PredictIt

[Arkady]

Political futures markets, like PredictIt, are interesting, since the requirement for someone to put his money where his mouth is keeps things a lot more realistic than among the chattering classes, where there's no penalty for saying something that ends up being wrong. To move the market, you need to be willing to lose money if you're wrong.

Right now on the question of who will be the Republican nominee, the most probable doesn't even come at 50% likelihood! Trump is at 48 cents, which effectively means 48%. Cruz is a distant second at 29%, but there are all sorts of other options, because of the possibility of a contested convention. Ryan's at 11%, Kasich at 10%, Romney at 3%, and Rubio at 1% (things don't add up to 100%, because pricing builds in a profit for the market).

On the Democratic side, things are simpler. Clinton is at 84%, Sanders at 17%, with some shreds left over for Biden and Warren, plus a long list of people at 1 cent.

For the overall election, people putting their money where their mouths are have set the likelihood of a Clinton win at 61%, Trump's at 19%, Cruz at 13%, Sanders 12%, then a lot of low probability offerings. 69% says a Democrat will win the White House, versus 32% for a Republican. Julian Castro and Perez are the highest ranked Democratic VP possiblites, while Kasich and Christie are the highest ranked republicans. There's a 27% likelihood of Trump running as a third party candidate.

The likelihood of a Democratic landslide for the White House is 40%, versus 12% for a GOP landslide. The probability is 63% that the GOP will lose the Senate. The probability of undivided Democratic government is only 18%.

 

[Lukey]

Well, since you are a lawyer, and not a mathematician, I suppose it would be helpful for the group if I point out that those are not actual "probabilities," despite your misleading suggestion that they are. They are odds. Like in sports gambling, it reflects the narrow collective opinions only of those actually placing wagers. That's not the same thing as, say the "odds" of a roulette ball landing on red vs. black. In fact, I would say it is way less sophisticated than sports bookmaking or, say, horse race handicapping, in that there's likely only a few thousand people actually placing wagers on that exchange. But hey, it's better than throwing darts at a wall...

 

[marvin martian]

Well, since you are a lawyer, and not a mathematician, I suppose it would be helpful for the group if I point out that those are not actual "probabilities," despite your misleading suggestion that they are. They are odds. Like in sports gambling, it reflects the narrow collective opinions only of those actually placing wagers. That's not the same thing as, say the "odds" of a roulette ball landing on red vs. black. In fact, I would say it is way less sophisticated than sports bookmaking or, say, horse race handicapping, in that there's likely only a few thousand people actually placing wagers on that exchange. But hey, it's better than throwing darts at a wall...

But, factoring in @Arkady's duty to propagandize for a DNC victory, it makes sense why he posted this.

 

[Lukey]

But, factoring in @Arkady's duty to propagandize for a DNC victory, it makes sense why he posted this.

Did you notice how he subtly conflates the two concepts and, by the end, he's referring to them as "likelihoods" rather than odds. It's like an amoral defense attorney trying to get his guilty client exonerated - he'll plant a seed in the jurors minds by implying that some evidence is more factual and weighty than it actually is. I guess they teach that sleezy stuff in law school, which is probably why it is considered such an ignoble profession...

 

[Arkady]

Well, since you are a lawyer, and not a mathematician, I suppose it would be helpful for the group if I point out that those are not actual "probabilities,"

Yes, they're not actually probabilities or I would have said they were actually probabilities, rather than effectively probabilities, as I said. And, as I pointed out, they don't add up to 100%, which obviously actual probabilities would. In a perfectly efficient market, though, the cost of the purchase would be equal to the collective assessment of the probability of pay out (if an 85 cent purchase wins you a dollar and you win 85% of the time, then it's correctly priced, since your likely return equals your investment). That's made more complicated by the transaction costs, plus the time value of money (given the delays on the pay out), plus the chunkiness from them not allowing sub-penny pricing, etc. But it's pretty darned close. If you think there's a big gap between the cost of the share, in pennies, and the probability of it paying out, in terms of percentage, then you have an arbitrage opportunity: go ahead and make a big purchase in whichever direction you think it's skewed relative to the probability. But if the market perceived that arbitrage opportunity, it would already be taken and that would already have shifted the price.

 

[Arkady]

Did you notice how he subtly conflates the two concepts and, by the end, he's referring to them as "likelihoods" rather than odds. It's like an amoral defense attorney trying to get his guilty client exonerated - he'll plant a seed in the jurors minds by implying that some evidence is more factual and weighty than it actually is. I guess they teach that sleezy stuff in law school, which is probably why it is considered such an ignoble profession...

If you were better at math you'd get this. But, as we both know, it's never been your strong subject.

 

[RickWA]

Yes, they're not actually probabilities or I would have said they were actually probabilities, rather than effectively probabilities, as I said. And, as I pointed out, they don't add up to 100%, which obviously actual probabilities would. In a perfectly efficient market, though, the cost of the purchase would be equal to the collective assessment of the probability of pay out (if an 85 cent purchase wins you a dollar and you win 85% of the time, then it's correctly priced, since your likely return equals your investment). That's made more complicated by the transaction costs, plus the time value of money (given the delays on the pay out), plus the chunkiness from them not allowing sub-penny pricing, etc. But it's pretty darned close. If you think there's a big gap between the cost of the share, in pennies, and the probability of it paying out, in terms of percentage, then you have an arbitrage opportunity: go ahead and make a big purchase in whichever direction you think it's skewed relative to the probability. But if the market perceived that arbitrage opportunity, it would already be taken and that would already have shifted the price.

Pretty shamelessly weaselly, Ark. Does your family know that you have soiled Pampers for a conscience?

 

[as seen on tv]

But then there's the computer, which has accurately forecast every election, programmed by a college professor.

the computer says Trump is a 97% "lock"

http://www.aol.com/article/2016/02/...the-nomination-hell-win-in-november/21319450/

predictit wasn't around in 2008, when the college computer got Obama right. And PredictIt simply measures the media noise being reflected off of its wagering playas. Some of who are voters, some are low information voters, and some have never voted. Not really the same things as likely voters, is it?

I'm not placing a bet on either the computer or PredictIt. but if it gives you comfort to tell yourself that the same crowd which drove BitCoin value up and down like a turbocharged yo-yo is now confident who will win the election, by all means be consoled with Hillary, LOL

 

[marvin martian]

Did you notice how he subtly conflates the two concepts and, by the end, he's referring to them as "likelihoods" rather than odds. It's like an amoral defense attorney trying to get his guilty client exonerated - he'll plant a seed in the jurors minds by implying that some evidence is more factual and weighty than it actually is. I guess they teach that sleezy stuff in law school, which is probably why it is considered such an ignoble profession...

Lying via omission is his go-to tactic, always has been. Lawyers are masters of plausible deniability (i.e. "I never actually said ______"). Sadly, it works on the, shall we say, less intellectually well-endowed members of the community.

 

[RickWA]

Sadly, it works on the, shall we say, less intellectually well-endowed members of the community.

Hence, its status as Democrat communication tool numero uno. That's a base that is forever ripe for the picking.

 

[Arkady]

Pretty shamelessly weaselly, Ark. Does your family know that you have soiled Pampers for a conscience?

Not shameless. Just math. Learn some.

 

[Arkady]

But then there's the computer, which has accurately forecast every election, programmed by a college professor.

ANYONE can come up with a model that correctly "predicts" every election after the fact simply by designing the model to fit the known data that's already on the books. With enough data to sift through, a consistent alignment, however random, will inevitably emerge. Maybe it'll be that whatever candidate's mother has the alphabetically later first initial of her middle name always wins, or something like that. Draw the bullseye around the cluster of known hits and it looks like accuracy. The real test is how well the model does it predicting UNKNOWN results.

Anyway, his last run in an election without an incumbent had him predicting an 0.2 percentage point win by Obama. The real results had a 7.2-point Obama win. So, he was off by a pretty big margin. We'll see if his model does any better this time.

But, if you're confident he's right, then mortgage your house and buy Trump on PredictIt, because right now it's a huge bargain and you can make a fortune!

 

[RickWA]

Not shameless. Just math. Learn some.

You didn't leverage math, you leveraged your usual integrity-deficient shysterism. Some things never change.

 

[resident_evil]

I hope Jim Webb runs as an independent.

 

[as seen on tv]

ANYONE can come up with a model that correctly "predicts" every election after the fact simply by designing the model to fit the known data that's already on the books. With enough data to sift through, a consistent alignment, however random, will inevitably emerge. Maybe it'll be that whatever candidate's mother has the alphabetically later first initial of her middle name always wins, or something like that. Draw the bullseye around the cluster of known hits and it looks like accuracy. The real test is how well the model does it predicting UNKNOWN results.

Anyway, his last run in an election without an incumbent had him predicting an 0.2 percentage point win by Obama. The real results had a 7.2-point Obama win. So, he was off by a pretty big margin. We'll see if his model does any better this time.

But, if you're confident he's right, then mortgage your house and buy Trump on PredictIt, because right now it's a huge bargain and you can make a fortune!

so are you wagering on Predic-Tit or not?

Put your money where your mouth is.

 

[Lukey]

Yes, they're not actually probabilities or I would have said they were actually probabilities, rather than effectively probabilities, as I said. And, as I pointed out, they don't add up to 100%, which obviously actual probabilities would. In a perfectly efficient market, though, the cost of the purchase would be equal to the collective assessment of the probability of pay out (if an 85 cent purchase wins you a dollar and you win 85% of the time, then it's correctly priced, since your likely return equals your investment). That's made more complicated by the transaction costs, plus the time value of money (given the delays on the pay out), plus the chunkiness from them not allowing sub-penny pricing, etc. But it's pretty darned close. If you think there's a big gap between the cost of the share, in pennies, and the probability of it paying out, in terms of percentage, then you have an arbitrage opportunity: go ahead and make a big purchase in whichever direction you think it's skewed relative to the probability. But if the market perceived that arbitrage opportunity, it would already be taken and that would already have shifted the price.

You most certainly implied it. Your "conclusion:"

The likelihood of a Democratic landslide for the White House is 40%, versus 12% for a GOP landslide. The probability is 63% that the GOP will lose the Senate. The probability of undivided Democratic government is only 18%.

That was a bald faced lie, was it not? Those are NOT "probabilities" at all. They are simply bookmaking "odds." Again, for those following along at home. The chances of a roulette ball falling on red or black (or 00) are "probabilities." The chances of drawing a King for an inside straight on a one card draw are a probability. The "odds" on the winner of the Superbowl or the Kentucky Derby are NOT "probabilities." They are guesses, and bookmakers will set the "odds" of such events based on the bets placed on each outcome. Bookmakers attempt to move the odds as bets come in to try and keep them balanced (so they make the vig). That tends to be way more accurate than blind guesses, which is what is happening here. This far out these "odds" are nothing but a collection of fantastical guesses made by a very small (gambling addicted) slice of the population. Your results may vary...

 

[Deleted member 21794]

ANYONE can come up with a model that correctly "predicts" every election after the fact simply by designing the model to fit the known data that's already on the books.

I agree with that. In my stock and futures trading travels, I read a good dozen or so books. Most of them had "systems" that proved their validity by back testing. All you had to do was follow these system rules for trading, and you'd be a multimillionaire in a few years- starting with just $10,000. Funny though, they didn't talk about how their system worked AFTER the author came up with it. Frankly, if I came up with a system like that, I'd be too busy piling up cash and traveling the world on the profits to bother writing a a silly book that might sell what a few tens of thousands, MAYBE 100,000 if you're lucky? Perhaps two of the ten books I read were honestly educational and realistic.

 

[Arkady]

so are you wagering on Predic-Tit or not?

Put your money where your mouth is.

It only makes sense to play Predictit if you think it's WRONG. If the collective wisdom of the market has mis-priced those shares relative to the actual probability of the outcome, then one or the other share is a bargain and a person should buy it. But if the collective wisdom has priced the shares pretty close to their real value based on probability of outcome, then there's no point buying them, since they're worth more or less what they'd cost to acquire. I see no reason to think Predictit has it wrong by much, so my money would be better put into stocks or some other vehicle with fundamental intrinsic value. But, for those who think those prices don't reflect real probabilities, this is a buying opportunity. So, will you put your money where your mouth is?

 

[Arkady]

That was a bald faced lie, was it not?

Clearly not. I was simply and clearly stating what the market is telling us. That doesn't mean the market is necessarily correct. If you think it's wrong, this is a buying opportunity and you can put your money where your mouth is by buying whichever share you think is under-valued relative to the real probable outcome. It's possible that this market isn't very good at predicting, especially if you're right about how small the market is. So, you have an opportunity. For my own part, though, I think the market's likely to be a fairly good appraisal of actual probabilities, so it would be dodgy for me to try to substitute my judgement and wager my money on the market being wrong one way or the other.

 

[Lukey]

I agree with that. In my stock and futures trading travels, I read a good dozen or so books. Most of them had "systems" that proved their validity by back testing. All you had to do was follow these system rules for trading, and you'd be a multimillionaire in a few years- starting with just $10,000. Funny though, they didn't talk about how their system worked AFTER the author came up with it. Frankly, if I came up with a system like that, I'd be too busy piling up cash and traveling the world on the profits to bother writing a a silly book that might sell what a few tens of thousands, MAYBE 100,000 if you're lucky? Perhaps two of the ten books I read were honestly educational and realistic.

Trading is different than investing, and it is almost impossible to use a "system" for trading. Because, as you point out, if the market was that (short term) predictable, everyone would be doing the same thing. Trading is gambling and some guys really can rack up a long term winning streak. But eventually most traders wipe out, unless they are proprietary traders in a bank where they are just making a market, in which case gains and losses will tend to offset (not a great way to make a profit). There are, however, some good technical systems for investing, the oldest being the Dow Theory. It's all based on probabilities and can have a good record over the long term but there are periods (such as now) where a whip sawing market will keep switching back and forth and you can accrue a series of small losses that add up to a big loss over time. There's no easy way to get rich (unless you are a Clinton).