I count what others feel | Prediction market portfolios | Sport·Math·Edge | Models, not magic | @CoinMarketCap https://xitter.cf/t.co/krp0rc4MKl
Joined May 2012
- Tweets15.1K
- Following750
- Followers2.4K
- Likes52K
Pinned Tweet
Prediction markets have a problem few people talk about
On smaller events the books are thin
@magicmarkets One of the best builders on Polymarket, 5th by 30d volume
>> 25+ liquidity providers
>> One order book
>> One of the best prices, from whichever source has it
>> Zero commission
NFL. NBA. MLB. Premier League. Champions League. Mexico Liga. NCAA. ATP. Esports
Thin books were never a prediction market problem
They were a fragmentation problem
Capital goes where the best price is at size
Now its all in one book
AI-generated code can work exactly as written and still do the wrong thing
I ran into this while building an AI agent: there was a contradiction in its logic that I hadn't caught myself. @prelint_ai helped me find it by checking the implementation against the intended product behavior
That's the part I care about. As agents take on more of the building, the risk isn't just broken code. It's shipping something that works technically but contradicts what the product is supposed to do
Your community makes one call. Every voter gets an NFT
@prophetthedog’s APE OR FADE turns a YES/NO vote into a collective onchain prediction
Sign in with X or connect a wallet, then vote YES (APE) or NO (FADE)
Voting is free, and Prophet creates a wallet if you need one
If the community gets it right, randomly selected voters win rewards
If it misses, you can burn your NFT in the Graveyard for a free Moonshot spin. Prizes are possible, not guaranteed
Live question:
Will PaperTrade TVL be above $35M on Tuesday, Oct 27?
Cast your vote: graveyard.prophetthedog.xyz/…
Easiest money on table if you knows what is Predictions.
So you can turn someone else’s rug into your own upside.
Predict Papertrade TVL: $100M —> $36M
On Prophet you can bet it stays ≤ $35M by Tuesday and get an NFT that shows your side.
graveyard.prophetthedog.xyz/…
> If you’re right → yummy airdrop in $PROPHET
> If you’re wrong → burn the NFT and spin the wheel for some tokens anyway.
Downside is basically zero, upside either way.
I set up my first search on @LeadpoetAI for prediction market traders and analytics teams
Instead of filtering by job title alone, I defined the audience, described the product they might need and added four buying signals: interest in analytics tools, building trading dashboards, hiring quant researchers, and discussing data or liquidity problems
The request is now sourcing 10 companies. The real test is whether those signals lead to relevant prospects, not just people who fit the profile on paper
Where do you find the best prediction-market trading tools and build a stack that fits your style?
> Platforms
> Wallet trackers
> Trading terminals
> Scanners
> Analytics
> Execution tools
611 projects. 18 categories. One map
Stop hunting tools one X post at a time
Bookmark this. @PredMarketWiki
After using @testdriveso across multiple campaigns, I can check individual post scores, performance bonuses and whether my review meets the brief
That covers the performance of each post, but I’d also like a clearer picture of my position in the creator community
One feature I’d add is a shareable Creator Card, similar to Stitch3’s scorecard. It could show your rank, strongest category, influence and top connections in one place
Stitch3’s card makes your position in its ecosystem visible at a glance
TestDrive could do the same for its creators, turning campaign activity into something easier to understand and share
UFC 333 tomorrow, Merab vs Yan
Heres how Thumper from @magicmarkets works
Same fight, three prices
>> 59c on the Polymarket ticket
>> 60.2c with fee
>> 58.5c on the MagicMarkets
For a trader the real gap is 1.7c: break even drops from 60.2% to 58.5% and $100 buys about 4.8 more shares
Cheaper entry is edge you dont need to prediction
1/ i built a chrome extension for @magicmarkets that shows you when polymarket isn't the cheapest place to take a sports position
it's called thumper. Free to use, no account required 🧵
Tried @excalidraw and went straight to drawing
The interface is simple, and Ctrl+Z works for undoing mistakes
Two small UX issues: the language switches based on your IP without asking, and I couldn't find a visible Undo button
Let users choose their language on entry and make Undo easier to spot
ghøst retweeted
Insurance claims run on photos. So what happens when the photo is lying?
I built my #PerturbVideo submission around that question, using Angle 3: The claim that never happened
A real example was reported by The Guardian in May 2024. A van photo was taken from a tradesman's social media, edited to show a cracked bumper, and sent to UK insurer LV= with a fake invoice for more than £1,000. LV= had concerns and passed it to its fraud team. Allianz UK said claims involving app-edited images, videos and documents rose 300% between 2021-22 and 2022-23
My video then plays out a clearly labeled, made-up scenario where an automated claims model approves a doctored photo. A few tuned pixels that a person would never notice can leave a model wrong and still confident
The cost doesn't just vanish. Insurers cover the payout, and honest policyholders feel it in their premiums. The Coalition Against Insurance Fraud estimates US insurance fraud at $308.6 billion a year in its 2022 study, which is an industry figure and not something Perturb measured
Perturb @perturbaix flips the usual order and breaks the model first, before a fraudster gets the chance. On Bittensor Subnet 26, miners search for the smallest change that makes a model's answer flip. Validators confirm the flip is real and the change stays invisible, copied work gets no reward, and whoever finds the best attacks earns the most $TAO
For the team behind the model, that testing turns into a robustness score, a vulnerability heatmap, verified attack data to retrain on and an on-chain certificate of testing. Details live at perturbai.io
As of now, image-model testing is live and the network has been running since May 2026. Text, LLM and multimodal attacks are still on the roadmap, and November presales are a target rather than a done deal
ghøst retweeted
The price on screen isnt always the price you pay
On Liverpool vs Villarreal, Polymarkets draw costs 21.8c all-in, while MagicMarkets offers 20.8c
Thats a $12 difference in potential payout on a $100 trade
Across 100 comparable trades, thats $1,200. Same stake, different execution price
Thumper is a free Chrome extension that puts @magicmarkets prices next to Polymarket sports markets, so you can compare the real cost before trading
Check the price
Then take the trade
1/ i built a chrome extension for @magicmarkets that shows you when polymarket isn't the cheapest place to take a sports position
it's called thumper. Free to use, no account required 🧵
AI doesnt operate in a lab
> Lighting changes
> Angles change
> Conditions change
Real-world inputs dont care how well your model scored on a clean benchmark
@perturbaix tests what happens when the inputs stop being predictable, finding weaknesses before they become real-world failures
Angle 1: The Taped Sign
A tiny piece of tape
A very different reading
In 2020, McAfee showed that tape could make an older vision system read 35 mph as 85 mph
For automakers, a misread sign can become a safety and liability risk. @perturbaix helps find these weaknesses before deployment
On Bittensor Subnet 26, miners compete to find tiny changes that fool AI models
Validators check that each attack works and remains imperceptible. Copies earn nothing
Customers get robustness scores, vulnerability heatmaps, and verified data for retraining
Image-model testing is live
Text, LLM, and multimodal attacks are still on the roadmap.
Find weaknesses before attackers do
#PerturbVideo
Tested @OpenBot_ on Windows with a free OpenCode model, no API key needed
I set up an agent to check Snitch3, list its key details and save them to a Markdown file
It got the job done in about three minutes, and the details matched the site
The rough part was watching it work: the chat showed broken fragments of text instead of a clear progress update
I ran the free Agentsy Report scan on Probly
The AI description was accurate: a prediction market platform covering politics, sports, crypto and finance
But PROBLY appeared in only 6 of 14 answers and its own site was cited in just 2
The clearest finding: across 8 answers to unbranded discovery questions, PROBLY appeared zero times
AI can describe a product correctly and still miss it when people search by category
Builder codes turn perps into a distribution layer for Polymarket
For terminals, the edge is capturing flow across event-driven and continuous markets under one roof
Glad to see you taking part in TestDrive!
I been doing the tests myself, and its nice to see ppl sharing their experiences with different tools
Curious to see what you end up testing and what you think of it
I've tested a lot of platforms, so I joined a rewarded Test Drive from @testdriveso to check out the Test Track.
The UI is clean, everything feels intuitive, and signing up with X took 2 seconds. I’d just like to understand the leaderboard criteria better. Really enjoyed the platform and would love to join more test drives in the future!
Sharps figured this out a long time ago
Nobody serious takes a first deposit bonus, because the real bonus is a sharper price on every single trade
Same logic here: a fee waiver is a one time discount, no fees on NBA is a lower cost basis on every position you ever take
The real edge in funding arbitrage isnt finding the highest rate
Its finding a spread that survives fees, slippage, margin requirements and funding reversals
40% annualized rate means little if you can only capture it for a few hours or need twice the capital to keep both legs safe
Net return on deployed capital matters more than the headline APY
Top 3 Funding Arbitrage Strategies
Bookmark this so you can test different methods
You can already do proper arbitrage inside one venue or across a few
1. Spot-Perp (classic)
One of the most common strategies in my book
When funding is positive (longs pay shorts): buy spot + short equal notional in the perpetual
Total cost of the hedge < the funding you collect
You need the same notional on both legs so the price movement cancels
Works cleanest on major coins with deep books
2. Perp-perp funding arb
Same asset, different funding rates on two exchanges
Long the venue with the lowest (or negative) funding + short the one with the highest
You collect the spread while price risk nets to zero
We used this heavily when one venue stayed elevated while the other mean-reverted
3. Lending + Short Perp
Deposit the asset into lending (earn APY) and short the same size in the perp
You collect two yields: lending interest + funding when it’s positive
The position stays delta-neutral (price movement cancels out)
Works when both rates are healthy and stable
Main cost is lower capital efficiency and extra risk: smart-contract on the lending side plus liquidation risk on the short
Exit if funding flips or the lending APY collapses
I'd nominate @desearch_ai (SN22) for a @testdriveso
It brings web and X search into the Bittensor ecosystem
and I test it on a practical research task:
can it find recent claims about a subnet and surface the original sources so you can verify them?
Search quality matters less if you still have to hunt down every source yourself
The price on screen isnt always the price you pay
On Liverpool vs Villarreal, Polymarkets draw costs 21.8c all-in, while MagicMarkets offers 20.8c
Thats a $12 difference in potential payout on a $100 trade
Across 100 comparable trades, thats $1,200. Same stake, different execution price
Thumper is a free Chrome extension that puts @magicmarkets prices next to Polymarket sports markets, so you can compare the real cost before trading
Check the price
Then take the trade
1/ i built a chrome extension for @magicmarkets that shows you when polymarket isn't the cheapest place to take a sports position
it's called thumper. Free to use, no account required 🧵
Table 6: in the sim the edge leaks with time and size
100 ms already costs 8.6 points of profit
Half participation keeps 56.5% of guaranteed profit at a higher return on capital
Orders with a loss climb from 14.2% at 100 ms to 22.1% at 1 second
The climb is execution, not variance
Variance you accept, errors you engineer out
A one-second delay left 22.1% of simulated Polymarket arbitrage orders losing money.
Researchers replayed full order books across linked contracts, accounting for spreads and fees. Add that delay, and only 54.3% of orders filled completely.
The problem: the profit guarantee depends on getting the whole basket. If some legs fill and others disappear, you’re left holding an exposed position.
The delayed strategy still retained 79.3% of its baseline profit overall. But a profitable strategy can contain plenty of losing executions.
These are historical simulations using June–August 2026 data. A useful reminder that spotting contradictory prices and actually collecting the difference are separate jobs.
