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 was about making clob.ts boring enough to survive a matching-engine rewrite.
Part 4 is the refusal list. Platforms ship surface when rails are unstable. Most of that surface does not belong inside a BTC 5-minute TWAP engine.
1. The Rule
If it does not help me:
- pin 60s TWAP
- enter near 50¢
- scratch decay
- reconcile fills
…it waits.
Q4 is a rails season. Rails seasons punish extra state machines.
2. Will Not Port: Combos / RFQ
Combos are a multi-leg quote product. They are not “5m Up/Down with more outcomes.”
Different matching.
Different clock.
Different residual risk.
Different capital rules.
I will not:
- call RFQ from
engine.ts - reuse 5m scratch timers on a parlay
- treat combo volume as proof the 5m loop should widen
If Combos ever get code, they get a separate adapter and a separate loss cap.
3. Will Not Port: Kalshi 15m Weights
Same word — 60-second average — different index, different clock, different strike convention.
A Polymarket 5m TWAP path follower is not a Kalshi final-minute CF reconstruction bot.
Sharing P(up) weights across those two books is how you manufacture fake confidence.
4. Will Not Port: Agent Autopilot
Exa Connect and builder tooling are useful as clients.
They are not the production trader.
I will not let an agent loop:
- skip freshness checks
- ignore rate-limit backoff
- place size because a prompt said “BTC looks up”
- bypass per-slot caps
An agent can read logs. It does not get an unsigned path to clob.ts.
5. Will Not Port: Perps, Passkeys, or “Unified DeFi” UX
Adjacent product work can matter to the company.
It does not change the 5m specialist thesis.
No perps margin logic in this repo.
No “one account trades everything” wrapper.
No new asset spray to hide a weaker BTC read.
6. Will Not Port: Snapshot Habits Back Into TWAP60
This one is a regression, not a feature.
Banned:
- CEX-close labels
- 30s open + 60s close
- t=0 size dumps as a default
- last-8-second “settlement snipes”
- hold-everything because the first
P(up)was 0.56
The official series is a rolling 60s average. Trading it like July 2026 snapshot flow is not conservative. It is stale.
7. What Can Be Shared
These travel without becoming a new strategy:
- kill switch
- daily loss cap
- structured slot logs
- idempotent order IDs
- Retry-After handling
- stale-feed hard skip
- Data API / Gamma reconciliation
That is the risk brain. It can supervise one production adapter.
8. The Production Boundary
Risk + logger
└─ BTC 5m adapter
twapFeed.ts → 60s official path
signal.ts → private P(up)
engine.ts → enter / scratch / redeem
clob.ts → matching adapter
Everything else stays outside the process until the current book is measured through a matching change, not just announced.
9. Why Refusal Is the Work
A rewritten CLOB plus a new product line creates the same temptation:
Add markets so volume looks alive while fills get worse.
I would rather keep the public footprint readable:
- BTC
- 5m
- 45–55¢
- later sells than buys
- inspectable wallet
If that footprint breaks, the machine changed. Shipping Combos into the same binary would only hide it.
10. What Part 5 Covers
The last post is the concrete patch list through Q4:
- what already landed after TWAP60
- what I am measuring before any CLOB cutover
- what stays private
- how to read @abrownfox001 while the venue rebuilds rails
Links
- Repo: https://github.com/abrownfox0/abrownfox001-twap60-prediction-trigger-system
- Video: https://www.youtube.com/watch?v=XzhugRL6BV4
- Live activity: https://polymarket.com/@abrownfox001?tab=activity
Not financial advice. New products are not a reason to mix state machines.
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 #TWAP #TradingBot #BTC #CLOB #AlgoTrading #PredictionMarkets
Top comments (0)