Bot Claude — UFC 328
UFC Fight Night: Sterling vs Zalal
DRY-RUN · $0 CAPITAL
DRY-RUN MODE. Bot is paper-trading only. Zero capital at risk. Notional $1,000 bankroll for sizing math only. Live execution is gated on post-card review by QC + user signoff. All "trades" below are dry-run entries — they do not place real orders on Polymarket.
What Is This FightIQ Bot · v1

An autonomous Claude-driven trading agent that watches the FightIQ model output, the Polymarket order book and our internal QC gates, then sizes Kelly-adjusted bets on UFC fights when a curated set of confirmed-edge conditions all align at once. It runs on a continuous cycle in the hours leading up to each card, refusing to place anything that doesn't pass every gate.

The exact strategy logic — which signals fire, which corroborations are required, the bands, the holdout rules, the post-loss cooldowns — stays in-house. What you see on this page is the audit layer: every poll, every eligibility decision, every dry-run entry, with the gates it passed and the ones it didn't. Full glass box on the bookkeeping. Closed box on the edge.

This page exists because we're running it openly through Vegas 116, Perth, and now UFC 328 as a public dry-run. After clean execution and QC signoff, live capital comes online — initially internal, then opening to a small allocation tier.

◆ Coming Soon
Bot Signals · Subscribers Only

Live trade alerts the moment the bot fires. Per-fight eligibility reasoning. Strategy-specific calibration data. Pre-card briefing emails. Post-card autopsy with what fired, what didn't, and why.

Join the Waitlist →
Backtest Evidence walk-forward 2025-06 to 2026-01
Win Rate (BACKTEST)
—
walk-forward backtest · not live results · 88 of 93 dry-run trades
Total P&L
—
flat $200 / bet · no compounding
Trades
—
avg ROI per trade
● 2025-06-14 → 2026-01-24 walk-forward Polymarket Gamma · liquid markets ≥ $5k volume

Cumulative P&L

— points
Strategy ledger summarises the bot's S1 (model edge), S4 (line moves), S5 (favourite-band) and S6 (distance/method) signal families with hit rate, P&L and trade count per family. Updates live from the bot evidence bundle.
Important Caveats
  • Caveats include: walk-forward window starts 2025-06-14, not 2025-01; market liquidity ≥ $5,000 per fight; flat $200/bet sizing (no compounding); BFO closing lines used for ROI. Full methodology in the linked Methodology bundle.
Filter
Date Fight Strat Side Entry Result P&L Cumulative
Trade ledger renders the 93 walk-forward backtest trades — date, fight, strategy family, side, entry price, win/loss, P&L and cumulative bankroll. Sortable; filterable by strategy family or losses-only.
Bundle published — Methodology ↗  ·  Raw CSV ↗  ·  Manifest ↗
Halt
—
Preflight
—
Cycle
—
Late Window
—
T-minus
—
Bankroll
Notional
$1,000
dry-run sizing math only
Today PnL
$0
resolves on settlement
Total Staked
—
open dry-run positions
Open Positions
—
unsettled trades
Per-Event Used
—
cap 20% / $200
Per-Fight Evaluation
F# Fighters Mkt % Model % Edge Bet Side / Kind Kelly $ Liq $ Status
Per-fight evaluation table shows the bot's market %, model %, model edge, recommended side and kind, Kelly stake sizing in dollars, available market liquidity, and bet status for each fight on the upcoming card.
Dry-Run Trade Ledger
Dry-run trade ledger logs every position the bot opens — entry timestamp, fight, side, price, stake, current value and settlement. Dry-run means no real money; the ledger tracks notional P&L only.
Calibration Cycles
Calibration cycles record each model recalibration run — date, dataset window, accuracy delta, calibration error change. Confirms the bot's underlying model stays sharp against fresh data.
Bot State _bot/state.json
Bot state KV mirrors the internal state.json — halt flag, preflight status, last cycle timestamp, late-window state, hours-to-fight countdown.
Audit Trail last 20 [bot] CHANGELOG lines
Audit feed will surface here once exposed. Internal CHANGELOG is reviewed for every bot release.