What is our Crystal Ball telling us?
The World’s 1st Anime Prediction Market
As the last season of Game of Thrones has shown, people love to speculate what happens next in their favourite TV shows. What if you could also win stuff if you correctly guessed the ending of the season? Could this engagement sustain a blockchain-powered prediction market ? We built a platform to find out. We call it Crystal Ball. We chose not to train it on GOTS8 but rather our beloved Japanese anime.
What is Japanese anime and why is it suited to predictions?
Every year Japan produces around 200 anime TV series (about 50 per quarter either new release or continuing seasons) Many of the shows broadcast 1 episode a week on Japanese linear TV (following that they may be simulcast on streaming services in and outside Japan). We know that many anime shows are video adaptations of popular manga that if you have already read you may be able to predict the outcome, however, many anime storylines diverge from the manga and some anime are originals that are not based on a manga at all. There are many Reddit forums dedicated to discussions of each episode. With Crystal Ball anybody can
(1) guess what will be the top show of the season;
(2) guess the answer to predictions created by fans;
(3) create their own prediction related to any show or topic.
Correctly guessing outcomes leads to winning our proprietary Punk Points and climbing the leaderboard. And what do Points make? … Prizes!
How do we decide the top show of the quarter ?
We decided to use the worldwide google trends ranking of the title based on search volume combined with the number of comments/upvotes on Reddit related to the show. This allows us to create a consistent global metric that is agnostic to broadcast or streaming.
What is a blockchain-powered prediction market ?
Paraphrasing from this nice description and review here: “Prediction markets are a special kind of betting platform that can be used for multiple purposes ranging from insuring a crop from bad weather to predicting the outcome of an event based on the wisdom of the crowd. Among other popular uses of prediction markets are hedging an investment, shorting the market, creating a futures market and predicting the outcome of an election.”
Why use blockchain in a prediction market ?
The exciting thing is that the outcome of a bet can be determined and resolved without using any central authority, through software and the community. The results can also potentially be more accurate because of the reliance on the wisdom of a crowd distributed around the world in the network.
Why prediction markets for film and TV ?
Data for Creators. If we can scale prediction markets to accurately track what story outcomes fans are predicting, creators can leverage this data to inform their choices when producing new shows. Creators could also make rollout of new episodes, seasons or sequels more interesting by providing incentives for guessing outcomes correctly. Points on our platform could also be used in exchange for merchandise or for driving engagement with the community. Hopefully the result is more fun for fans and has real benefits for creators.
What were the results ?
We received over 1,300 story predictions or votes from over 1,100 people and we are still in the middle of the Spring 2019 season.
What have we learnt so far ?
- Fans love to predict (and discuss) story outcomes.
- Predicting story outcomes is more popular than predicting the most popular show.
- 3% of fans will create their own predictions (versus voting on others predictions).
- Fans love to win free stuff (duh).
- Fans not so interested in winning money from betting on predictions.
- Rich opportunity to mine data as evidenced by the wonderful tableau_kun (Reddit) and his awesome universal anime rankings.
- Checking the outcome of every single prediction is labour-intensive and something that could potentially be driven by the community in a more de-centralised implementation of the market.
- We devote a not insignificant amount of time to editing the questions and communicating with predictors to clarify the meaning. Many predictions submitted were questions with non-binary outcomes indicating that it is not as easy to grasp the concept of a prediction versus a simple opinion poll. This makes us think that Prediction markets are much more suited to sports where there is a win/lose/draw outcome that is easy to match your answer to and create a prediction for.
- Google Trends is a blunt instrument that is good for a high-level comparison of up to 5 search terms but requires some creativity and hacking in order to compare more than 5 search terms on a regular basis (in different languages at that).
Can I predict the future of Crystal Ball?
I can predict with high accuracy that we will continue Crystal Ball as a fun way to track all the new anime releases, but this will not be our core business. A true tokenised prediction market earns value for the operators either through (a) the value of the token or (b) through commissions on the amount of money/tokens staked in the prediction. (a) requires a significant investment in tokenising the platform and adopting or competing with existing well-funded protocols (Eg. Gnosis, Augur, Veil) (b) requires navigating betting licenses.
Our Greatest Crystal Ball Story
Let me leave you with this reminder of how wonderful the Internet is. We gave out BlockPunk Merch as prizes to weekly points-winners. Check out our cool hoodie below as an example.
One of the winners is an anime fan living in Vietnam. So my friend Meraj is on a business trip to Ho Chi Minh and I suddenly receive this Whatsapp message:
I reached out to our friend in Vietnam and confirmed that he did indeed have a random guy shouting at him on a bike recently! We have agreed to have a drink in Saigon between the 3 of us sometime to celebrate our cosmic connection. Even Mystic Meg couldn’t have predicted that outcome :)
