Bellwether. PAPER

Dashboard

loading...

Open positions

1
$114.17 committed

Bankroll

$9,135.26
+82.71% from $5,000 start

Realized PnL

+$4,135.26
today +$0.00 · 258 closes

Win rate

42.6%
110W / 144L / 4 even

Decisions today

0
0 tradeable (conv ≥70)

AI spend today

$0.0000
avg conv - · top -
boss: loading...

Trading today

0 / 0
opens / closes

Risk gate today

0 / 0
allowed / blocked

Recent closes

paper PnL last 6 closes · all trades →
Closed Market Side Closed by PnL
Sun Jul 12 2:00am polymarket
Will Roberto Sánchez Palomino win the...
no no +$10.93
Sun Jul 12 2:00am polymarket
Strait of Hormuz traffic returns to n...
no no +$14.30
Sun Jul 5 2:00am polymarket
Iran agrees to end enrichment of ura...
yes no $-35.00
Sun Jul 5 2:00am polymarket
Iran agrees to end enrichment of ura...
no no +$9.87
Sun Jul 5 2:00am polymarket
Will the Iranian regime fall by June 30?
no no +$2.17
Sun Jul 5 2:00am polymarket
Will Spencer Pratt win the 2026 Los A...
no no +$10.45

Upcoming resolutions

open positions sorted by resolve date
Resolves Market Side Notional Max win
Mon Aug 17 5:00pm polymarket
Will Byron Donalds be the Republican ...
no $114.17 +$799.17 (8.00x)

Recent activity

last 6h - full feed →
loading...

Engine

...
checking

Database

...
MySQL connection

AI providers

Claude-
Gemini-
ChatGPT-

Markets ingested

1,353,307
1,348,178 K / 5,129 P

Last scan

loading...

Project timeline

major model + strategy changes
Fri May 22 Real stall fix: every Boss signal in 3 days was at fill <0.10 and getting blocked. Reopened deep-underdog (min_fill_price 0.10→0.05) with damage control: stake capped at 1% bankroll when fill <0.15 regardless of conviction tier
Wed May 20 Volume push: conviction floor 70→60; new experimental tier (60-69) at 1% bankroll; payout floor 1.5x→1.25x; slot capacity 50→64; slate cadence 30→20min - targeting 15-20 trades/day for data collection
Sat May 16 Major patch: duplicate-position gate (was bleeding $1,333 across 7 markets); skip fills < 0.10 (5 losses in a row); skip late-stage extreme markets (Warsh shape); slate conviction floor 70→60; slate sizing 1.5x; whale_copy notional 50→35
Wed May 13 Pause MM (silent $1,108 loss) + arb discovery (0 pairs found); activate 7 political multi-outcome slates (CA Gov, IA/ME/NM/FL primaries, Sweden PM, Colombia R1) for slate-style trade volume
Wed May 13 Crypto exclusion (KXBTC/KXETH/KXSOL + Polymarket bitcoin/ethereum subcategories); fat_pitch tier sizing downgraded to high (0.04) after 5 losses on deep-mispricing pattern; trade 115 closed manually. See docs/deep_review_2026-05-13.md
Tue May 12 Audit: exclude Kalshi CPI prints (3-for-3 losses) + Polymarket sports/entertainment (Eurovision/EPL/Wimbledon/etc); Eurovision slate deactivated. See docs/audit_2026-05-12.md
Sun May 10 Activity feed: hide skip-only decisions by default; toggle to show with slate batches collapsed
Sun May 10 Whale-copy Tier 1: 60min hold window + exit threshold $200 to $1000
Sun May 10 Loosened filters: 2x payout to 1.5x; stale data 10min to 30min; max positions 30 to 200 (cap was clamped)
Fri May 8 Resolution checker cadence 6h to 3h; panel login removed
Thu May 7 Kalshi 429 backoff; arb discovery 12h to 24h
Mon May 4 Loss-pattern filter (cut Kalshi quant-price markets); position size scales with bankroll
Mon May 4 Position cap scales with bankroll, not a fixed integer
Wed Apr 29 Min payout multiple gate: refuse fills worse than 2x risk-reward
Tue Apr 28 Polymarket resolution fix: closed=true filter; unblocked 251 stuck markets
Mon Apr 27 AI spend optimization (~14x reduction); whale-copy weight 20% to 40%
Sun Apr 26 Multi-provider Boss (Claude/Gemini/ChatGPT); runtime settings page
Sat Apr 25 Project started: Bellwether scaffold, initial schema, engine + panel