Point Parlay at any prediction market, perp, or game. It reads a web of live sources and returns one number you can trust, Buy, Sell, or Skip, with the fair price and every source it read. Most of the time it says skip. When there's a real edge, you'll know. And every call is public before it resolves, win or loss.
a web of live sources read · one calibrated verdict · every call published · we never touch your money
You're trading prediction markets and perps on vibes, a half-read headline, and a chart you glanced at. The "signal groups" selling you a record quietly delete their losses, so you're buying a highlight reel you can't verify. The hard part was never finding a confident call. It's telling a real edge from noise, honestly. That gap is the whole reason Parlay exists.
No way to tell if any of it is real.
Now you can check whether it's real.
One market, start to finish: the sources stream in, the gates tick through, and it lands on a verdict. This is the same pipeline behind every call in the terminal, including the one outcome most tools will never show you.
Parlay forward-tests every strategy in the open. Each call is paper-traded and timestamped before it resolves, then graded against the market. The winners and the losers both stay on the board, with their post-mortem. Nothing is quietly removed. That's how you know the wins are real.
See a strategy a creator is pushing on X or TikTok? Ask us to test it. We forward-test it here in the open and publish how it really holds up, win or loss.
Live, straight from the forward-test ledger. Paper-traded, small samples, labeled as such. Read the shape, not the headline: a low win rate with positive PnL is the tail signature, many small losses paid back by a few big wins.
This is what makes a verdict worth acting on, and it's the only honest proof of an AI track record: a reliability diagram, not a vanity "accuracy %". The dashed line is perfect calibration; the closer the points sit to it, the more the number means. Where they drift, we recalibrate.
This diagram is illustrative — the points and sample sizes are an explainer, not live data. Until a strategy's calibration holds and it beats the market on Brier it stays labeled unproven, including on the ledger above.
That's the read. Now here's the instrument.
A sample snapshot. Scan the board, open any market, and see the verdict plus the live information panel, the exact cited sources behind it. The track record is public; the live verdicts and edge board are what a subscription unlocks.
Read every market through one calibrated lens, all backed by the same cited sources. Here are the four you'll use most.
Hundreds of markets read in one pass, skips the efficient ones and flags the rare edge.
Independent models price each market separately, then the overconfidence is shrunk toward history.
News, on-chain flow, order books, weather models, odds and filings, all read into a single calibrated call.
Forward-test strategies run live, each with Kelly sizing, loss caps and kill-switches, plus a record you can see.
Those four are the daily drivers. Here's the rest of the desk — every tool reading from the same calibrated engine, one login.
Reads the whole board in one pass and flags the rare edge.
Funding, basis and momentum on perpetual futures.
Lines across books, de-juiced to a true price.
Equities and macro events read through the same lens.
Several models price each market, then shrunk to history.
The day's highest-conviction reads, ranked.
Cross-market and structural mispricings as they surface.
Every live forward-test position, timestamped in the open.
~20 sources read and cited behind every verdict.
Large flow and smart-money positioning, surfaced live.
What the market-movers are saying and doing.
Star markets and track them in one place.
Every call logged, win or loss, for honest review.
Get pinged when an edge crosses your threshold.
Learn to read markets the calibrated way — short lessons and drills on edges, vig, resolution rules and sizing, with your own calibration score that grows as you go. Free to start.
Start learning →Your personal reliability diagram, graded over time.
Track what you've mastered, lesson by lesson.
We never link to your exchange or venue accounts. No API keys requested, ever.
Your money stays on the venues, in your control. Parlay is research, education, and paper testing.
The bots here trade paper. You place your own trades, with your own judgment.
Parlay runs in the browser. The full dense terminal on desktop, your verdicts and live ledger in your pocket. Sign in and it's there.
Lock the low beta price now. It rises out of beta, once the record has proven itself. You're paying for the read, the cited sources, and an honest track record. A tool, not a tip service.
See the read and the full public record. Free, not a trial.
The everyday workhorse.
The full desk.
On most liquid markets, nothing does, which is exactly why our verdicts mostly say SKIP. The edge we're testing lives in niche, structural corners: weather models, funding mechanics, fine-print resolution rules. We don't claim to beat the market yet; the ledger above is the honest answer in progress, win or loss.
No. Research and education. We publish our testing so you can judge the tool, not follow it blindly. You place your own trades on the venues, with your own accounts.
Every call is logged and timestamped before the market resolves, so nothing can be added or edited after the fact. You can read each entry, the price we logged, and how it scored against the market.
Monthly, cancel anytime; access runs to the end of the period. Beta pricing stays locked as long as you stay subscribed.
Parlay is built in the open and improving constantly. Tell us what's missing. We appreciate all feedback and act on it.