Ken: MIN_RESOLVED 100->50 (Steve, 2026-08-03) — let real trading start sooner on the fresh sim; keeps 2-winner consensus + independence denylist as the real safety. High-conviction winners (90% win) are well above noise at N=50.

29760751609cfd2096023402f9487d3ca6d2d364 · 2026-08-03 10:43:00 -0700 · steve@designerwallcoverings.com

Files touched

M kalshi-dash/server.js

Diff

commit 29760751609cfd2096023402f9487d3ca6d2d364
Author: steve@designerwallcoverings.com <steve@designerwallcoverings.com>
Date:   Mon Aug 3 10:43:00 2026 -0700

    Ken: MIN_RESOLVED 100->50 (Steve, 2026-08-03) — let real trading start sooner on the fresh sim; keeps 2-winner consensus + independence denylist as the real safety. High-conviction winners (90% win) are well above noise at N=50.
---
 kalshi-dash/server.js | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/kalshi-dash/server.js b/kalshi-dash/server.js
index 3766cd3..aa3edf9 100644
--- a/kalshi-dash/server.js
+++ b/kalshi-dash/server.js
@@ -7499,10 +7499,12 @@ async function getTraderState() {
 // currently hold the SAME direction. entry_price is refreshed to the LIVE Kalshi YES
 // price so the trader's yes/no price math is correct (paper stores NO-cost, not YES).
 async function getWinnerConsensusSignals() {
-  // MIN_RESOLVED raised 20→100 (2026-08-03, Cody gate): at 60% win gate, N=20 is clearable
-  // by pure variance — P(>=12 wins | n=20, p=0.5) ~= 25%, so ~1-in-4 RANDOM portfolios would
-  // qualify as a "winner" that real money then follows. N=100 pushes the gate above noise.
-  const MIN_RESOLVED = 100, WIN_GATE = 0.60, MIN_AGREE = 2, MAX_MARKETS = 8;
+  // MIN_RESOLVED: 20 was pure noise (P(>=12/20|0.5)~=25%); 100 is fully above noise. Steve set
+  // 50 (2026-08-03) to let real trading start sooner on the fresh sim while keeping the 2-winner
+  // consensus + independence denylist as the real safety. At N=50 a barely-60% portfolio is only
+  // ~10% likely by luck, but requiring TWO independent winners to AGREE makes the joint
+  // false-positive small — and high-conviction portfolios (e.g. 90% win) are nowhere near noise.
+  const MIN_RESOLVED = 50, WIN_GATE = 0.60, MIN_AGREE = 2, MAX_MARKETS = 8;
   // INDEPENDENCE GATE (2026-08-03, Cody gate): real bets follow "≥2 winners agree", but these
   // strategies carry NO independent signal — they accept on pure randomness or take everything,
   // so two of them co-holding a market is noise, not consensus. Exclude them from driving real

← b2420a9 Ken: migrate live buy order to Kalshi v2 create-order (POST · back to Ken · Ken: near-term + real-book filter (Steve) — scan now only tr b2c914b →