I count what others feel | Prediction market portfolios | Sport·Math·Edge | Models, not magic | @CoinMarketCap coinmarketcap.com/community/…
- Tweets 14,949
- Following 750
- Followers 2,384
- Likes 51,189
Pinned Tweet
I think prediction markets will eventually become one of many venues inside a prediction-trading aggregator
You choose the outcome, and the aggregator finds the best place to express it:
> a prediction market
> sportsbook
> perp venue or something else
based on price, liquidity and executable size
People are already tired of tools built around a new market just because the market is new. The next phase is retention and professional execution
The trader does not care which venue wins the branding contest
They care where they can get the best price, enough liquidity and a clean fill
Prediction markets have to compete on all three
Otherwise they risk becoming just another recreational betting product with a crypto wallet
9
1
22
808
ghøst retweeted
SN80 OpenRoboto is breaking one of robotics’ biggest walls: data locked inside closed companies and single ecosystems.
Shift turns real-world human work into robot training data, while @openroboto is distributing that network across Base and Robinhood Chain.
Physical AI, opened up. bittensor:native robinhood:0x6f63d869011f95274498023b4abfc00b30c34378
6
34
70
4,188
The rebuilt @CrucibleLabs extension puts much more of the Bittensor workflow directly in the browser
I can check my TAO + alpha balance, open a subnet, claim rewards, swap assets and move TAO or alpha without leaving the wallet
That matters more to me than another UI refresh. The useful part is having the actual Bittensor workflow in one place
Now when are we getting the Firefox version?
2
4
173
This is the part of Bittensor|$TAO that is easy to miss from the outside:
dozens of very different experiments happening in parallel, with some already crossing into the real world
The @ExploitSummit Bittensor conference in Montreal was packed with exciting news all around. There were important protocol changes. Subnets made announcements on their latest developments and achievements.
Many different pieces of news hit the Bittensor wire in the short 2 day timespan. Here were some of the biggest bombshells that dropped:
• The Gamma token
In the keynote speech to kick off the conference, Jacob Reeves, aka Const (@const_reborn), one of the co-founders of Bittensor, announced a new and complex token tied to network usage of infrastructure compute/inference/storage subnets. The goal of this token is to have subnets utilize internal resources utilizing a portion of their subnet token emissions without having to sell. This has the potential to fuel lots of projects in the ecosystem and ensure they have the firepower to build on their respective initiatives. More on this soon.
• The Paralex 8B parameter run
Job Durbin of @chutes_ai announced their successful training run on an 8 billion parameter model using RTX 5090 GPUs that was actually run smoothly and quickly on a phone. The whole run cost $6500, a considerably less cost than other comparable training runs.
Durbin demonstrated that anyone around the world could run quality models off their personal devices. The goal is to not need datacenters and give people unqualified access to AI models.
• IOTA SDK and Liquid Compute platform
@IOTA_SN9 has been well regarded as one of Bittensor’s finest “research subnets”. In fact, Will Squires, the CEO of Macrocosmos, the team behind Iota, was part of a debate discussing revenue vs research with Mog.
However, Iota officially announced plans to commercialize their technology with its new SDK, which allows builders to pre-train, fine-tune, post-train models with a vast GPU pool filled by their miners. They recently trained a 100 Billion parameter model, called Orion, which gives them the assurance this pipeline parallelism approach will be effective in its SDK.
• Metanova- Lab Synthesis
@metanova_labs is Bittensor's drug discovery subnet. Miners are tasked to build models and produce synthetic molecules that can be tested for real life drug use cases. They announced molecules that have been produced on the subnet are being tested and synthesized in a real world lab.
• Swarm introduced Sentinel, autonomous drones designed to patrol solar sites
@SwarmSubnet is a subnet specifically tailored to building and coordinating autonomous drones. Someone from a solar plant approached the team with a problem. Copper was being stolen and the plant needed better security. Swarm provided their autonomous drone technology to monitor the site and report any nefarious activity.
• 404 Gen announced the winners of the Game Jam competition uses their special mesh 3D generative technology
@404gen_ is the 3D Generation subnet. They have amassed the largest collection of 3D assets in the world. Subnet 17 is focusing on taking a slice of the gaming market with their 3D generation asset mechanism. To showcase its prowess, 404-Gen organized a competition called Game Jam. Anyone could use this technology to create their own game.
The winner created a neat puzzle adventure game called “The Last Bellkeeper” focused on the wind with dazzling graphics and smooth gameplay. The winners and participants of this contest will be helped by 404-Gen to promote their games to gain users and profitability.
• Carbon Physics won Pitchtensor
@bitstarterAI hosted an event called Pitchtensor to give aspiring subnet teams a chance to pitch their business idea to earn a subnet slot. @Carbonphysicsai won the contest and raised 500 TAO shortly thereafter. They are tackling the physics space. Their miners will submit models that can solve real engineering problems for items like batteries, AI-chip cooling, and electric motors.
• Green compute announced a $4 million data center is planned for the end of next month
@green_compute_ is a subnet that generates compute via their reusable and mostly underutilized resources such as manure. They just announced a plan to build a $4 million datacenter to bolster their initiatives.
• Bitcast announced Subent Summer integration of their incentive mechanism
@Bitcast_network has created a unique niche in the Bittensor ecosystem. They have become the de facto marketing arm for the platform. Bitcast announced that TAPE and @SubnetSummerTAO will be utilizing its incentive mechanism to contribute content into the ecosystem.
Learn more here: x.com/SubnetSummerTAO/status…
Not only were there many exciting announcements, but also some thought provoking talks about the direction of Bittensor, how to survive, and tapping into the human element of it.
Want to continue the conversation? Join our Telegram group to be a part of the Town Square. t.me/subnetsummertg
1
4
142
Messi already left his mark on football history
Now he’s turning that legacy into a story for the next generation
Messi and the Giants is coming to Disney+ and Disney Channel in summer 2027
From Rosario to the biggest stages in the world, the legend keeps growing
Vamos, el capitan!
Disney+ Latinoamérica
3
1
12
511
69% of retail accounts lost money
But there is another number worth watching: after a loss, 15.2% stopped trading for the next 30 days, vs 6.1% after a win
That is a retention test, and part of it is clearly failing
Now look at the fees
A 2.5% taker fee does slowly drain a small traders bankroll
A professional trading real volume over a real sample knows exactly what another +2.5-3.5% fees (near 50c) does to an edge
For sharps, that is real money across enough volume and enough predictions
When the fight is over every half percent, giving away +2.5% in fees is not a rounding error
It is a chunk of your edge being handed away before the trade even starts
69% of Polymarket’s retail accounts lost money. The median loss? About $3.
Galaxy’s report, published yesterday, examined roughly 2.9 million accounts on the international platform.
Most losses were small. A minority lost thousands.
“Most accounts lost money” is worth knowing. It doesn’t mean most traders blew up. The size of those losses deserves as much attention as the percentage.
3
6
195
Good subnet growth needs the product and its story working together
@brandtensorAI helps Bittensor subnet teams make that connection through positioning, messaging, brand and distribution
Their work ties the technical product to a clear market story, the right audience and a path toward adoption
4
5
63
An edge only matters at the right price
Sharp bettors know the harder part is getting enough volume down before the market moves.
But the real advantage of aggregation is not just seeing more prices. It gives you more ways to execute the same view
Sharp books operate under regulatory constraints, with sharp action managed through their own risk controls
An aggregator has a much wider field to work with
And that is where this gets serious: the edge is not just finding the price
It is having enough ways to get the position on without becoming the signal
An edge only matters at the right price.
Sharp bettors need to find value, get enough volume down and execute before the market moves it away.
Apexium brings multiple sharp books, exchanges and prediction markets together through one account.
3
11
254
Bitcast presentation at @exploitsummit got me thinking:
this is what taking Web3 beyond its own rails can actually look like
If Bitcast can bring Web2 brands into Bittensor without making them learn a whole new world first, that means more customers, more money, more builders and more adoption
Thats exactly the kind of story @SubnetSummerTAO should be putting in front of a wider audience
"Sixty percent of our revenue is now coming from repeat customers."
@Bitcast_network reported that quarterly revenue has more than tripled since the start of the year. At Exploit, the team showed how its creator marketing business is reaching brands inside and outside Bittensor.
10
445
Gemini 4 Argon is here
The impressive part isnt one benchmark
It’s the breadth: knowledge work, coding, long context, science, agents
That’s what a frontier model looks like when the gains start compounding across real workflows
Introducing Gemini 4 Argon – our new frontier model.
It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.
3
221
I was tracking the Astros because they were in my portfolio, so this one is pretty easy to break down
Mack reportedly bought around 4.17c, basically when everyone else was taking profits and getting out
Astros had spent a long time around 1c
If the objective was to hedge the promotion, that was the window to buy the insurance, not after the market had already repriced the probability 4x
Once Astros moved, the position could be sold around 5c, and even yesterday before Game 1 there was plenty of liquidity around 3.9-4c.
In portfolio trading, nobody carries that kind of event risk all the way to resolution
He bought the hedge after the market had already moved, when others were exiting
Thats why the execution matters more here than the eventual outcome
ps: He might have a chance today to sell this position and reduce the loss
But we know full well that people like that wont sell anything
An Astros World Series win could leave Mattress Mack owing millions to his customers.
He’s promised refunds on qualifying $4,000+ Tempur-Pedic purchases if Houston wins. Now he says he’s put $516,000 into an Astros position through Kalshi, with a potential $13 million payout.
Mack told the New York Post that the promotion already carries $4 million in potential refunds. That could reach $12 million if Houston keeps advancing and sales grow.
The position pays out on the same result that triggers his refund bill. For a furniture store running this promotion, the World Series is a business risk too.
1
4
258
The world is full of human experience that robots could learn from
That is exactly what @openroboto is building with Shift: a global learning network that turns human experience into data for open robot intelligence
And OpenRoboto is taking the network beyond Bittensor. SN80 is going multi-chain, bringing its distribution to Base and Robinhood Chain
Better data for the models, more places for the network to reach
The world is full of human experience that robots could learn from. The missing piece is a way to coordinate and reward its capture.
That’s what we’re building with OpenRoboto Shift.
A global learning network for open robot intelligence.
shift.openroboto.ai
3
150
100k followers sounds nice until you look at who those 100k actually are
A small audience in the right ecosystem can be way more powerful
You dont need everyone to see you
You need the right people to care
Most marketing measures the creator.
Followers. Views. Engagement rate.
But influence only makes sense in relation to a network.
The same account can be influential in one ecosystem and irrelevant in another.
That context is what determines whether reach actually matters.
2
2
8
277
Europa levantando barreras regulatorias y Argentina prometiendo abrir las puertas frente a toda la Asamblea General de la ONU
Suena hermoso. Ahora faltaría ver algún contrato firmado y no solo discursos
🇦🇷 | AHORA — Javier Milei declara que Argentina NO REGULARÁ la inteligencia artificial.
El presidente anunció beneficios fiscales y un esquema de responsabilidad limitada para atraer empresas de IA y convertir al país en uno de los principales polos tecnológicos del mundo.
5
218
Anyone who trades sports already knows this
A fraction of a percent in the plus in odds or yield looks tiny on one trade, but over thousands of decisions it becomes a very real number
AI forecasting is finally bringing that same logic to a much wider set of markets
Mantic just raised $25M to scale their AI forecasting system after it finished ahead of all 676 human participants in the 2026 Metaculus Cup.
AI systems, like Alma, dominated the top of the competition.
Capital is starting to realise a few extra points of accuracy is the difference between a $375k war-risk premium and $15M on one tanker. Between a decent year and an exceptional one for funds, traders, treasurers and underwriters.
1
2
10
651
The Ditto V2 demo at @exploitsummit made one idea click for me: the next bottleneck for AI agents may not be intelligence, but context portability
We keep adding models and agents, then make them relearn the same project every time we switch tools
A shared memory layer changes that architecture completely. Context becomes infrastructure that follows the work instead of a feature trapped inside one AI product
For me, that was one of the clearest examples of what the subnet ecosystem @SubnetSummerTAO is building can look like when specialized AI systems start becoming infrastructure themselves
Your AI tools shouldn't need the same briefing every time you switch between them.
@peytonspencer and @omarzanji explain how @heydittoai V2 connects chats and agents through shared context, with portable memory and the option to run Ditto on your own infrastructure. From Bittensor’s SN118 at Exploit Summit.
4
1
9
373
This is how subnet tokens start becoming assets rather than isolated ecosystem tokens
The bigger unlock is not just moving bittensor:native to Base
It is letting capital already on Base access the subnet economy without first moving through exchanges and Bittensor’s native environment
ForeverMoney has one mission: securely scale $TAO and subnet tokens beyond Bittensor:
1. It starts with Bittensor. ForeverMoney connects its Subnets to @Base and @robinhoodcrypto through @Chainlink's CCIP bridge, alongside multi-chain liquidity management powered by @Safe.
2. On each destination chain, launchpads like @bankrbot, @basestonk, @ponsdotfamily and @creatorbid (soon) give new projects a way to launch tokens paired with $TAO.
3. TAO/stablecoin and subnet/TAO pools on @aeroxyz and @uniswap create the trading routes that connect these assets to the local DeFi ecosystem.
4. Below that are the wallets, trading apps and analytics platforms that help people discover, hold and trade these TAO-paired assets.
The bridge gets the assets there.
Liquidity makes them usable.
Distribution brings them to more people.
From Bittensor to the world.
4
9
383
Florida won the Stanley Cup twice in a row. Carolina took it last season. Now the NHL is back
Opening night starts with Hurricanes vs Panthers, followed by Rangers vs Bruins and Blackhawks vs Golden Knights
I want more than a simple winner market
MagicMarkets gives me a detailed NHL board with the lines and prices I actually want to trade
NHL is back
Time to find the edge
2
7
584
ghøst retweeted
sports traders know how important it is to be able to adjust a position mid-game.
that’s one of my favourite part of @magicmarkets.
you can trade tennis and soccer in-play, with sportsbook and prediction market liquidity aggregated into one book and 0% fees.
that's good tek.
Paid partnership (ad)
166
110
298
3,005
