DEV Community

ABROWNFOX001
ABROWNFOX001

Posted on

abrownfox001: A TWAP-Anchored BTC 5-Minute Directional Engine on Polymarket

Repo: abrownfox001-twap60-prediction-trigger-system

I run a live automated bot on Polymarket focused almost exclusively on BTC 5-minute Up/Down markets.

It is not a 94–99¢ convergence farm.

It is a directional system that enters the coin-flip band (~45–55¢), holds only high-conviction legs, and scratches the rest near breakeven.

Profile: @abrownfox001

Wallet (proxy): 0x12b712029efeee7efc4066e98e4e1b6beb8b1b0b

Active since: Jan 12, 2026

Executive Summary

Convergence farmer (typical) abrownfox001
Price zone 94–99¢ ~45–55¢ (~90% of trades)
Edge source Structural near-certainty TWAP-anchored directional forecast
Monetization Tiny edge × huge volume ~61% held-side win rate × fair entries

Resolved sample (150 markets, entries ≈ 50¢, Jul 2026 public data):

Metric Value
Held-side win rate 60.7% (91W / 59L)
ROI on capital at resolution ≈ +24.5%
Avg entry on held legs ~0.50

The core result: turn a market most people treat as 50/50 into roughly 61/39 on the book that actually settles.

The Three Problems This System Solves

1. Calibrated short-horizon signal (private)

BTC 5m books sit near 50¢ because short-horizon direction is almost random. A sustained ~60% read requires forecasting where settlement is heading under TWAP rules — live Chainlink TWAP vs slot-open reference, plus lead-lag, flow, and time-in-slot features — not a noisy CEX tick.

2. Fair-value entry at size

A view is useless if you overpay. Observed behavior:

  • Mean entry ~0.503
  • Clips roughly $50–175
  • Entries spread across the slot (median buy ~47s after open)

3. Breakeven conviction culling

In a large public sample, ~930 markets show buy and sell at essentially the same price (avg buy 0.504 / sell 0.503). Those are scratch exits when conviction decays. Only strong reads ride to redemption. That filter is why resolved win rate can exceed raw signal accuracy at near-zero scratch cost.

One line: TWAP-anchored 5m BTC direction, executed at fair ~50¢, with scratches that cull weak tickets so the settlement book is mostly the good reads.

What It Trades

Dimension Pattern
Asset BTC (~99% of activity)
Window 5-minute Up/Down (~96%)
Price zone ~90.6% of fills in 45–55¢
Other Occasional other assets — noise, not core

Single-market specialist. No politics, no long-dated convergence sweeps as core business.

Measured Behavior (Public Sample)

Metric Value
Trades 5,000 · ~126 days · ~40/day
Buy : Sell 3,643 : 1,357 (72.9% / 27.1%) — active scratches
Median notional $57.75 · p90 $176.52
Unique markets 2,570
Round-tripped (scratch) ~930 · spread ≈ 0

Intra-slot timing:

  • Buys skew earlier (median ~47s) — accumulate near 50¢
  • Sells skew later (median ~226s) — conviction cull as the slot matures

Slot Lifecycle 

SlotOpen
  → Signal: P(up) vs 0.5
  → Accumulate: BUY near 50¢ if edge clears
  → ReScore every 10–30s
       ├─ decayed → Scratch ~breakeven
       └─ intact  → Hold → Redeem winner
Enter fullscreen mode Exit fullscreen mode

Style: directional micro-forecasting + risk culling — not market making, not 99¢ farming.

Economics at ~50¢

For a binary held to $1 at entry ( p \approx 0.50 ) and win rate ( w ):

[
EV \approx 0.50 \cdot (2w - 1)
]

At ( w = 0.607 ): naïve EV ≈ +10.7% of stake per resolution.

The higher sample ROI can look richer because winners are often accumulated below 0.50 and losers are scratched before a full $0 outcome.

Fees bind far less at 50¢ than at 99¢. The real constraints are signal accuracy, latency, and fill quality.

System Pipeline

  1. Signal (private) — latest Chainlink TWAP vs slot-open reference → calibrated P(up)
  2. Entry — if edge clears threshold and ask is in 45–55¢, accumulate small clips
  3. Cull — re-score every 10–30s; scratch on decay; else hold to redeem

Reference TypeScript modules:

Module Role
engine.ts Slot lifecycle
twapFeed.ts Chainlink TWAP stream (RTDS)
signal.ts P(up) vs open reference
cexFeed.ts Auxiliary lead-lag
markets.ts Gamma discovery of btc-updown-5m-*
clob.ts CLOB v2 orders

Standalone Windows binary (Abrownfox001.exe) available with a 2-day trial. Source runs via npm start (dry-run by default).

Risks & Caveats

  1. The edge is the signal. Without a real ~60% read, 50¢ trading is coin-flipping minus fees.
  2. Latency & adverse selection — slow infra buys the wrong leg.
  3. Regime decay — crowded books and volatility shifts erode calibration.
  4. Sample limits — figures above are from public windows, not audited lifetime P&L.
  5. Scratch needs depth — thin books turn “breakeven” into real spread cost.
  6. TWAP settlement — 5-minute crypto markets settle on Chainlink TWAP (60s window as of the Aug 2026 update). Models trained on single-price snapshots need recalibration.

Why the Signal Stays Private

The moat is the calibrated 5m read and how it is executed at size near 50¢. Publishing that competes the edge away. This repo proves authorship and documents the archetype — it does not open-source alpha.

Verification is public: profile, wallet, and API-visible fills.

curl "https://gamma-api.polymarket.com/public-profile?address=0x12b712029efeee7efc4066e98e4e1b6beb8b1b0b"
curl "https://data-api.polymarket.com/trades?user=0x12b712029efeee7efc4066e98e4e1b6beb8b1b0b&limit=1000"
Enter fullscreen mode Exit fullscreen mode

Bottom Line

abrownfox001 is a specialized BTC 5-minute directional engine:

  • Anchored to Chainlink TWAP settlement
  • Enters only in the fair mid-band
  • Actively scratches weak tickets
  • Holds only higher-conviction legs to resolution

Not financial advice. Paper or micro-size first. Latency, regime shifts, and fill quality still matter.

Repo: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system

If you have more questions, please feel free to contact me at any time: https://t.me/abrownfox001

My Polymarket Activity: https://polymarket.com/@abrownfox001?tab=activity

#Polymarket #BTC #TWAP #TradingBot #AlgoTrading #Chainlink #PredictionMarkets #QuantTrading
Enter fullscreen mode Exit fullscreen mode

Top comments (0)