HivePredict has had a proper visual overhaul. Same platform, same on-chain guarantees, but the interface is now much better at showing you what the crowd actually thinks.
If you have not tried it: you predict outcomes on real-world events (crypto prices, sport, Hive itself), stakes pool together, and winners split the losing pool proportionally. Log in with Keychain or HiveSigner, no signup, no email.
Red means one thing now
The old brand colour was red. The NO side of every market was also red. So the strongest colour on the screen was carrying two meanings that contradict each other, and nothing on the page was clearly HivePredict rather than clearly a losing bet.
The brand is now honey amber. Red is reserved for NO and nothing else, which meant pulling it out of the status labels and the category tags too. Voided markets are grey now, not red.
Auto markets
The platform will now automatically create markets to predict in (with more on the way). The platform itself now will stake on both sides of the default markets, so there are many varied markets to predict within including some Aussie sports markets, crypto price markets and so on.
Creator commissions
A market creator earns 1% of their market's final pool if they act as a two-sided maker. To qualify, at resolution they must still hold an open position on every outcome, each at or above the minimum opening bet, with the smallest side worth at least half the largest.
Seeding both sides costs the creator money: whichever way the market resolves, the losing side of their own seed is gone. The commission is what pays for that, and it rises with the volume the market attracts. Holding every outcome also means the creator has nothing to gain from the market landing one way rather than another, which matters on manual markets where the creator submits the resolution proof.
Cashing out of an outcome before resolution forfeits the commission for that market.
Fee caps
Fees on the platform are capped at a total of 3%. The 3% total is capped at the losing-side pool. This ensures that winners always receive at least their original stake back, even when one side dominates the pool. Creator commission is taken from within that cap, so it can never reduce a winner's payout.
One bar instead of two numbers
Cards used to stack YES and NO as two separate pills. To work out which way a market was leaning you had to read two numbers and compare them in your head. On every card. On a board of twenty markets that is forty numbers to get through before you find anything you want to bet on.
There is now a single split bar per card. One look tells you the direction and roughly the margin. The footer underneath carries the pool size, how many people are in, and the countdown, so you can tell a 9 HIVE market with four predictors apart from a 2 HIVE market that only the house has touched.
Three rows of filters became one
Browse and Watching, then nine categories, then New / Ending / Volume, then Open / Awaiting / Settled, then Core / HE / All. Five groups of controls, three rows deep, every one of them a red pill competing with the actual markets underneath.
Sort and status are compact dropdowns now, categories are one scrollable row, and only the selected category is highlighted. Controls should get out of the way of the thing you came for.
The odds chart was running to 120 per cent
The odds history chart was auto-scaling its own axis, so a chart of a probability was cheerfully drawing tick marks above 100 per cent.
It is pinned to a fixed 0 to 100 per cent scale now, with a dashed line at even odds so you can see immediately which side a market has crossed to.
Colour is not the only signal any more
Green for YES and red for NO is the convention in prediction markets, and it is also close to the worst possible pair for the most common kind of colour blindness. Roughly one in twelve men cannot reliably separate them.
I kept green and red, because breaking a convention has its own cost, but colour is no longer the only thing doing the work:
- YES is always first. Left on the bar, left in the labels, left in the bet panel. Position carries the meaning even if the hue does not.
- Every percentage has a Yes or No label sitting next to it. No bare coloured numbers anywhere.
- A small gap splits the two halves of the bar, so the boundary is visible without relying on colour at all.
Everything else
- Predictor counts on market cards, so you can see how many people are actually in a pool rather than guessing from the size of it.
- Loading skeletons that match the shape of a real card, so the board stops jumping around as data lands.
- Proper empty states when a filter turns up nothing, with a way back out instead of one grey line of text.
- Categories are monochrome now. Nine categories is more than any colour palette can keep apart, so the colour budget goes to status and outcome where it means something.
- Actual type hierarchy. Page titles, card titles and metadata used to sit within a couple of pixels of each other, which is why scanning the board felt like reading a spreadsheet.
What is next
Category counts on the filter row are the obvious gap. I left them off rather than show you a number that ignores whichever other filters you have set, which is worse than no number at all.
Everything from the last update still stands: Spanish support, a referral program, a suggest-a-market flow, and pushing the community resolution system further.
Have a look at hivepredict.app and tell me what you think.
