That, to me would be a better indicator of their strength of prediction.

by adamterando on Sat Sep 30, 2006 at 10:23:07 AM EST
[ Parent ]

Re: I'm with Chris (none / 0)

"There is some validity to that thought. My biggest point is that even if you don't think the markets are spot on I don't see how you can't think they are in the ball park."

Were the much larger, more diverse - and hence theoretically more accurate - stock markets "in the ballpark" on the proper valuation of telecom companies in 1998? Were they right - no, scratch that - were they "in the ballpark" on Enron in 2000?

Just hypothetically speaking, if it were revealed tomorrow that 90% of the traders on Tradesports were Republican officials or donors would that change your view as to whether we should acknowledge that they are in "in the ballpark"?

I've had this argument with a Republican friend who thinks that the exit polls were FUBAR in 2004 and yet loves Tradesports as a predictor. He maintains that asking hundred of thousands of people what they did 5 minutes in the PAST can't be trusted because of selection problems that may make the selection non-representative. Yet he insists that a *self-selected, GUARANTEED non-representative* set of traders making guesses as to an event four weeks in the future have some sort of magic predictive power.

I think he's batshit nuts.

by chaboard on Sat Sep 30, 2006 at 09:20:19 PM EST
[ Parent ]

He's not nuts (none / 0)

But it is time to refer everyone over Seth Godin's theories about marketing...

All Marketers Are Liars is a decent book in this regard. Just don't believe it all.

Godin's main point is that most people are simply shopping for the lie that best suits their ego. Your Republican friend likes TradeSports because in 2004 it sold him the answer he wanted to hear.

Likewise, he's worried about the exit polls, because we all damn well know if this were the Ukraine and not the US, those exit polls would have been the undoing of Bush and the GOP.

But, even in the failure of Americans to respond to the exit poll discrepancy, we see people shopping for a lie they want to hear.

Americans want to hear that America is a solid, stable, free, happy, rich, tall, sexy, wonderful, friendly democracy.

And we'll tke that lie at hand long before we'll settle for rioting in the streets to demand a fair election.

The GOP knows this, and they bust their asses every election to milk it just a little longer.

by jcjcjc on Sat Sep 30, 2006 at 11:26:11 PM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

"If anyone can predict exactly how 100,000,000 million people will decide to act between now and November 7th, then that person probably isn't human anyway."

There go the fields of sociology, social psychology, and city planning!

by lokiloki on Sat Sep 30, 2006 at 02:16:13 AM EST

Re: The Great Predictive Value of Elections Future (none / 0)

I don't remember any of those fields being able to predict exactly how things will happen.

by Chris Bowers on Sat Sep 30, 2006 at 02:29:00 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Uh, to clarify things, 100,000,000 million is 100 trillion. Which is several orders of magnitude larger than the population of the planet.

I don't think that's covered in "city planning".

by RickD on Sat Sep 30, 2006 at 06:18:40 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

As an urban planning student (city planning is really the British term), I can vouch for that.

by Sandwich Repairman on Sat Sep 30, 2006 at 08:41:41 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Incidentally, speaking of the prediction markets, we DID get some nice movement on tradesports today, re: house control. Dropped 6 points and stands at 52% GOP win (50.9/52.8 bid/ask).

Incidentally, has anyone else noticed some really weird movements in the tradesports senate races? Seems as though someone might be purposely buying at a very inflated price (say, 1 contract) and then when the market corrects to the natural bid/ask tradesports shows a massive change... e.g., RI-GOP +10, TN-DEM +4, etc etc. Reminds me of when someone tried to buy all the bush contracts on Iowa in 2004.

by lokiloki on Sat Sep 30, 2006 at 02:21:44 AM EST

Re: The Great Predictive Value of Elections Future (none / 0)

Of course in highly illiquid markets as the individual senate race markets are, manipulation is easy and often occurs. Its interesting how they tend to level back to somewhat reasonable levels though, even though the spreads are pretty wide sometimes. This is very difficult to accomplish in highly liquid markets such as control of the House and Senate. Those are two of the most liquid political markets on Tradesports.

Concerning Bush in 2004 Iowa Markets, what do you mean someone tried to buy all of the Bush contracts. I traded in that market in 2004. There were 150,000 dollars trading in that market and individual account balances were limited to 500 dollars. There is no way for anyone to corner a market of that size with 500 dollars. I did not see anything that looked like a corner attempt. I saw Bush contracts trading too high and I profited by selling the Bush contracts and buying them back at lower prices but I didn't see any attempt to by all the Bush contracts.

Could you explain what you saw. Just curious.

by lynx on Sat Sep 30, 2006 at 02:29:49 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

i will look for the graph but one day there was a HUGE spike for kerry win...

by lokiloki on Sat Sep 30, 2006 at 02:59:36 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Ah, here it is...

http://128.255.244.60/graphs/graph_Pres0 4_VS.cfm

Check that out and tell me that isn't the strangest darn thing...

by lokiloki on Sat Sep 30, 2006 at 03:01:09 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

That spike is easily explained.

IEM is a low volume market and the chart posts end of day prices not average prices for the day.

So what happens in an instance like that is someone buys a Kerry contract late in the day and takes out the last seller at the price he was trading at. For some reason there is an absence in pent up shares listed for sale and the next best share available for sale is way above the market price. All someone has to do is wait until 11:59:59 and buy one share at the price and that will be the print for the day.

It happens all too often on IEM and I wish they would chart the daily averages rather than the closing price because people do make such trades on purpose to make the charts look goofy.

It has no bearing on how the trading was actually going. Someone just exploited an absence of shares for sale at the end of the day. Thats why it dropped right back down the next day because when people woke up the next morning and saw the price it traded at overnight (again probably 1 share) the shares for sale start flooding the market such that that absence of shares for sale is not there the next day and cannot be repeated.

by lynx on Sat Sep 30, 2006 at 03:39:15 AM EST
[ Parent ]

Rigging (none / 0)

It isn't that hard to engage in rigging a contest, especially online.

First off, you can get shills ... anyone on eBay has seen it happen. eBay is a game of chicken among buyers with sellers placing shills into the game to heave them all over the cliff at the end.

Second, you can use multiple accounts. A little tougher if the game is thoroughly policing.

Third, you can mask your ID.

An aggressive player can buy a significant portion of the market if he is so inclined and others are not.

Worse, you can generate proxies and shills for practically nothing.

Look at the penny stock spammers, who frequently make a good markup on junk stocks by encouraging people to buy into them at a mark up with just spam emailings.

You get the flood, you sell your holdings, and then the stock tanks when the wave is over. People are pissed, and usually no one gets prosecuted, because anyone who buys stocks based on a spam email is retard. Yet, there are enough suckers out there that the scam continues to churn.

Rigging requires nothing more than the boldness to do it.

by jcjcjc on Sat Sep 30, 2006 at 11:43:25 PM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

OK. Can someone interpret what the markets are saying in terms of probabilities for the House and Senate control?

by ab initio on Sat Sep 30, 2006 at 03:22:52 AM EST

Re: The Great Predictive Value of Elections Future (none / 0)

that the GOP will retain both the house and the senate. the house teeters back and forth at around 52% GOP win liklihood. the Senate is around 80%.

go to tradesports.com and click on Politics on the left, then click on US 06 GOP Control, then click again on the submenu... last prices were senate-gop 79%, house-gop 52%.

Iowa is another market and is more bullish on GOP prospects sadly. however, the trend is definitely moving to the Dem side for the house...

http://iemweb.biz.uiowa.edu/graphs/graph _House06.cfm

Here is the senate graph:

http://iemweb.biz.uiowa.edu/graphs/graph _Senate06.cfm

Here is the overall graph:

http://iemweb.biz.uiowa.edu/graphs/graph _Congress06.cfm

by lokiloki on Sat Sep 30, 2006 at 03:31:23 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

The red line indicates people owning shares of Republicans gain seats in the House. Are these just idiots with money, or what?

by SaratogaProf on Sat Sep 30, 2006 at 10:16:07 AM EST
[ Parent ]

Re: The Great Predictive Value of Elections Future (none / 0)

Did you not read the previous discussion on odds?