POLYBOT — Trading Console

Band	Side	Contracts	Avg cost	Mark	MTM
— no open positions —

Band	PM (final)	Status
— loading —

Singapore —

Weather

System

R0 · Session State —

Mode

—

phase —

Settle ETA

—

Daily P&L

$—

realized — · unr —

Forecaster Status

—

R1 · Today's High

Floor —

—

METAR running — · anticipated —

peak —

Next METAR —

—

last —°C · —

—

Phase-E Forecast (+30m) —

—

σ — · ensemble —

R2 · Live Observations

S24 · Changi Met —

—°C

5m — · 15m —

—

UVI · NEA Changi —

—

Δ15m — · peak —

—

METAR · WSSS Changi —

—°C

dewp — · spread — · wind —

—

R3 · Multi-Step Joint Forecast —

+— min

—

+— min

—

+— min

—

+— min

—

R4 · Floor Distribution & Daily-High Bands click band row → forecast trajectory

Floor Distribution

no data

—

Daily-High Bands

Band	Model	PM	Edge	P(NO)	YES	NO
no band data

R5 · Spatial & Forecast Consensus

Singapore · Weather Stations

— —kt — · —

NEA Radar · Changi —

— mm/hr

vicinity — mm/hr · +15m — · +30m —

motion —

Forecast High · Consensus —

—°C

WU — · NEA —

OWM — · OpenM — · WA —

—

NEA 2hr · Changi —

—

Upstream of S24 —

computing…

R7 · OWM Path to Settle · hourly temp → 16:00 SGT OWM · —

Hourly temp vs settlement bands —

━ owm fcst ┄ owm reading ━ s24 actual ━ metar actual ━ wu fcst ▮ pop ▮ rain likely ● now — integer °C = band edges

Projected Peak—

—°

band — · at —

vs daily max —

Rain Threat to Peak—

—%

—

Hours to Settle

—

settle 16:00 SGT

R7c · Recent METAR observations · WSSS Changi —

Last few raw reports, newest first. A SPECI is an off-schedule special report — issued when conditions change between the routine half-hourly METARs (e.g. a temperature jump or rain onset).

R7b · Rain Nowcast (60 min) · R8 · OWM vs Model peak band

Precip outlook—

—

Minute precipitation (mm/h)

OWM peak band

—

vs Modelmodel —

—

R9 · 8-Day Outlook · today + day-ahead highlighted

Today: —

Tomorrow: —

R10 · External Conditions (OWM) —

External Temp—

—°

feels —° · dewp —°

OWM grid · not Changi METAR

Humidity / Dewpoint

—%

dewp —° · spread —°

Pressure

— hPa

vis —

Cloud / Wind—

—%

wind — · gust —

—

Monitor Log tail · live

—loading log tail…

Executor Logtail · live

—loading log tail…

Process Tree —

File Freshness —

Portfolio Risk · cross-book —

Terminal Commands · cheat sheet click to copy

Setup

cd /Users/JonathanMitchell/PolymarketBot first — get into bot directory

Bot lifecycle

./bot start scheduler + dashboard (preflight first; pass --no-dashboard for headless)

./bot stop kill EVERYTHING (scheduler / monitor / executor / observer / dashboard)

./bot status processes alive? + last 20 log lines

./bot tail follow scheduler.log live

./bot test pytest preflight only

During session

./bot pnl P&L breakdown by tag (mm / flip / cert / signal / lottery)

./bot mm tail today's executor MM activity (filtered)

Reports

./bot report auto-pick (morning before noon, EOD after) — same as 06:00 + 18:00 cron

./bot report morning overnight observer + rolling calibration

./bot report eod fires + late-day stability + conviction + P&L

Dashboard

./bot dashboard launch on http://127.0.0.1:8765/

./bot dashboard 9000 custom port

pkill -f dashboard/server.py kill dashboard only (leave bot running)

Diagnostics · on-demand

python3 tools/calibrate_hawkes.py MLE fit fill arrival rate · run after ~2 weeks of MM fills

python3 tools/calibration_rolling.py --window 14 --out logs/calibration_rolling.json rolling Brier vs PM · auto-runs at EOD; this manual form refreshes the dashboard cache

python3 tools/calibrate_pws_bias.py --window-days 14 --out logs/pws_bias_schedule.json refit Hougang→Changi bias schedule · auto-runs at EOD; manual run picked up next monitor cycle

python3 tools/calibration.py single-day calibration · ad hoc

python3 tools/analyze_overnight.py --window 7 overnight observer stats · already in morning report

python3 tools/audit_late_day_stability.py late-day thrash detector · already in EOD

python3 tools/audit_conviction_manager.py cancel/reprice audit · already in EOD

python3 tools/verify_fixes_fired.py how many times each safety guard fired today · in EOD

python3 tools/verify_smoothing.py --hour 12 --temp 32 Dirichlet / Gaussian smoothing diagnostic · ad hoc

python3 tools/ruin_bounds.py --bankroll 500 --days 30 Monte Carlo P(bankroll<0) · run after changing risk caps

python3 tools/smoke_multi_city.py CityConfig contract validator · only when adding a city

On-chain

python3 redeem_positions.py --dry-run test redeem without on-chain submit

python3 redeem_positions.py live redeem · auto-runs daily at 09:00 SGT in scheduler

EMP-RV · PM vs M1 Settle (BUY-NO, both tails of PM peak) —

PM PEAK

—

MODAL (model)

HOUR · RUN-HIGH

—

MAX LOSS @ SETTLE

—

RESTING (orders)

—

FIRES · SKIPS

—

NET P&L

—

Per-band desk · BUY-NO when PM over-prices our M1 settle prob hover a row for the live order book book — —

Band	Δ	NO bid ×sz	NO ask ×sz	depth $	PM	p̂ ref	p_wil (n)	v1m	M1 ●	E·mid	E·ask	NO pos / tgt	resting	last
no live bands

Settlement payout matrix — total P&L net of fees if day settles here EV —

Settle band	V1M	M1	PM	P&L @ settle
no positions

Fills & resting orders · emp_rv_no —

Time	Band	Side	Price	Size	Status	P&L
no fills or orders

Realized—

Open (mark)—

Settle EV—

Fees paid—

NET—