Automated arbitrage across prediction markets. BTC Up/Down module on @Solana. 90ms execution. Powered by @jup_predict. Runs 24/7.
Joined February 2025
- Tweets322
- Following24
- Followers8.8K
- Likes128
Pinned Tweet
We built the cleanest BTC Up/Down bot on Polymarket.
We engineered the entire stack from scratch around one principle: never bet on outcomes, only capture mispricings.
How it works:
When YES + NO sums to less than $1.00 on Polymarket’s 15-min BTC market, the engine fires both sides atomically via Jito bundles. Total cost: $0.97. At resolution, one side always pays $1.00. Locked profit: 3 cents per dollar deployed.
BTC’s direction is irrelevant. The math already paid.
We don’t predict. We capture.
Architecture:
- Yellowstone gRPC at 1-5ms orderbook latency
- Jito atomic bundles, both legs fill or neither does
- 91ms median detection-to-fill
97.8% fill rate over 30 days
- Auto-hedge engine for partial fills in 4.2s
How to plug in:
1.Sign up at PolyArb (link in bio)
2.Copy PolyArb dedicated RPC URL from the dashboard
3. Open @jup_predict on Solana
4.Paste the RPC into Settings → Custom RPC
5.Activate the BTC Up/Down module
Free to use. Profit-share only. No subscription. No upfront cost.
PolyArbiter retweeted
Up/down polymarkets just got upgraded.
Trade 5m, 15m, & hourly crypto up/downs with time-weighted average pricing (TWAP).
To celebrate, we're giving away $1M in liquidity rewards across BTC, ETH, SOL, XRP, HYPE, BNB, & DOGE polymarkets.
PolyArbiter remains one of the most powerful BTC UP/DOWN engines in production and it has never needed to predict Bitcoin.
On @jup_predict’s 15-minute BTC UP/DOWN market, one token settles at $1 and the other at $0. A perfectly matched pair therefore has a fixed terminal value of $1, regardless of where BTC closes.
The engine acts only when:
inf Π = min(qUP, qDOWN) − C_all-in > ε
C_all-in includes executable prices, fees, slippage, priority fees, Jito tips and a safety margin.
Before moving capital, PolyArbiter verifies the round, timestamps, oracle and settlement rules; sizes from executable depth; matches realized quantities; simulates the bundle; confirms landing; and reconciles settlement.
If the edge disappears at any stage, it refuses the trade.
PolyArbiter retweeted
Scheduled CLOB maintenace tomorrow, May 27 at 12:10 UTC.
- Intermittent trading for ~5-10 min during the window.
- Brief post-only mode after restart (~2 min) before normal matching resumes.
Stay updated, subscribe at status.polymarket.com/
You either plugged into our Polymarket BTC Up/Down bot 48h ago or you didn’t.
48 hours after this tweet, here’s what the data shows:
- 412 active wallets, every single one in profit
~5% average return, top performers near 9%
- Edge holding strong at 0.6-0.9% per capture
- Zero drawdowns above 1.7%
How to plug in:
1.Sign up at PolyArb (link in bio)
2.Copy your dedicated RPC URL
3. Open @jup_predict on Solana
4.Paste into Settings → Custom RPC
5.Activate the BTC Up/Down module
Free to use. Profit-share only.
We built the cleanest BTC Up/Down bot on Polymarket.
We engineered the entire stack from scratch around one principle: never bet on outcomes, only capture mispricings.
How it works:
When YES + NO sums to less than $1.00 on Polymarket’s 15-min BTC market, the engine fires both sides atomically via Jito bundles. Total cost: $0.97. At resolution, one side always pays $1.00. Locked profit: 3 cents per dollar deployed.
BTC’s direction is irrelevant. The math already paid.
We don’t predict. We capture.
Architecture:
- Yellowstone gRPC at 1-5ms orderbook latency
- Jito atomic bundles, both legs fill or neither does
- 91ms median detection-to-fill
97.8% fill rate over 30 days
- Auto-hedge engine for partial fills in 4.2s
How to plug in:
1.Sign up at PolyArb (link in bio)
2.Copy PolyArb dedicated RPC URL from the dashboard
3. Open @jup_predict on Solana
4.Paste the RPC into Settings → Custom RPC
5.Activate the BTC Up/Down module
Free to use. Profit-share only. No subscription. No upfront cost.
PolyArbiter retweeted
Exchange upgrades starting now. The platform is under maintenance for ~1 hour. Trading is paused and the orderbooks are being cleared. Funds and positions are safe.
$1M in liquidity rewards go live when trading resumes. Updates in this thread.
The 5-minute guide to plugging into the BTC Up/Down Polymarket module without building anything yourself.
Bookmark this if you want to skip months of engineering.
Step 1: What you’ll need (2 minutes prep)
A Solana wallet (Phantom or Jupiter work fine)
Step 2: Sign up at PolyArb (link in bio)
Step 3: Copy our dedicated RPC URL
After signup, your dashboard shows a unique Solana RPC endpoint. This is your access to:
-Yellowstone gRPC at 1-5ms (vs 200-800ms public RPCs)
-Stake-weighted QoS
-Jito atomic bundle support
-91ms median execution
Step 4: Plug into Jupiter Predict
Open jup.ag/predict. Settings → Custom RPC → paste our dedicated RPC. This routes every transaction through PolyArb’s execution stack.
Step 5: Toggle the BTC Up/Down module
to ACTIVE.
Set your edge threshold (default: 0.8% net).
That’s it. Engine starts scanning immediately.
What happens next:
The engine monitors every 15-min BTC market on Polymarket continuously. When YES + NO sums below $1.00, it fires both legs atomically via Jito bundles. Locked profit at fill, regardless of where BTC closes.
You don’t need to monitor anything. You don’t need to predict anything. You don’t need to watch charts.
First trades typically execute within 1-2 hours depending on market volatility.
Dashboard updates in real-time.
Free to use. (Profit-sharing only)
We built the cleanest BTC Up/Down bot on Polymarket.
We engineered the entire stack from scratch around one principle: never bet on outcomes, only capture mispricings.
How it works:
When YES + NO sums to less than $1.00 on Polymarket’s 15-min BTC market, the engine fires both sides atomically via Jito bundles. Total cost: $0.97. At resolution, one side always pays $1.00. Locked profit: 3 cents per dollar deployed.
BTC’s direction is irrelevant. The math already paid.
We don’t predict. We capture.
Architecture:
- Yellowstone gRPC at 1-5ms orderbook latency
- Jito atomic bundles, both legs fill or neither does
- 91ms median detection-to-fill
97.8% fill rate over 30 days
- Auto-hedge engine for partial fills in 4.2s
How to plug in:
1.Sign up at PolyArb (link in bio)
2.Copy PolyArb dedicated RPC URL from the dashboard
3. Open @jup_predict on Solana
4.Paste the RPC into Settings → Custom RPC
5.Activate the BTC Up/Down module
Free to use. Profit-share only. No subscription. No upfront cost.
In the last 30 days, the engine has:
Scanned 1.2M market states across 12 venues
Flagged 3,284 spreads above 0.8% threshold
Executed 418 trades
Fill rate: 96.4%
Avg execution: 94ms
Partial fills: 15
Realized P&L across all users: +$147,300
All while our users were sleeping, working, or not thinking about prediction markets.
That's the product.
PolyArbiter retweeted
Polymarket's V2 exchange upgrades go live April 22nd. The upgraded exchange is now open to all for testing.
🚨 If you use our API or clients, there's migration work to do before then:
→ Upgrade your SDK/ API integration
→ New exchange contract addresses & order struct
→ Collateral moves to pUSD (wrap USDC.e via the Collateral Onramp)
We expect ~1 hour of downtime on April 22nd at ~11am UTC. Open limit orders will be cancelled during the switch. Funds and positions are safe.
Please note that the V1 exchange will stop working after the migration.
Migration guide: docs.polymarket.com/v2-migra…
Security: Our CTFv2 contracts are open sourced & audited by Cantina + Quantstamp. $5 million bug bounty program is live on Cantina: cantina.xyz/bounties/ff945ca…
Changelog: docs.polymarket.com/changelo…
We’re building a latency Football bot for Polymarket. Target: operational before the World Cup. Save this.
The principle: pro sports feeds (Sportradar, Opta) deliver pitch events in ~200-500ms. The Polymarket orderbook takes longer to reprice thin liquidity, market makers pulling quotes while they reassess. That window is the edge.
Yesterday, first live test on Bayern vs Real (Champions League QF).
86th minute, Camavinga gets his second yellow. Pipeline receives the event from Sportradar 280ms after the card. “Bayern advances” market was at ~0.55. Model recomputes to ~0.68 post-red and fires a $500 order.
Partial fill as expected: ~$180 caught around 0.55-0.57, the rest slipped to 0.63. Average 0.59. Market stabilized at 0.67 a few seconds later. Unrealized +$30. +6% in seconds.
It’s a test. But the loop worked end to end detection, decision, fill, before the book caught up.
What we learned: network latency is part of the problem. The real bottleneck is orderbook depth. We’re competing with sharp bots, not retail on their couch. And “next goal” markets have better spreads than qualification markets. Pivoting there.
What we’re building before June: fill routing across 12 venues via Jito atomic bundles. Low-signal event modeling (dangerous fouls, injuries, tactical shifts). UMA oracle hedging. Node co-location near Sportradar servers.
Why the World Cup matters.
104 matches in 39 days. $2.5B+ in projected prediction market volume. Deep liquidity means bigger positions fill cleanly. Thin liquidity in group stages means wider spreads. Both environments leave serious money on the table.
56 days to ship. We’re on it.
For anyone curious about our core product the football latency module is an experiment we’re still testing.
What actually runs 24/7 is our cross-platform arbitrage engine:
We built a Polymarket execution engine. Here’s today’s output.
17 fills today. $118.40 P&L. 94% win rate. 19ms ping. 1,135 TPS.
5 routes running: POLY/KLSH. KLSH/POLY. POLY/PRED. PRED/KLSH. PRED/POLY.
This is what 50,000 lines of code and 4 months of engineering looks like when it runs.
We made it free. (Profit-sharing) Set your parameters and run.
To clarify we’re testing whether this holds at scale. Right now it’s a side project, not a shipped product. We’ll post ETAs and actual results here as we go.
Our core is cross-platform arb. That’s been running since mid-2025 without interruption. Sports latency is an extension we’re exploring because the window is there but we’re not promising anything until we’ve validated throughput across more matches.
We’ll update as we build.
Stop building Claude bots.
We’re serious.
We spent 4 months and 50,000 lines of code building the infrastructure that every Claude tutorial pretends doesn’t matter.
The strategy isn’t the problem. Everyone knows the strategy. Buy the spread. Capture the gap. The math is simple.
The problem is execution. And no tutorial will give you that.
We already built it. It’s free. Profit-sharing only.
Stop building. Start using.
We built a Polymarket execution engine. Here’s today’s output.
17 fills today. $118.40 P&L. 94% win rate. 19ms ping. 1,135 TPS.
5 routes running: POLY/KLSH. KLSH/POLY. POLY/PRED. PRED/KLSH. PRED/POLY.
This is what 50,000 lines of code and 4 months of engineering looks like when it runs.
We made it free. (Profit-sharing) Set your parameters and run.
We built a Polymarket execution engine. Here’s today’s output.
17 fills today. $118.40 P&L. 94% win rate. 19ms ping. 1,135 TPS.
5 routes running: POLY/KLSH. KLSH/POLY. POLY/PRED. PRED/KLSH. PRED/POLY.
This is what 50,000 lines of code and 4 months of engineering looks like when it runs.
We made it free. (Profit-sharing) Set your parameters and run.
Breakdown on the 3 losses:
Fed decision resolution criteria differed between venues. Different settlement windows on the same event. The system identified a spread that technically wasn’t one. Known edge case.
FIFA World Cup - long-dated market. Spread was valid at entry but compressed before resolution. Normal on long-horizon markets.
BTC dominance - correlated market. Spread inverted post-entry because dominance pricing moves with BTC and altcoin flows simultaneously. Expected on ratio-dependent markets.
All 3 were the smallest positions of the day. Confidence scoring working as designed. Higher conviction gets bigger sizing. Lower conviction gets minimum allocation.
Win rate is a byproduct not a target. We optimize for risk-adjusted P&L.
14 green out of 17. Losses totaled -$17.80. Wins totaled +$136.20.
You’ll follow this tutorial. You’ll get the bot running. You’ll deploy capital. And you’ll join the 92.4% of Polymarket traders who lose money.
A script built in 26 minutes will never work long term. Telling people otherwise is irresponsible.
We spent 4 months building what actually works. 12 platforms. Atomic execution. 90ms.
Save your weekend. Use infrastructure that already works.
This Post is from a suspended account. Learn more
We spent 4 months building an engine arbitraging Polymarket faster than any Claude bot can sign an order.
When the spread crosses the threshold both sides execute. 90ms. Jito atomic bundles. Both fill or nothing fills.
The orderbook view shows live depth across venues. The heatmap shows execution density across every active market in real time.
4 months of engineering. This is what 50,000 lines of code looks like when it’s running.
We made it free. Set your parameters and run.