USDC in, USDC out
You bet in USDC and pay the gas in USDC too. No “go buy a gas token first” detour. Arc settles in stablecoins natively.
Macro prediction markets · priced by the framework
Two prices on every macro market. The crowd's, and the framework's.
Polymarket and Kalshi show one number, what bettors are paying. Pharos opens binary markets on CPI, payrolls, GDP and the Fed, and posts a second number beside it: an independent 12-pillar macro model's fair value, computed from official data, point-in-time, and written on-chain before anyone trades. The gap between the two is the product.
2 prices
on every market
12 pillars
2,500-series macro model
100% on-chain
settles on official data
Arc testnet
USDC in, USDC out
The spread, live
◆ Flagship market · live on Arc testneton-chain
US headline CPI trend for May 2026 prints >= 3.00% (3m/6m saar)?
Settles on the official BLS print · June 10
0255075100
the crowd
50.0%
the framework
83.2%
+33.2ppthe framework reads YES as underpriced. A whale can shove the crowd price. It cannot move the framework number. That one is computed from official data, point-in-time, and posted on-chain before anyone shows up.
Inflation Heat (PCI)calibrated on 241 printsIC +0.30point-in-time 2026-05-22framework price on-chain
View this market on Arcscan ↗
Live on Arc testnet. The framework price and the crowd price are both written on the contract.
A price nobody can buy
The crowd price moves with money. The framework price moves with data. A seven-figure bet can push the first one. It cannot push the second. A $500 account and a $500M book read the same number.
It shows its receipts
Every market names the exact pillar pricing it, how many prints it was calibrated on, and how strong the edge is. No demonstrated edge, no faked conviction. It shrinks to fifty and says so.
The scoreboard is on-chain
Every market resolves on the official BLS, BEA or Fed print, written to chain by an agent, permanently, with no hand on the scale. Trust me becomes check it yourself.
The board
Graded by how much it trusts itself
We only claim conviction where the model beats a naive baseline out of sample. Everywhere else it says so, on the card.
GDPCALIBRATED
US real GDP for Q1 2026 (2nd) prints >= 2.00% (saar)?
crowd 50.0%
model 55.7%
+5.7ppmodel: YES underpriced
Activity Pulse + Capex Thrustn 104IC +0.29
View this market on Arcscan ↗
Live on Arc testnet. The framework price and the crowd price are both written on the contract.
point-in-time 2026-05-22resolves May 28framework on-chain
PAYROLLSABSTAINS
US nonfarm payrolls for May 2026 print >= 125k?
crowd 50.0%
model 46.1%
model 46.1%framework abstains here
The in-sample fit looks strong but does not survive out of sample. The framework declines to post conviction here.
View this market on Arcscan ↗
Live on Arc testnet. The framework price and the crowd price are both written on the contract.
point-in-time 2026-05-22resolves Jun 5framework on-chain
FOMCMODELED
Does the FOMC cut the target by >= 25bps at the Jun 2026 meeting?
crowd 50.0%
model 26.6%
model 26.6%modeled, not calibrated
A transparent rate reaction function, not a calibrated market. Shown for completeness.
View this market on Arcscan ↗
Live on Arc testnet. The framework price and the crowd price are both written on the contract.
point-in-time 2026-05-22resolves Jun 17framework on-chain
The moat
How the second number is made
Anyone can list a CPI market. Pricing it with a real macro model, honestly, is the part that is hard to copy.
01 / point-in-time
No peeking
Only data the world had already seen on the day the market opened. The quote is an information state, not a revised final value.
02 / consistent trend
Signal, not noise
We read annualized trends across months, not one jumpy print, so a single noisy release does not whipsaw the fair value.
03 / calibrate
History sets the tilt
The pillar-to-outcome relationship is fit on aligned history, and we report how well it has actually predicted this print.
04 / shrink
It can say I don't know
When a pillar has no proven out-of-sample edge, the model pulls toward 50% and tells you. Honesty is the feature.
Why Arc
Built for the way macro money actually moves
Your size stays private
Arc's encrypted mempool means a large macro position does not broadcast its intent before it fills. The institutional wedge.
Settles on the print
When the number lands, an agent resolves the market on the official figure and pays winners on-chain. No bridge, no wait.
Why we built it. Every market is a live, adversarial, out-of-sample test of our framework, funded by other people's conviction. The crowd is the red team. Every miss feeds the next version. This is not a casino. It is a research process, graded in public, in real time, by the economy itself.