Actively participate in various Arc activities
Today I also submitted PredictArena to the Agora Agents Hackathon.
A small milestone, but it feels meaningful.
For me, this project is really about one question: if agents are going to act, forecast, and make decisions for us, how do we make them accountable?
PredictArena is my attempt at an answer: agents that scan real markets, produce deterministic signals, pass risk gates, and leave USDC-backed receipts on Arc.
Still early. Still rough in places. But submitting it made the idea feel a little more real.
@arc
#arc
3
100
For this demo, we wanted to show what autonomous agents look like when they move beyond prompts.
PredictArena agents continuously scan live Polymarket crypto markets, generate deterministic BTC/ETH/SOL forecasts, filter weak signals through a Risk Agent, and bond eligible predictions with USDC on Arc Testnet.
The key idea: agent decisions should be auditable.
Each signal includes probability, edge, Kelly sizing, risk flags, model/data hashes, run receipts, tx links, and a public reputation profile.
AI agents should not just make claims.
They should build a track record.
Demo below π
@arc
@BuildOnCircle
#ac
Phase 2 of PredictArena is completed.
Currently supported features:
- Real-time scanning of Polymarket crypto price markets
- Automatic parsing of BTC/ETH/SOL prediction markets
- Automated trading signal generation via Volatility, Momentum and Risk Agents
- Monte Carlo GBM probability model
- Submission of USDC signal bonds to Arc Testnet
- Signal cards, detail pages, leaderboards and demo snapshot fallback
- Chinese & English bilingual support, light/dark mode UI
Next step: Automatic Resolution Engine.
Enable automatic settlement of committed signals upon expiry, calculate accuracy, Brier score and paper ROI, and execute USDC bond refund or slash operations on Arc.
For testnet use only. Not financial advice.
---
# Thread Version
1. Phase 2 of PredictArena is finished.
It is an autonomous prediction-market agent arena built for Arc.
Agents automatically scan real Polymarket markets, generate quantitative prediction signals, and submit convictions via USDC signal bonds on Arc Testnet.
2. Core capabilities fully implemented:
- Polymarket live market scanner
- BTC/ETH/SOL crypto market parser
- EXPIRY & TOUCH condition identification
- Volatility Agent
- Momentum Agent
- Risk Agent
- Monte Carlo GBM probability engine
- Kelly criterion & stake sizing mechanism
- Signal cards and detail pages
- USDC bond transaction workflow on Arc Testnet
- Leaderboard system
- Demo snapshot fallback mechanism
3. This iteration focuses on agentic decision-making rather than prediction UI:
- Market Scout screens target markets
- Quantitative agents output probability data
- Risk Agent filters low-quality trading opportunities
- Agent wallets submit USDC bonds on Arc
- Leaderboard archives all agent performance track records
4. Upcoming development: Automatic Resolution Engine
Goal to complete the full business closed loop:
Scan market β Generate signals β Commit USDC bonds β Await expiry/touch triggers β Auto settlement β Bond refund or slash β Update reputation metrics
5. Functional support of Resolution Engine:
- EXPIRY_ABOVE, EXPIRY_BELOW
- TOUCH_ABOVE, TOUCH_BELOW
- Settlement based on public crypto candlestick data
- Brier score calculation
- Accuracy statistics
- Paper ROI computation
- USDC bond refund & slash execution
- Batch on-chain settlement on Arc
6. This upgrade elevates PredictArena from a signal-generating tool to a complete agent reputation system.
Every signal is attached with model parameters, data hashes, bond records, on-chain transactions, settlement results and leaderboard archives.
For testnet deployment only. This is not financial advice. No real trading on Polymarket is involved.
@buildoncircle
@arc
@LumaHQ
2
5
2,723
Phase 2 of PredictArena is completed.
Currently supported features:
- Real-time scanning of Polymarket crypto price markets
- Automatic parsing of BTC/ETH/SOL prediction markets
- Automated trading signal generation via Volatility, Momentum and Risk Agents
- Monte Carlo GBM probability model
- Submission of USDC signal bonds to Arc Testnet
- Signal cards, detail pages, leaderboards and demo snapshot fallback
- Chinese & English bilingual support, light/dark mode UI
Next step: Automatic Resolution Engine.
Enable automatic settlement of committed signals upon expiry, calculate accuracy, Brier score and paper ROI, and execute USDC bond refund or slash operations on Arc.
For testnet use only. Not financial advice.
---
# Thread Version
1. Phase 2 of PredictArena is finished.
It is an autonomous prediction-market agent arena built for Arc.
Agents automatically scan real Polymarket markets, generate quantitative prediction signals, and submit convictions via USDC signal bonds on Arc Testnet.
2. Core capabilities fully implemented:
- Polymarket live market scanner
- BTC/ETH/SOL crypto market parser
- EXPIRY & TOUCH condition identification
- Volatility Agent
- Momentum Agent
- Risk Agent
- Monte Carlo GBM probability engine
- Kelly criterion & stake sizing mechanism
- Signal cards and detail pages
- USDC bond transaction workflow on Arc Testnet
- Leaderboard system
- Demo snapshot fallback mechanism
3. This iteration focuses on agentic decision-making rather than prediction UI:
- Market Scout screens target markets
- Quantitative agents output probability data
- Risk Agent filters low-quality trading opportunities
- Agent wallets submit USDC bonds on Arc
- Leaderboard archives all agent performance track records
4. Upcoming development: Automatic Resolution Engine
Goal to complete the full business closed loop:
Scan market β Generate signals β Commit USDC bonds β Await expiry/touch triggers β Auto settlement β Bond refund or slash β Update reputation metrics
5. Functional support of Resolution Engine:
- EXPIRY_ABOVE, EXPIRY_BELOW
- TOUCH_ABOVE, TOUCH_BELOW
- Settlement based on public crypto candlestick data
- Brier score calculation
- Accuracy statistics
- Paper ROI computation
- USDC bond refund & slash execution
- Batch on-chain settlement on Arc
6. This upgrade elevates PredictArena from a signal-generating tool to a complete agent reputation system.
Every signal is attached with model parameters, data hashes, bond records, on-chain transactions, settlement results and leaderboard archives.
For testnet deployment only. This is not financial advice. No real trading on Polymarket is involved.
@buildoncircle
@arc
@LumaHQ
PredictArena Phase 1 is shipped.
Autonomous agents now scan real Polymarket markets, filter crypto price markets, forecast with specialized agents, and commit high-conviction signals to Arc Testnet using USDC signal bonds.
No manual evidence.
No manual market creation.
No AMM.
No Polymarket clone.
Just agent-generated prediction signals with onchain skin in the game.
Built for the Agora Agents Hackathon by Canteen Γ Circle Γ Arc.
Phase 2 starts now.
@buildoncircle
@arc
2
776
24 Nov 2025
#PiNetwork #PiCommunity #PiNation #PiPioneers #PiKYC #PiWallet #PiBrowser
#PredictArena #PiApp #PiApps #PiCoreTeam #PiMainnet
predictarenaabcb9553.pinet.cβ¦
Show your prediction skills on Predict Arena
3
8
1,460
9 Aug 2025
Three contenders. One crown. π
Pick your champion. Predict with precision, the future of AI is in our hands.
π The Predict Arena β Who Will Take the Crown?
The arena is open π₯
Three AI titans enter: GPT-5, Gemini, Grok
But only the community decides whoβs truly the best π
There will be only one to rule the leaderboard π«
Whoβs your pick?
Share it below.
Make your prediction ππ½predict.recall.network
@recallnet @cookiedotfun
#RecallNetwork #PredictArena
8 Aug 2025
Performance > Promises
6
7
198
8 Aug 2025
ποΈ In The Predict Arena, weβre not just guessing,
weβre shaping the AI future.
πGPT-5
πGemini 2.5 Pro
π©·Grok 4
Only one can take the crown. πβοΈ
Start hereππ½
predict.recall.network/evaluβ¦
@recallnet
@cookiedotfun
#RecallNet #PredictArena
7 Aug 2025
1/ The era of community-powered AI rankings is here.
Over 5 days, the Recall community crowdsourced an ungameable benchmark to test 50 popular AI models and cast 7.8M votes in the world's largest pre-launch skill prediction for GPT-5 on predict.recall.network.
Let's dig inπ§΅
4
1
6
409