The Manifold Team is doing a hackathon tomorrow (July 12).
What are some cool ideas? They should be scoped to something an engineer could complete in one day.
From Discord
Hey everyone,
I was mulling over an idea that I think could be an interesting feature for Manifold Markets. You're all probably familiar with the VIX in the stock market, which serves as a measure of volatility. What if we developed a similar index for Manifold, one that quantifies the unpredictability in our prediction markets?
Here's how I picture it working: We select a representative group of markets, possibly those with high liquidity or determined by a committee, steering clear of self-resolving markets, derivative markets, and the like. The goal is to get a set of markets that truly capture the essence of Manifold.
The "Randomness Index" itself would be designed to capture not just the individual market uncertainties, but also the correlation between them. We would use the forecasted probabilities (referred to as 'theta') from each market and assume they follow a multivariate normal distribution. The 'mu' (mean) would be the current estimated probabilities and the 'sigma' (covariance matrix) would be estimated from the historical data.
To measure the spread of this distribution, we would use the Mahalanobis distance, which gives a sense of the overall dispersion given the covariance structure. By generating a multitude of potential 'future states', we can calculate this distance for each, then summarize these distances (average, max, median, etc.) to form the "Randomness Index".
This would provide a quantitative measure of overall 'randomness' or unpredictability across our chosen markets. From a marketing perspective, it could also offer a unique feature - a real-time measure of global uncertainty, unique to Manifold. Besides, it might offer users an intriguing way to visualize and comprehend the dynamics of our platform.
I'm curious to hear what you think about this idea. Would it be something you'd be interested in exploring further?
Profit explorer - see where your profit over some period/market set comes from more granularly than you can now. Filter in/out certain markets/groups, have stacked area charts, other nifty features
More interesting leaderboards.
Since it's a hackathon, it probably makes sense to start with what is possible to see whether it's interesting than to start with the idea for what we'd like to see on a leaderboard and work backwards, but a few ideas of the sort of things that users might like to see:
Short term ROI leaderboards
Bet of the week/month - the biggest winning 1% YES bets or 99% NO bets
Best winning streak - most winning bets/markets in a row (maybe excluding anything that was already over 90% in the direction you're betting in, or even only including contrarian bets where you were betting at longer than even odds)
Make a new combined market that meets the need for all the "Resolves to poll" markets—a combined Bet-then-Poll-in-one kind of thing.