DEV Community

ABROWNFOX001
ABROWNFOX001

Posted on

5m CLOB, Combos, and US Rails Are Not One Bot

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

YouTube walkthrough: https://www.youtube.com/watch?v=XzhugRL6BV4

Live profile: https://polymarket.com/@abrownfox001

Part 3 stayed inside the 5-minute matching loop.

Part 4 is the product split. Polymarket now looks like one brand with several machines. A BTC 5m engine that absorbs all of them will not survive a Q4 CLOB rewrite — it will drown in adapters.

1. Three Surfaces, Three State Machines

Surface Matching Settlement / payoff Bot implication
Crypto 5m / 15m / 4h Up/Down CLOB 60s Chainlink TWAP Current abrownfox001 core
Combos (Polymarket US) RFQ, not a resting CLOB All legs must win Separate risk system
Broader US / onchain product push Mixed Category rules Infra + compliance, not 5m alpha

“Polymarket shipped it” is not a reason to extend engine.ts.

2. Why Combos Do Not Inherit TWAP60 Logic

A Combo is a multi-leg parlay:

  • 2–10 legs
  • one payout only if every leg wins
  • quote request, short response window
  • residual-leg risk if the package does not land clean
  • maker-side capital features that 5m taker clips do not use

A 5m BTC ticket is the opposite:

  • one market
  • one official TWAP path
  • one 300-second clock
  • scratch back to flat on the same book

You can share a logger.

You cannot share P(up), scratch timers, or delay assumptions.

3. Why US Rails Are Also Not a Strategy Patch

The US product push, debit-card / withdrawal operations, and onchain/DeFi unification are real company work.

They change:

  • who can onboard
  • how money moves
  • which app a user sees
  • which compliance checks exist

They do not change whether btc-updown-5m-* pays Chainlink TWAP60.

If you retune a directional 5m model because of a US-rail headline, you mixed corporate news into settlement math.

4. Builder Surface Is Noise Until It Touches Fills

Pitch days, grants, Exa Connect, agent wrappers — useful for distribution.

For a live specialist they are clients of the engine, not the engine.

Expected side effect: more sloppy order loops and more rate-limit pressure.

That is an ops problem. It is not a reason to add assets or abandon mid-band entries.

5. The Adapter Map I Want Before Q4

Risk + logger
  ├─ 5m BTC CLOB adapter     → TWAP60 + delay-aware taker   (production)
  ├─ data/verify adapter     → Gamma + Data API             (always)
  ├─ Combos/RFQ adapter      → separate repo                (research only)
  └─ agent/connector client  → read-only or hard-capped     (optional)
Enter fullscreen mode Exit fullscreen mode

When the Rust matching engine lands, only the 5m CLOB adapter should move first.

If Combos and 5m share a class hierarchy, the cutover will ship both bugs at once.

6. What I Will Not Port

  • 5m TWAP features onto Combos
  • Combos quote timeouts onto BTC scratches
  • US withdrawal headlines into signal.ts
  • agent auto-placement as production
  • “more markets” as a hedge against CLOB uncertainty

Specialization is the point of @abrownfox001.

The public footprint is still BTC, 5m, 45–55¢, with later sells than buys. That is the product.

7. Checklist Before You Touch Another Surface

  • [ ] 5m discovery still resolves to live TWAP60 slugs
  • [ ] positions reconcile via Data API
  • [ ] Combos code cannot place 5m orders
  • [ ] agent keys cannot raise size
  • [ ] one kill switch can halt every adapter

8. What Part 5 Covers

The last post is the operating plan:

  • keep the specialist loop
  • freeze strategy weights across a matching cutover
  • patch only feeds, delay logs, and adapters
  • leave the calibrated read closed

Links

Not financial advice. Product surfaces move — confirm against live docs.

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 #Combos #TWAP #TradingBot #BTC #AlgoTrading #PredictionMarkets
Enter fullscreen mode Exit fullscreen mode

Top comments (0)