greggles's blog

After working on the system for a bit I finally re-launched Open Prediction Marktets Beta site.

The beta.openpredictionmarkets.org site has two goals:

• To serve as the place running the leading edge of my prediction market code
• To help the Drupal community make more accurate predictions about its future

So, I read the wikipedia page on parimutuel betting and felt like that presents a similarly predictive market without the complexities of trying to build a markets concept with trading and all the fun that comes from matching trades. I really got stuck last time on two elements:

• The trading concept requires a lot of extra calculations and overhead, those lead to a complex system which leveraged a lot of contributed modules...

I just realized there are now a few of these, so I wanted to create a list:

• Bookmaker which has releases for 6.x and 5.x, but no commits in the last year.
• Betting which has a 6.x release, and the last commit was 5 weeks ago, but there is an issue in the queue that says it doesn't work.
• Prediction Market no releases and no open bugs, active 28 weeks ago.

Drupal user Rob Knight has begun work on a new module - http://drupal.org/project/pm - which aims to provide a market maker in Drupal.

A quick review of his TODO.txt shows a bit of work left (userpoints integration, in particular, is important).

I'm a big fan of HubDub. The user interface for events and trades is very easy to understand. And, lucky for me I'm in the top 10% of the leaderboard (my hubdub profile).

Recently, Simon Owens interviewed Nigel Eccles (who has no home page to link to...). The interview covered both HubDub and the sub-site PunditWatch which tracks pundits.

The "gaming" of anything is generally bad. In the case of prediction markets, there are occasionally some which easily be gamed by a broad audience. There is a new market on hubdub for:

How many results will a Google search for 'Switchy McSwitchenstein' yield by the end of April 2008?

Happy Path Trading

The "happy path" is done (has been for a while actually). You can buy/sell portfolios of contracts. You can buy/sell individual contracts as long as the number you want to buy/sell matches perfectly with someone else. This is pretty exciting to me. I've even got some validation at order entry and trade execution to make sure things "good."

Some Oddball Paths

Partial match (and vice versa):
1. Order placed for 100 items
2. Match placed for 50 items
3. Match satisfies the first 50 and is executed/closed

There were recently two launches of prediction markets that are worth writing about: HubDub and The Industry Standard.

Prediction Market with News Aggregator

No code discussion here - just information on what prediction markets are. Especially in light of some recent criticisms from the Democratic New Hampshire primary election, this overview post on the Mercury blog is really informative. It walks through how to interpret prediction markets

One of the hardest things to do is understand how often you can lose when betting on a favourite. Let’s do some simple math:...95% - 1 in 20 favourites will lose

I think I might drop "refactoring" from my descriptions because frankly small amounts of that are going to happen way to much to bother bringing up every time. But again, I did some ;)

Workflow States

I decided on the Workflow module which is simple to use, simple to setup and provides everything I need. It was particularly useful watching the video (available from the project homepage) .

More refactoring