Welcome back
Here's your last autonomous run on Solana devnet.
Realized P&L
settled from the real score
+0.00USDC
—
—
Position—
Trustless proof-settle · oracle-pinned
Closing line Brier
0.193
vs 0.240 base · skill +0.195
Verifier
6 / 6
credential-free · public RPC
Kill-switch
Armed
feed-health + exposure guard
Sizing
¼-Kelly
0% ruin · 81% of full-Kelly growth
Autonomous bets
this run · quarter-Kelly#ActionMarket YES → fairEdgeAmount
The edge — real, and honestly bounded
Closing line is sharp
real data0.193 Brier vs 0.240 base · skill +0.195
The price EdgeBot bets toward is genuinely sharp — reliability per probability bin (dot area ∝ n).
Why N=20 can't be “significant”
powerN=20 has only 10.5% power, so a non-significant result is expected — not evidence of no edge. Confirming needs N≈200, the gated feed's ceiling, for everyone.
Why quarter-Kelly
sim · labelled¼-Kelly captures 81% of full-Kelly's log-growth with 0% ruin and ~40% less drawdown — the case for the shipped sizing.
Under the validated model, edge clears zero
sim · labelled+8.89pp mean CLV · 95% CI [+8.85, +8.93]
50,000 synthetic markets on a model fit to real data and validated against the real Brier (0.177 vs 0.193) before any conclusion — honestly labelled synthetic.
Verify it yourself — no trust required
Read-only · public devnet + TxLINE · no keys$ npm run judge:verify
Market is Settled on-chainon-chain statepass
On-chain pools match the runpool reconcilepass
Real TxLINE score re-fetchedindependent feedpass
Derived outcome = on-chain resolutionno typed winnerpass
CLV backtest reproducescommitted cachepass
6 of 6 pass — read from public RPC, the real TxLINE feed, and this repo.settled market ↗