TV2 // POLYBOT TERMINAL

▸ SYSTEM STATUS Current run stats only

🤖 ALL BOTS auto-discovered — new bots & strategies appear here automatically

Loading…

MAIN BOT

PAPER

BALANCE

—

TOTAL P&L (all-time, mixed fill regime)

—

ROI —

LEGACY PAPER P&L (EXECUTION-CONTAMINATED)

—

— trades

WIN RATE

—

— W / — L

TRADES

—

avg $— /trade

TODAY — —

GEORGE

PAPER · CL-ANCHOR

BALANCE

—

TOTAL P&L

—

all-time

WIN RATE

—

— W / — L

TRADES

—

— open

BORG

SHARED $500 · PILOT

SHARED BALANCE

—

P&L 1×

—

capital + liquidity constrained

WIN RATE

—

— W / — L

ADMITTED

—

— pending

Shared Research P&L

—

one $500 account · pilot only

Main Today (Diagnostic)

—

last 24 h · main bot

Shared Win Rate

—

— W / — L

Admitted Orders

—

— · — · —

Skip Rate

— / 85% target

On Target

Avg EV_adj

— avg

On Target

Gate 1 Trigger Rate

— rate

Active

Avg Trade Size

—

Sizing

Strategy GBM SIGNAL ENGINE

MARKET

Polymarket BTC 5-min binary

Up/Down markets · USDC · Polygon

SIGNAL SOURCE

Binance BTC/USDT WebSocket — price momentum + order book imbalance
BTC Δ threshold: — / 30s window

SIZING

Kelly criterion · manual · cap — · max $— · max daily loss $—

GATE PIPELINE

Pre-filter Stale lag >20s · No-chase · EV decay

Gate 1 Microstructure confidence ≥ —

Gate 2 EV_adj ≥ —% floor · spread cap 3%

Gate 3 BTC momentum direction · min Δ —%

▶ TRADE Entry at best ask/bid · Kelly size · TP/SL managed

Gate Performance — 24h

Pre-Filters

A: Stale lag—

B: Chase—

C: EV decay—

Avg lag—

Gate 1: Micro

Trigger rate—

Fails (G1+1.5)—

Last seen—

Gate 2: EV+Cost

EV_adj avg—

Cost avg—

Pass rate—

Fails (G2/2.5)—

Gate 3: BTC Mom

Trades executed—

Trade rate—

Fails (G3)—

Avg spread—

Recent Signals (Last 10)

TimeDirEV_adjVerdictReason

No signals yet

Feed Status

Binance: Waiting...

Chainlink: Waiting...

Polymarket: Loading...

Wallet: Not configured

USDC: —

Claude AI: Not configured

▸ TRADES & SIGNALS

Revenue (Wins)

—

— trades

Costs (Losses)

—

— trades

Net P&L

—

— trades

Win Rate

—

— /trade

Advanced Metrics ▼

Sharpe Ratio

—

Max Drawdown

—

Expectancy

—

Profit Factor

—

Win Streak

—

Loss Streak

—

Avg Win

—

Avg Loss

—

This Session

Decision Stream

Time	Dir	Filter	Lag	EV adj	Spread	Verdict	Reason
Loading…

Decisions (24h)

—

Skip Rate

—

Avg EV_adj

—

Avg Lag Age

—

Avg Spread

—

Last Signal

—

Pre-Filters A/B/C

A: Lag stale (>20s)—

B: Chase (>8%)—

C: EV decaying—

Avg Lag Age—

Gate 1: Microstructure

Confidence—

Pass Rate—

Fails (G1 + G1.5)—

Last detected—

Gate 2: EV + Cost

EV_adjusted avg—

Pass Rate—

Fails (G2/2.5)—

Gate 3: BTC Momentum

EMA pass rate—

Fails (G3)—

Trades executed—

Trade rate—

Skip Breakdown (24h)

Loading…

Live Market Feed ● DISCONNECTED

BTC Price

—

BTC Δ 30s

—

YES Price

—

NO Price

—

EV_adj

—

EV History

OB Imbal

—

Flips

—

Balance

—

Decision Stream

TimeDirFilterLagEV_adjSpreadVerdictReason

No signals yet

Live on-chain trades from your Polymarket wallet

TimeSideOutcomeMarketPriceUSDC

Click Refresh to load your Polymarket activity

🔬 RETROSPECTIVE SCENARIO LAB — NOT A CAUSAL BACKTEST

Strategy Parameters

Gate 1 Threshold (Conf)

Gate 2 EV Floor (%)

Kelly Cap (%)

Max Trade Size ($)

Initial Balance ($)

Days History

Simulated Performance

TOTAL PnL

—

WIN RATE

—

MAX DRAWDOWN

—

TRADES

—

Run scenario to view results

🧪 GEORGE — SPLIT TEST (CHAINLINK MODEL, PAPER ONLY)

George Balance

—

stake $— flat

Total P&L

—

vs main bot: see Trades tab

Win Rate

—

— W / — L

Trades

—

— open

σ (EWMA 5m)

—

— CL rounds seen

GEORGE SETTINGS

Trade Size ($)

Paper Balance ($)

Min Edge

Cost Buffer

CL Deviation %

Min Sigma (5m)

Max Sigma (5m)

Max Remaining (s)

Min Remaining (s)

Max Daily Loss ($)

ORACLE MODEL STATEresolution print vs spot

Chainlink now: —

Binance spot: —

Divergence: —

CL(open) anchor: —

p(UP): —

p(oracle updates): —

Last decision: —

GEORGE TRADESflat stake, hold to resolution

#	Time	Dir	Entry	Exit	P&L	p(UP)	pUpd	CL open	CL entry	Spot	Reason
No George trades yet

GEORGE LOG

Start George to see the decision log.

🛰️ BORG — SHADOW EDGE DISCOVERY + H53 LIVE OVERRIDE

Book Feed

—

tape — · takers —

Binance Feed

—

chainlink —

Markets

—

— live-captured opens

Q3 Disagreements

—

outcome vs Binance sign

Snapshot Gaps 24h

—

uptime —

Shared $500 Simulator

—

one account · capital + depth competition

Shared P&L (1×)

—

after pessimistic costs

Shadow Fills

—

— scored · — pending

Win Rate

—

— W / — L

Last Scored

—

auto-scores every 5 min

H53 — ACCIDENTAL 5M NEAR-EVEN FAVORITE · LIVE OVERRIDEUNPROVEN · $10 MAX · EXACT FAK LIMIT · ONE/MARKET

Mode

—

heartbeat —

Collateral

—

authenticated wallet

Live decisions

—

— skipped

Matched

—

ack —

Realized P&L

—

— unresolved

Time	Token	Signal	Requested	Status	Matched	Fill	Ack	Outcome	Realized
Waiting for H53 live decisions

SHADOW SCOREBOARDback-of-queue fills, current crypto taker-fee curve at 0.5×/1×/2× stress — pilot rows are machinery tuning, not evidence (EVAL_PROTOCOL §3)

Strategy	Phase	Places	Cancels	Scored	Fills	Fill %	PnL gross	PnL 0.5×	PnL 1×	PnL 2×	Quality A/B	Fidelity A/B	Last
No shadow orders yet

RECENT SHADOW ORDERSwhat BORG would have done — scored after resolution

Time	Strategy	Action	Side	Px	Size	TTE	Queue	Filled	Quality	Fidelity	PnL 1×	Note
No shadow orders yet

BORG LOGcollector + shadow events

Waiting for events…

🌊 FLOW LAB — PUBLIC TAKER-SWEEP RESEARCH + CAPPED CANARY

Global Trades (run)

—

Realtime Panel

—

last print —

Independent Sweeps

—

goal ≥300 + 30 days

Paper Fills

—

— scored

Mode

PILOT

$500 research · $10 target

FINAL-10S ABSORPTION — FORWARD V3 + EXECUTION CANARYunproven · waiting for live causal intents

Time	Market	TTE	Target	Arrival Ask	Size	Notional	Min	Intent	Order	Ack	Reason
No V3 boundary intents yet

LATENCY × ARM SCOREBOARDcontinuation vs fade control · exact entry + exit taker fees · rows overlap, never sum as a portfolio

Arm	Latency	Sweeps	Scored	Fills	Fill %	Win 5s	PnL 1s	PnL 2s	PnL 5s	PnL 10s	Mean 5s	Quality A/B
Collecting prospective public-flow observations…

RECENT PAPER SIGNALSa signal begins only after a completed trade is publicly observable

Time	Market	Trigger	Notional	Arm	Latency	Target	Limit	Fill	PnL 5s	Quality
No eligible sweeps yet

What this tests: whether a publicly completed taker sweep has a short-lived continuation markout after realistic latency, displayed liquidity and both taker fees. The fade arm is the control. The global scanner sees completed trades across Polymarket; sub-second scoring is restricted to the bounded realtime panel because public Data API timestamps are second-resolution. Nothing here observes a pending user order, guarantees exit liquidity, or submits an order.

📚 BOOK LAB — ALL-MARKET L2 + STRUCTURAL ARBITRAGE (PAPER ONLY)

Universe

—

— selected

Intents 1h

—

— fills

Reaction

—

target 20–50 ms internal

Structural

—

economic leads · non-atomic

Mode

PAPER

$500 cohort · no DB hot path

L2 / PASSIVE-MAKER SCOREBOARDfull-depth fills + fees · latency arms overlap and must never be summed

Strategy	Arm	Intents	Fills	Markets	Win 5s	PnL 1s	PnL 5s	PnL 30s	Mean 5s	Toxicity/share	Quality A/B
Collecting forward observations…

PAIRED COMPLETE-SET MAKERchecking dedicated paper process…

Version / arm	Cycles	Filled	Merged	Markets	Merged shares	Locked PnL	Orphan PnL	Realized PnL	Modeled reward*	Actual + model*	Open orphans	Unscored
Collecting queue-realistic complementary fills…

RECENT PAIRED CYCLES$1 merge payout · unmatched shares remain directional risk

Time	Market	Arm	Status	Initial pair	Leg fills	Merged	Locked	Orphan	Total	Modeled reward*	Quality
No paired cycles yet

STRUCTURAL PAYOFF IDENTITIES“economic” is not locked arbitrage; atomic-qualified is the hard standard

Structure	Latency	Evaluations	Economic leads	Atomic qualified	Max displayed 2× profit	Latest
No structural evaluations yet

RECENT SIMULATED ORDER INTENTSpost-only queue model, partial fills, and executable bid-depth markouts

Time	Market	Strategy	Action	Price × Size	Queue	Status	Fill	PnL 5s	Quality
No intents yet

Evidence contract: thresholds are mechanism defaults frozen before this cohort. *Modeled reward uses public aggregate L2 and is not earned, claimed, or admissible for promotion; maker rebates remain zero until account-level reconciliation exists. Require at least 300 independent markets, 30 calendar days, both time halves positive, 2×-cost robustness and a market-clustered lower confidence bound above zero. A zero or negative edge is an accepted result.

⇄ CROSS-VENUE LAB — POLYMARKET × KALSHI (LIVE DATA · PAPER ONLY)

Polymarket

—

active binary markets

Kalshi

—

active binary markets

Pending / Approved

—

— reviewed rejected · — monitored

Forward Relation Events

—

— ever economic · — disappeared

Capital Model

$250 + $250

prefunded · $500 total

EXECUTABLE TWO-LEG SCOREBOARDtyped payoff proof · bankroll/depth optimized · fees + one tick/leg stress

Direction	Observations	Pairs	Proved episodes	Lockable	Max parity control	Max proved stress	Max control ROI	Max proved ROI	Max qty	Mean control	Latest
Collecting synchronized books…

CAPITAL RELEASE — EXECUTABLE CONVERGENCEfour fees · right-censored · 30-day horizon

Cohort	Episodes	Pairs	Profitable exits	Censored	Observed median	P ≤1h	P ≤6h	P ≤24h	P ≤7d
Collecting fixed-size entry and liquidation marks…

CONTRACT-RELATION REVIEW QUEUEequivalence, implication, exclusion and exhaustiveness are distinct proofs

Polymarket	Kalshi	Score	Status	Decision	Time Δ	Audit flags	Monitored
Discovering candidate pairs…

FORWARD RELATION-EVENT LEDGERone row per approved relation + state activation + safe bundle; quote flicker is not sample size

First seen	Relation	Direction	Lifecycle	Quotes	Economic	Disappear	Max qty	Best 2× stress	Worst orphan stress	Quality
Waiting for a forward approved relation event…

RECENT PAPER OPPORTUNITIESunproved parity is a control; only a deterministic payoff relation can become economic

Time	Pair	Direction	Qty	Poly	Kalshi	Capital	Raw lock	Raw ROI	Stress ROI	Orphan abort	Status	Quality
No synchronized evaluations yet

How to read this: exact identities support opposite-outcome bundles, but non-identical contracts can also support a proved implication. If event A implies event B, NO(A) + YES(B) pays at least $1 in every allowed terminal state. The engine now enumerates those states and rejects any bundle whose minimum payout is zero. A state-conditioned relation activates only after timestamped evidence satisfies every frozen gate. Convergence does not create terminal edge; it can release capital early. Cross-venue orders are never atomic. Current Kalshi transport may be public REST, so quality can be capped at B; authenticated WebSocket capture is preferred. The Dublin host is research-only for Kalshi because of current geographic restrictions.

⚡ PYTH RESOLVER BOUNDARY — FORWARD PAPER MEASUREMENT

Certified markets

—

— symbols

Signals 24h

—

— independent markets

Arrivals

—

20 / 50 / 100 / 250 / 500ms

Resolved

—

terminal-scored markets

Probe

$10 PAPER

depth + fees + $0.10 settlement stress

PRE-REGISTERED ARM SCOREBOARDoverlapping counterfactuals — PnL columns are not additive

Trigger	T−sec	Side	Latency	Arrivals	Markets	Executable	Mean entry	5s scored	Mean 5s	Terminal n	Terminal win	Mean terminal
Fresh forward cohort has not reached a checkpoint yet.

RECENT RESOLVER SIGNALSlive Pyth source clock; carried-forward and snapshots excluded

Time	Market	Symbol	Trigger	Side	Resolver / boundary	Distance	TTE	Executable arms
Waiting for a live checkpoint or sign flip…

Evidence contract: this does not assume that the current resolver side predicts the close. It measures whether boundary transfer survives executable asks, current fees, exit depth, 20–500ms arrival delays and terminal settlement cost. Require 300 independent markets, 14 calendar days, both chronological halves positive under doubled costs, Holm correction across arms and a market/day-clustered lower confidence bound above zero. Zero edge is an accepted result.

▸ CLAUDE AI — STRATEGY ANALYSIS

Analysis Control

Trades Analyzed

0

Next Trigger

After 50 more trades

Auto-Analysis

OFF

Latest Analysis

No analyses yet. Click "Analyze Now" to get started.

Date	Trade Count	P&L	Status	Action
No analysis history yet

⚗ LAB & SETTINGS

TOTAL PnL

—

WIN RATE

—

PROFIT FACTOR

—

EXPECTANCY

—

SHARPE

—

MAX DRAWDOWN

—

TRADES

—

AVG PnL/TRADE

—

per closed trade

Cumulative P&L Curve

No closed trades yet

Strategy Parameters

EV Floor (Gate 2)1.5%

Lower = more trades, higher = fewer but cleaner

BTC Momentum Threshold0.015%

Min 30s BTC move to trigger directional signal

Kelly Cap10%

Max % of balance per trade

Signal Interval8s

Max Trade Size (USD)$10

Gate 3 (BTC direction)

Hitting APPLY saves settings and restarts the bot

By Direction

No data

By Exit Reason

No data

By EV at Entry

No data

Performance by Hour (UTC)

Open Positions

No open positions

Recent Closed Trades

#	Time	Dir	Entry	Exit	Size	PnL	EV	Exit Reason

No closed trades in selected period

Gate Thresholds

Gate 1: Microstructure Confidence

0.45

Informational only — Gate 2 EV is the real filter

Gate 2: EV Hard Floor

0.5%

Minimum EV% required to trade (0.2–1.5% recommended) — lower = more trades

Gate 3: BTC Momentum Direction Confirmation

Min Kelly Edge (trade entry threshold)

5%

Minimum Kelly edge (modelProb − marketProb) required to enter

Gate 3: Min BTC Δ (momentum threshold)

0.010%

Min 30s BTC % move for Gate 3 direction confirmation (0.005–0.02% recommended)

Min BTC Flat Filter (btcFlat threshold)

0.005%

Skip if BTC is completely flat below this threshold (unless Gamma displaced)

Signal Interval

8s

Seconds between signal evaluations (5–15s recommended)

Order Execution

Order Timeout

60s

Max time to hold a resting limit order before cancellation

Adverse Selection Tolerance (ticks)

8

Cancel BUY if market drops this many ticks below limit price while order rests

Orders rest as GTC limits at lastTradePrice + 1 tick. Paper orders use stochastic fill simulation — fills are NOT guaranteed.

Flip & EV Exit

Flip Threshold (%)

5%

Min EV gain required to flip an open position. 5% filters BTC noise (~3-4% EV swings).

EV Decay Exit Ratio

2.0

EV velocity floor multiplier (0.5–2.0 recommended) — higher = less blocking of oscillating signals

Opening Window (seconds after market open)

100s

Trade in the first N seconds after market opens — fresh mispricing window

Closing Window (seconds before market close)

600s

Trade in the final N seconds before market closes — resolution momentum locked in

Position Sizing Mode

Manual — you set kelly cap & max size Adaptive — computed from trade stats

Kelly Cap (manual)

10%

Fractional Kelly cap — 10% recommended for Phase A

Max Trade Size (USD)

Hard dollar cap per trade — Kelly fraction applied to balance then capped here

Max Daily Loss (USD) Override daily loss limit (useful for paper testing)

Paper Balance (USD)

Starting balance for paper trading

Virtual Loss Filter

Enable Virtual Loss — simulate entries until N virtual losses, then trade for real

While not armed: same bot rules, but no real/paper dry balance impact. After one real WIN, filter resets.

Required virtual losses before real trade

Paper Trading Mode

Paper Trading (uncheck = LIVE real money)

Paper Balance ($)

Paper trading ACTIVE — all trades are simulated, no real money

Chainlink RPC Config

RPC Endpoint (optional)

Leave blank to use default public nodes

Claude AI Configuration

Enable Claude AI Analysis

Claude API Key

Anthropic API key — stored encrypted

Model

Auto-analyze every 50 trades

API Key not configured

Polymarket Wallet

Wallet Address

Signature Type

Для V2 должен совпадать с типом аккаунта в Polymarket.

Funder Address

Если пусто — используется Wallet Address. Для proxy/deposit flow укажи funder явно.

Builder Code (optional)

Private Key (not set)

Required for live trading. Stored encrypted. Never shared.

Geo Block Token (not set)

Required if server IP is geo-blocked by Polymarket. Get from browser DevTools → any clob.polymarket.com POST request → geo_block_token query param.

CLOB Proxy URL (not set)

Route order placement through a local HTTP proxy to bypass server-side geo-block. Run a proxy on a non-blocked machine (e.g. your local PC with VPN) and expose it via ngrok. Format: http://host:port

USDC Balance: Not connected

Auto-redeem resolved positions (LIVE)

После резолва рынка отправляет on-chain redeem через Polygon (нужен MATIC на кошельке с приватным ключом). Частота — ниже.

Auto-redeem interval (sec)

Минимум 120 с. По умолчанию раз в 10 минут.

Require Whale Convergence

Only trade if copy targets are trading in same direction

Enable Latency Arb Mode (optional)

Separate strategy mode: Binance implied probability vs Polymarket price. Disabled = current Blast logic unchanged.

Simple last-minute mode

Отдельный режим: только последняя минута до экспирации (≤60 с), вход market-buy по текущей цене, если сторона ≥ 0.99 (ask или Gamma). Пока включён — обычный Φ/EV пайплайн не используется.

Latency edge threshold (pp)

Signal only if |impliedProb - polyPrice| >= threshold. 0.10 = 10 percentage points.

Pre-close window (sec)

Only trade when time remaining is inside this final window before expiry.

Slope guard window (sec)

Short BTC momentum confirmation window to avoid reversal entries.

⚙ Admin Panel

TOTAL USERS

—

ACTIVE BOTS

—

TOTAL TRADES

—

PLATFORM P&L

—

TRADES (24H)

—

P&L (24H)

—

WIN RATE

—

Users

Email	Role	Bot	Mode	Trades	Win%	P&L	Actions
Loading…

All Trades

User	Market	Dir	Entry	Size	P&L	Result	Time
Loading…

Admin Action Log

Live Bot Logs ● LIVE Auto-refresh

      Loading logs…
    