Mercury's Blog

I’ve previously discussed in my post “Developing a business case for Prediction Markets” that the costs of forecasting error should be relatively straightforward. I wanted to expand on that a little more, and today I added a page to this blog that helps you calculate a forecasting error.

By navigating over to my new page “Forecasting Error Calculator,” (also on the top right under “Pages” if you’re not reading this in an RSS reader) you’ll see an embedded spreadsheet. The example demonstrates that for one particular company with 800 million in sales, each percentage point of forecasting error costs between 3.03 and 3.65 million.

It’s not unusual for a company to have a forecasting error of several percentage points, thus potentially costing them tens of millions in sales. But this shows the opportunity for a prediction market! If by reducing a forecasting error by just one percent, it could save this particular company between 3.03 and 3.65 million. It’s also not unusual for a prediction market to reduce error by much more than 1-2%, thus saving much more.

When you check out this new page, you will also see that I’ve provided a link for you to download the original spreadsheet yourself so you can experiment with it.  (No payment or registration needed!)  Different industries have significantly different cost structures, so results may vary.

I would appreciate your feedback if this is helpful for you, particularly as you think about building a business case for prediction markets. Would having the calculator in a different format (such as Flash or a web-based form) be more helpful? Please get in touch.

———————–

I’m happy to announce a new look for the Mercury Research & Consulting website. If any of you tried to go to the homepage in the last few days, you will have received an error as I was shifting servers. It’s all up and running again, and I invite you to take a look. If there’s a project that you’d like to discuss, please get in touch.

———————–

People who are reading this post on the website (instead of an RSS reader) will see a widget in my sidebar: my “Shared Items” from Google Reader. These are blog posts that I’ve read that really impacted me and that I wanted to share with you. Feel free to click on this link to see them all. If you read blogs in an RSS reader, click here to subscribe to the feed.

I think that you’ll find some thought-provoking and inspirational posts for your reading pleasure.

Yesterday, Chris Masse of Midas Oracle pointed out that someone had asked a question regarding prediction markets on LinkedIn. (Link to question.) I posted an answer, and this is what I wrote:

[...] I’m a consultant in prediction markets. I work with companies to find where they have business problems that a prediction market could solve, and help them implement a solution. Getting a prediction market website up and running is technically fairly easy; getting it working correctly and providing good data and a good experience is much tougher. [...]

Unfortunately, Chris has potentially mis-interpreted what I’ve said. He wrote in a post today:

[...] what he said could be interpreted as meaning that the software vendors don’t provide the right kind of advice. It could be the case sometimes, but it’s a sure thing to say that the software vendors that are the market leaders, because they have seen so many business cases, do provide great insights.[...]

This is absolutely NOT what I was saying if you read my original answer on LinkedIn. The prediction market software vendors that provide consulting services are great resources because of their extensive experience. What I was trying to point out is that it’s easy to start a prediction market these days. There are a number of software solutions, and though many are proprietary, there are open-source packages, too. Getting a market running from the IT side is very straightforward.

The value I and other prediction market consultants bring to the table, whether associated with PM software companies or academia, is all of the supporting structure a market needs to run. I’d like to refer to my post on the book “Reinventing the Bazaar,” which explains some of these concepts. Despite the rhetoric, markets need a well-thought-out supporting structure in order to operate effectively, and this is true for prediction markets just as it is for Dutch auctions in flowers or the NYSE. Without it, a prediction market will likely be unsuccessful.

What I bring to the table in my consulting is a total focus on my client’s needs, since I’m not trying to sell them any given software solution. I do not offer prediction market software because there are simply too many good packages already out there, and it’s not my specialty. Instead, I work with my clients to develop their needs and determine how to best serve them.

I hope this explains my thoughts in more detail. Again, please contact me if you’re interested in discussing your prediction markets project.