Big updates on Trifectly: forums, daily challenge, late re-analysis, new UI, and a bit about where this is heading
First, a sincere thank you to everyone who's been using the site and sending feedback. The DMs, the bug reports, the "have you thought about X" suggestions, all of it has shaped what's in this update. I read every message even when I'm slow to reply.
Here's what's new:
Forums. There's now a place on the site to talk about plays, share angles, argue about pace scenarios, post bad beats. Trying to keep it lightweight. No karma games, no algorithm, just a board. Not trying to replace this reddit sub- more Trifectly specific stuff. Really hoping this gets some traction.
Daily Challenge. A free contest that runs every day. The way it works: I release full analysis on one track every day, free to all users, no credits required. You make your selections, and whoever scores best on the day wins 1 credit added to their account. The top weekly score wins 2 credits. The point is to give people a reason to stay engaged on the slow racing days when cards are thin, and to let new users get a real look at what the analysis actually looks like without spending anything.
Late re-analysis. The analysis now gets re-run twice inside the 30 minutes before post. Late scratches, jockey changes, big tote moves, weather, all of it gets folded back in so what you're looking at five minutes before the gate isn't what was generated six hours ago.
Completely revised UI. Smoother, more intuitive, fewer taps to get to what you actually need. Tried to cut clutter without dumbing anything down.
Native iOS app in progress. Building it properly in Swift, not just wrapping the web app. Apple Pencil support is on the list. Being able to mark up a program the way you would on paper is something I've wanted forever.
A few things I want to be upfront about, because I think it matters.
This is a labor of love. I'm not trying to build an empire or chase some unicorn exit. I'm a stats, probabilities, and analysis geek and I genuinely enjoy doing this and improving the system. That's most of the reward for me. I try to be as transparent as I can about how things work, what the model is doing, where it's confident, and where it isn't.
To everyone who has purchased credits or a subscription, thank you. Genuinely. That support is what covers hosting, the proprietary ML models, the LLM costs, the data pipeline, and the time to keep building. Without it none of this exists.
I'm not trying to hand you finish orders or tell you what to bet. The goal is to surface every signal I can think of so you can fold them into your own play, whether you handicap Brohamer-style energy distribution, Quinn-style class and figures, Sartin pace, trip notes, sheets-style figures, or just sniff for real-time overlays at the windows. Different players use different angles, and the site is trying to feed all of them.
One thing people ask about a lot: the past performances aren't licensed from a data provider. I rebuild the dataset myself from publicly posted results. No external proprietary data goes into it. The upside of doing it the hard way is that it let me build a pretty deep research database on top, and all of it is accessible right inside the webapp. Track bias, seasonality, pedigree, jockey/trainer combos, individual horse profiles, all queryable without leaving the site or paying for another tool.
If you've been on the fence, give it a look. If you're already using it, the new stuff is live now. And keep the feedback coming. That's the loop that makes the site better.
Thanks again.
Mike