Prediction is hard, or perhaps I should say that accurate prediction is hard. The more I study prediction markets, betting odds and the like the more I wonder how transferable those theories are to startups. So I decided to run a small experiment….

Perhaps that work at the Sporting Life did actually rub off on me.

Parimutuel Betting with TechBet.

The rules of the road are quite simple:

  • You’re trying to predict the winner of Invent2016.
  • There’s definitely NO money involved.
  • Right now there’s no signing up to play, just make a bet and move on.
  • The betting is open until 7:30pm when the event starts, hopefully this means that no sneaky judges can pile a load of bets on at the last minute.

I’ve limited bet amounts to between 1 and 100 TechBet credits (i.e. a virtual currency with no value, just bragging rights when the winners are listed).

PS: This has nothing to do with Catalyst Inc. apart from they were running the event and it was the perfect candidate to use as a test bed, so if anyone from Catalyst is seething at the sides then I do apologise.


If you want to play then you can, just click on this link.

How Does The Pool Work?

There are no odds in parimutuel betting, what happens is that players place their bet by choosing the winner, during the course of the bet each entry will have an amount of money against them. Here’s an event with four outcomes and the amount that so far has been placed on each.

Name Total Bets
Entry 1 25
Entry 2 78
Entry 3 44
Entry 4 12

If we sum up the amounts bet we get the total amount in the pool. It’s just a sum of all the bets placed.


So in this instance there is a total of 159 “coins” in this event’s pool, this will be split against the winning entry, the payout is then calculated on the number of players and what they individually wagered on the event. So, for example, if Entry 4 won the event, there was a total of 12 “coins” wagered on Entry 4 winning.

159 / 12 = 13.25 per 1 unit bet (the original value of 1 coin bet placed and a profit of 12.25, so the odds are basically 12.25/1

If Player1 had bet 9 on Entry 4 and Player25 had bet 3 on Entry 4 then the pay out would look like.

Player1 = 13.25 * 9 = 119.25

Player25 = 13.25 * 3 = 39.75

Now I’m assuming there’s a 100% payout on the pool, in any normal situation is concerned tax and commission would have been skimmed off the pool before the payout was calculated.

Why On Earth Are You Doing This?

Why not? A couple of reasons, firstly I want to see if people are willing to take a punt with no risk. It also gives me data on the distribution of bets, do people already have a favourite winner of Invent2016? Within this one competition there is a lot of data I can learn from.

Secondly from a proof of concept point of view this was built in a couple of hours and was a good exercise in pinpointing what actually needed building, a tight specification and a narrow view of the deliverable. Yeah it’s on Bootstrap, yeah it looks a bit stale but it does the job.

Thirdly there was some deployment things in Amazon Web Services I’ve been trying to figure out, that ended up being the hardest part of the night, security groups can be a nightmare.

Go and have fun and good luck to all the finalists tonight. Once the winner is announced, I’ll collate all the bets and publish the results on Friday.