Prediction Market Tutorials
Does It Really Work? Wisdom Of The Crowds...
As James Surowiecki brilliantly describes in his book "The Wisdom of the Crowds", the prediction markets concept allows aggregating opinions of a group of people in a very elegant & precise manner.For this to work it requires that your market participants are not all with the same opinions and habits, and you need to give them the freedom to buy & sell as they believe - without being judged or pressured (anonymous betting are a good way of achieving the last one).
Experts also agree that market participants must be given an incentives for this to work.
However an incentive does not have to be money - it could be status or respect (see our Qmarkets Groups tutorial).
And YES - if you use this tool wisely, it will provide you with excellent results.
And unlike surveys, where you'll need a very large group of people in order to get reliable results - prediction markets usage shows that a small group (even 50 people) are more than enough.
Go through our Resources section to see articles, researches & news stories about usage of prediction markets in the real world.
Our blog also contains some editorials that we write, with examples of usage and more technical information that could be very helpful.
More Complex Markets - Multiple Choice
In our example so far we have looked at one example, which was rather simple.However, prediction markets at Qmarkets can address much more complex questions - and still keep the simplicity to the market participants.
For example, a more complete question in our example will be:
"Which of the following products will generate more revenue for our company on 2009?"
This is a multiple choice question with several possible answers:
- MP3 Player That Makes Coffee
- Coffee Maker That Plays Songs
- Solar MP4 Player
- Anoter Product
Each of these answers (or "Stocks" in QMarkets) will behave as a separate market - each will have a value, which reflects the probability that this answer will become true.
For instance, if the "MP3 Player That Makes Coffee" has value of 35, it means there is 35% probability that this product will generate the biggest revenue.
However, there is one difference - total sum of values for these 4 answers will always equal 100 - since there is always a 100% probability that one of these answer will be true, since these are all the options (including the "Another Product" option).
So, any time a market participant decides to "Buy" one of the answers, its value will increase. And in order to keep everything equal to 100%, it means that the other answers' values will decrease accordingly.
And on the other direction, if you want to "Sell" one of the answers (i.e, you think the probability is lower than current value), the other answers' stock values will increase accordingly.
This is all done automatically at Qmarkets, so you don't have to do anything for this to work - just let your users participate, and we will take care of the rest.
What About Markets Predicting Sales? Or Other Numbers bigger than 100?
For this method Qmarkets provides you with another kind of question - numeric questions.In these questions, you have to provide a range that is different than 0 to 100, and your market values will fluctuate between these numbers.
For instance, if you want your market to answer the following question:
"How Many MP3 Players will our company sell by end of 2007?", or
"What Will Be Our Company's Revenue for 2007?"
You define a range for your market - minimum value and maximum value.
For our "MP3 players" example, we could use 1-3 Million units.
You should pay attention to the range you provide - if your range is too big, your precision will drop dramatically, since every buy or sell operation will move prices more.
So if theoretically range could be 1-3 million, but you are very positive that real values are between 1.4 to 2.3 Million - use those numbers.
Another alternative to numeric questions, will be to define a multiple choice market for the same question:
"How Many MP3 Players will our company sell by end of 2007".
Your answers (stocks) in this case will be:
- Less Than 1 Million
- Between 1-1.5 Million
- Between 1.5-2.2 Million
- Between 2.2-3.0 Million
- More Than 3 Million
As an optimization, you can take the winning range, and use it to define a new numerical question for that range, in order to find the real number inside this range.
Combining the two options will really give you best results.
Previous <== What Happens If I Don't Sell?
Next ==> Qmarket Groups