Please help me figure this one out…

If I look at the outcome of a question like “Who will win this years competition?” you will have indeed one winner with a probability of 90 or above when the cashing out is about to happen…

That’s all ok, but if you look at the market in its whole, there’s only 10% left to be devided by other stocks and as such there is no comparison possible between the other stocks.

Also, if we state that team-a (the winner) was predicted with “there’s a 96% change that they will win.” and we look at the runner up which states “there’s a 6% chance that they will be the winner.” doesn’t that look wrong?
To me it seems that the runner up actually had a 85% chance of winning the competition and the winner actually 96% chance.

• winner of competition will be granted $100
• each team that end into the top-10 also get a $100 but only if the position in the competition is the same as the position in the market.
  So if team-x is 3rd in the competition and in the market it has the 3rd highest value, it will be awarded $100

Do you think the result of this market would represent really much better?
Or do I completely misunderstood prediction markets?
I also thought that the teams in the middle more often switch places and as such the market would be more alive if people were trying to predict those places.

Thanks for the question. But I’m probably misunderstanding your point.

To me it seems that the runner up actually had a 85% chance of winning the competition and the winner actually 96% chance.

The probability of all outcomes of something all have to add up to 1 (100%). So if the contest you describe only has one winner how can 2 people have a (.85 + .96 = 1.81) 181% probability of winning?

I wanted to have correct initial prices for my question “who will be the winner of this competition?”
So I was thinking how to do that… Luckaly the competition already started some months ago and as such 8 points could already be gained.
Assume that this is the current outcome:
T1|7
T2|6
T3|6
T4|3
T5|2
T6|0
So how can you know what the probability is of each team being the winner?
I think: (points * 100) / maxpoints
so:
T1|7|700/8|87.5
T2|6|600/8|75
T3|6|600/8|75
T4|3|300/8|37.5
T5|2|200/8|25
T6|0|0/8|0

As such T1 has 87.5% chance of winning the competition and T2 75% chance of winning the competition…

Then I though… Well isn’t this what the market should predict? rather then just saying T1 95% and T2 5% and all other 0%...
To me indeed T1 seems to be the winner and as such it should have 100% probability to be the winner… But I wonder about the other stocks… Their value must be really representative…
The 3rd in our example has a 75% chance of winning the competition… but using a single answer I cannot show this in my market…

That’s when I thought I should change the question and make it have multiple answers and ask “who will make the top-4?”.
In the details I then specified some payout rules…

I’m sure I’m getting something wrong here… so please help me figure this one out…

Assume different…
In business they could have a market of ideas and people are predicting which idea will be launched first by the company.
After a while we will have a clear top winner of 95% all others are merely having any percentage. They close the market.
Now the company launches their top idea, but another company launches more or less the same idea just before them.

Now they have NO idea what to launch because they don’t have a clue what the runner up idea is…

If they chose to have a multiple correct answer outcome, they would certainly have a runner up…

They could ask: which will be the next 3 idea’s that we will launch?

It sounds link you want to predict a ranking of proposed projects, and then use the rankings to decide which projects to choose, and then payoff the shares for the projects chosen.

But it gets a little tricky since the market turns into trying to predict which 3 proposals will be in top three proposals at the end of the market period. The problem with running such a market when the market uses an automated market maker (at least the version that Inkling uses) is that the last trader in the market can effectively pick which three projects end up with the highest price and therefore become selected (if market price is the sole criteria for project selection, and assuming the trader has sufficient funds). So rather than ‘wisdom of crowds’ you get the decision of the last trader in the market.

Haven’t thought about this much, but a continuous double auction style prediction market may be less subject to this criticism when used for this purpose.

Pretty good remark…
I do believe that the influence of 1 single person indeed needs to be limited like x% of your funds can be put into one single stock and x% of your funds can be invested in a single market…

Those with the most funds still can have a bigger impact on the market, but then again they have bigger funds because they are better in making predictions… Just the altitude of the impact is limited…

For the double auction style you need to map a buy with a sell and to most people, this is too complicated…

I did read something that was interesting where they said that a person with a larger amount of funds gets weigthed higher automatically when calculating the new rate of the stock because his track record was better. This extra coefficient was actually applied per category so that in one category you have a higer rate then another…

I’m still puzzled if multiple answers are better then a single outcome both in business and pleasure markets…
Pikking a single outcome sounds too much like a poll to me…
I guess I need to do some more reading on the subject…

onemike: Inkling is meant to help you predict something you can measure later on. Will something happen? How many of X will happen? Etc. This thing that you measure at the end you want to be independent of how the prediction market works.

If you are just asking, “which project should we pick”, this is a much better question for a traditional poll or take a vote.

This isn’t the fault of our market maker or using some other mechansim. It’s just not a good fit for what a prediction market is meant for. A prediction market is meant to predict something. If you ask your market to just trade and we’ll pick the highest priced stock as a project and then pay off that highest priced stock just because it’s the highest priced stock, people are now just trying to predict what other people are going to pick as the highest priced stock. There’s no actual feedback loop on anything important.

If the project ends up losing you a million dollars, how do you “punish” your traders for choosing a project with a negative ROI? You’ve already rewarded them for just picking the stock they thought everyone else was going to pick, not on some criteria that is completely independent from the market.

ednique: I’m still having a tough time understanding what your getting at. I don’t really know what (points * 100) / maxpoints is an equation for. Probability is the measure of how likely an event is. All the probabilities of all the possible outcomes of an event have to add up to 1.

Let’s pretend the 6 teams in your example, were complete identical (like the side s of a coin) you would say they all have an equal chance of winning. What is the probability of T1 winning: 1/6. That’s about 17%. How about T2. It’s the same 1/6 17%. Add them all up and it equals 1 or 100%. That’s all you can work with.

Also, if we state that team-a (the winner) was predicted with “there’s a 96% change that they will win.” and we look at the runner up which states “there’s a 6% chance that they will be the winner.” doesn’t that look wrong?

If the market actually thinks that this team has a 96% of winning the rest of the field has a 4% of winning since 96+4 = 100. There’s nothing wrong with how this looks. The crowd feels that team 1 is awesome, and almost a sure thing of winning this thing. The rest of the field doesn’t have much of a chance. Well they have a 4% chance.

If I look at the outcome of a question like “Who will win this years competition?” you will have indeed one winner with a probability of 90 or above when the cashing out is about to happen…

Inkling isn’t always spitting out these sure thing predictions because life isn’t filled with these sure things. Look at:

http://home.inklingmarkets.com/market/show/6569

The Sopranos definitely had a better chance of winning than any other show in their category. But the rest of the field as a whole had a chance of winning of over 50%.

ednique: It looks like the real goal of your market is not to dwell in the world of probabilities but you might just want to compare how many points your teams are going to score. You could just create a market by choosing “The answer to your question will be a number.” and then “Determine a specific number (e.g. xx.yy)” and ask “How many points will these teams score in the competition.” This way you can create stocks for each of your teams, and have traders pick actual point scores of each team. If you only have a max of 8 points in the game you are describing, make sure you pick a scale of .1 so you traders have a bit more room to play. So a price of $80 will equal 8 points.

Nate: I didn’t mean to imply criticism of the market maker or anything other than, as you put it, “It’s just not a good fit.” I’m a big fan of the LMSR. ;)

You guys are great… Thanx!
I now figured out what my problem really is… The thing is that in this particular competition the number 1 spot is so dead easy to predict that I was puzzled about the possible outcome and wanted to make the prediction more interesting.

The proposal of nate might be just what I am looking for!
Asking the right question is more important I realize now.
and Nate… I wasn’t complaining about the soft eighther…

Nate: the equation is actually “how much % of the maximum points did a particulart team makes up till now” where I assumed that if they perform exactly the same troughout the competition, that they have that % chance of winning the competition as 100% is winning all points.

But as I realize now, I’m asking the wrong question. The how many points question will much better represent what I am looking for.
In that way people will be trading on all stocks instead of the few possible candidates and each weeks the points will change and as such people will have an urge to trade week after week during the whole competition.

My “which project should we pick” question is indeed more of a poll although in a poll you ask the voter’s choice whereas in a market people will trade in a stock where they think most people will trade in…
which is not neccesary their own choice…
Maybe you people have some document or something else where sample questions are given of good working markets for business?