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 1 said the rails may be rewritten in Q4.
Part 2 is the layer that should not get rewritten just because the CLOB might.
If a 5-minute crypto bot un-pins settlement while the matching engine is in flux, it will blame “the new CLOB” for losses that were just the old mixed-window bug.
1. What Is Still Live
As of mid-September 2026, crypto Up/Down markets still resolve on Chainlink TWAP:
| Market | Official window |
|---|---|
| 5-minute | 60-second TWAP |
| 15-minute | 60-second TWAP |
| 4-hour | 60-second TWAP |
5m markets started at 30s on Aug 7, then moved to 60s on Aug 14.
Hourly / daily crypto are a different product family. Do not train a 5m engine on those closes.
60 seconds is a lookback, not a heartbeat.
2. The Question Did Not Change
Wrong:
Where does Binance print at the bell?
Right:
Where does the official 60s average sit versus the slot-open reference?
A Q4 matching rewrite can change how fast your order lands.
It does not change which series decides Up vs Down — unless the market rules themselves change.
Until the live market text says otherwise, pin TWAP60.
3. The State That Must Survive a CLOB Cutover
twap_open = 60s Chainlink TWAP at slot start
twap_now = latest 60s Chainlink TWAP
t_left = seconds remaining
resolution_src = rtds_twap | gamma
That row belongs above clob.ts, not inside it.
If the matching engine is replaced and you have to swap order adapters, this state should still be valid.
If you buried open-pinning inside order-placement code, the rewrite will delete your settlement identity.
4. Feed Rules That Still Apply
twapFeed.ts job is unchanged:
- RTDS 60s series /
windowSeconds: 60 - filter like
{"symbol":"btc/usd"} - value + observation timestamp
- reconnect with backoff
- refuse stale data
- exact decimal / bigint math
RTDS blip ≠ use CEX last.
RTDS blip = no official series for that slot.
Gamma remains the post-resolution fallback. It is not a live path.
5. The Bug People Will Reintroduce During the Rewrite
When teams “prepare for the new CLOB,” they often touch too much:
| Tempting shortcut | What it actually does |
|---|---|
| Reuse old 30s notebooks | Contaminated labels |
| Pin open from Binance, score TWAP | Fake edge |
| Keep snapshot research “just for features” | Teaches the model the retired target |
Wait to log resolution_src until Q4 |
You will not know what broke |
The CLOB can change. Mixed windows stay fatal.
6. What I Am Doing on This Layer
For @abrownfox001:
- production 5m path stays 60s TWAP only
- open and live score stay same-series
- CEX mid stays auxiliary lead-lag
- every completed slot logs
resolution_src - no settlement experiment bundled into an execution refactor
The private P(up) can be retuned on official labels.
It should not be retuned on a matching rumor.
7. What Part 3 Covers
Once settlement is pinned, the moving part is execution:
- current 150ms crypto taker delay
- volume-tier limits
- a possible Rust CLOB with different p99 behavior
- which latency measurements will survive that cutover
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. Confirm the live market’s official resolution source before you trade it.
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 #Chainlink #BTC #TradingBot #AlgoTrading #PredictionMarkets
Top comments (0)