The ARC
Playbook.

It's the probability that the manager on the other side would accept this trade, based on 47 features — KTC values, player age and trajectory, trade structure, and most importantly, how that specific manager has traded in the past. It's not just "is this fair." It's "will this particular person click accept."

Acceptance probability and negotiation likelihood are two different things. A trade can be unlikely to get accepted outright but still be worth sending because it will start a conversation. The Conversation Range tells you which zone you're in — Too Far Off, Long Shot, Conversation Starter, Likely Deal, or Strong Deal Zone.

Yes — and honestly this might be our favorite use case. While we're personally not fans of vetoes (as long as the trade is made in good faith, let people trade), we get it — some commishes need convincing. Just be warned — if the community votes to reject your trade too, that's useful information as well.

After you analyze any trade, you'll see a "Submit for Community Review" button on the result card. Click it, tell us whether you sent this offer or received it (this matters — received offers give us the best data on what actually gets rejected), and it goes into the voting queue.

That's exactly the kind of thing we track. If their profile shows they consistently undervalue picks, or can't figure out the strengths and weaknesses of their own roster, that's actionable information.

Connect your Sleeper account on the Vote Feed page. Once connected, we pull your league's trade history and build profiles for every manager you've shared a league with. The more leagues you connect, the more complete the profiles become.

The community vote is how we collect data on what real managers would actually do with a trade. Accept and Reject are straightforward. Counter is the interesting one — it means "I wouldn't accept this outright but I'd engage with it." That's valuable signal that the model doesn't get from accepted trade data alone.

Yes. Votes are weighted — not all votes count equally. Your weight increases when you: This filters out random clickers and makes sure the signal comes from people who actually play dynasty football.

Rejected trade data is the hardest thing for us to get. Sleeper only exposes accepted trades through their API — there's no endpoint for "trades that got ignored." Every rejected trade you submit helps us build a more accurate model on both sides of the equation.

No — it's completely optional. You can submit a trade with or without naming the manager you sent it to. Both are useful.

Player values move a lot over time. A trade that looked like a bad deal six months ago might look completely different today — and vice versa. Including a rough date (even just the month and year) lets us map the trade to the KTC values at the time it happened, rather than current values. That makes the data point dramatically more accurate for training the model.

Yes, and please do. Counter offers are a signal we're actively trying to collect more of. A counter means the trade type is in the right neighborhood — someone engaged with it — but the specific terms didn't land. That's a completely different signal from an outright rejection, and right now we have very little of it.