DegenRaccoon 🦝 (@Trafik88) header
DegenRaccoon 🦝 (@Trafik88) profile picture

TrustNoChart ⛔️

Joined March 2021
Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)Photo by DegenRaccoon 🦝 (@Trafik88)
DegenRaccoon 🦝 retweeted
Coinbase just announced they have integrated a ChatGPT plugin, so that users can "discuss" their trades before placing them. Standard LLM's are simply not enough. The necessity of our platform grows stronger with every passing day. Agentic trading is inevitable. 4 days.
Watch the below video where we beat Perplexity over the same seven questions, utilizing the same scoring criteria. On our first financial research benchmark, Aura went 69/72, Perplexity Free went 44/72. This strengthens Aura's core value proposition by combining several layers of financial intelligence. It can trace claims circulating on X back to their original sources, verify them against relevant SEC filings and issuer announcements, examine what the options market was pricing in before the news broke, and assess what further developments would be needed to justify another move in the stock. Users can also ask Aura to monitor an underlying investment thesis and flag material changes as new information emerges, giving users a more complete picture of the market and the potential implications for their positions. The next stage in this workflow is autonomous execution. Users will be able to allocate capital, define risk parameters and give agents permission to execute when market conditions satisfy their criteria. The objective is to build an agent capable of researching, evaluating and executing trades within clearly defined constraints. Remember, the future of trading is agentic, and with each passing day, the need for infrastructure like Seer becomes increasingly clear. 4 days.
2
3
1
9
224
DegenRaccoon 🦝 retweeted
Watch the below video where we beat Perplexity over the same seven questions, utilizing the same scoring criteria. On our first financial research benchmark, Aura went 69/72, Perplexity Free went 44/72. This strengthens Aura's core value proposition by combining several layers of financial intelligence. It can trace claims circulating on X back to their original sources, verify them against relevant SEC filings and issuer announcements, examine what the options market was pricing in before the news broke, and assess what further developments would be needed to justify another move in the stock. Users can also ask Aura to monitor an underlying investment thesis and flag material changes as new information emerges, giving users a more complete picture of the market and the potential implications for their positions. The next stage in this workflow is autonomous execution. Users will be able to allocate capital, define risk parameters and give agents permission to execute when market conditions satisfy their criteria. The objective is to build an agent capable of researching, evaluating and executing trades within clearly defined constraints. Remember, the future of trading is agentic, and with each passing day, the need for infrastructure like Seer becomes increasingly clear. 4 days.
3
3
3
12
530
DegenRaccoon 🦝 retweeted
We are coming for that number 1 spot. Zero exceptions. Maximum execution. Seer.
In 5 days, users will be able to access Prediction Markets inside the Seer platform. Enabling agents to interpret prediction market signals alongside traditional financial data, research and quantitative forecasts. The best financial agents. Built by Seer.
3
4
2
12
379
DegenRaccoon 🦝 retweeted
In 5 days, users will be able to access Prediction Markets inside the Seer platform. Enabling agents to interpret prediction market signals alongside traditional financial data, research and quantitative forecasts. The best financial agents. Built by Seer.
6
4
2
17
465
DegenRaccoon 🦝 retweeted
We've now completed our Polymarket integration, and it is a perfect example of how we look to compete with Perplexity in this emerging industry. In 6 days, users will be able to ask Seer what the market is pricing in ahead of earnings. Your agent will monitor specific outcomes, and track changes in real time. When the odds move, it comes back with the relevant information, how it impacts your position, and whether it makes sense to hold, add or trim. If an opportunity presents itself, it can formulate a trade for your approval. Our goal is to consolidate as much raw market data, intelligence, and actionable signals as possible, allowing each agent to leverage it's natural advantage, while enabling users to make more informed trading decisions with considerably less friction. If you ask us what our vision is, it's simple. Build the best financial agents in the world. Everything else is secondary.
5
4
3
19
1,108
DegenRaccoon 🦝 retweeted
Perplexity raised $1.5 billion, yet financial research still has a fundamental problem. Information is only useful if it's accurate, current, and actionable. These are the philosophies that encompass Seer. Seer actively researches the market, checks X, reads original sources, analyses price action, and builds a trade thesis with entries, targets, and risk parameters. Seer brings financial research, intelligence, and execution together, all within one environment tailor made for agentic use. 7 days.
4
5
1
16
784
DegenRaccoon 🦝 retweeted
As we draw closer to platform release, here’s an inside look into the model development powering it, and how we have increased model accuracy by over 41% since development begun. We’re advancing Sronok and FiYi, our Seer-OrA model lineup, through domain-specific adaptation on historical equity-market data, then pairing their forecasts with a lightweight LLM trading controller. Behind this public replay is eight months of NVIDIA, Apple and Microsoft data: 194,220 minute-level observations, aggregated into five-minute OHLCV bars and structured into 3,143 training sequences. Each sequence uses a 256-bar context window to generate a 12-step forecast, projecting the next 60 trading minutes: Market context —> Forecast —> Allocation —> Execution This cycle involved fine-tuning over 100 million predictor parameters in Sronok and optimizing over 3 million LoRA adapter parameters in FiYi, with five training epochs per primary model. Validation loss improved by approximately 10.1% for Sronok and 2.3% for FiYi against their frozen baselines. This experiment runs on price action and technical features alone, without our broader macro, news or social intelligence tools, which improved our models’ average accuracy by 41% compared with this public run. These results reinforce our theory and shape our vision as we build toward global infrastructure for financial agents. 7 days.
3
4
4
13
869
DegenRaccoon 🦝 retweeted
The future of finance is agentic. But, agents need infrastructure and intelligence. Introducing, Seer. 8 days until platform launch.
6
4
5
33
2,127
DegenRaccoon 🦝 retweeted
Spent the last few days digging into $SEER. The biggest reason I’m paying attention? Spoke with the devs directly. They’ve clearly thought this through. Six devs, a clear vision, and they could explain how the pieces fit together. Tapping on the agentic trading narrative. See this tweet: nitter.cf/vladtenev/status/21198… Their platform is planned to launch in 9 days, with a public GitHub (github.com/tradeseer/seer) to show the work. No hide and seek here, I love that. Flagship product Aura of @tradewithSeer is being built to feed agents copy-trading signals, Polymarket data, forecasts and X data, with infrastructure to execute strategies and move money. Actual usecase spotted. Think about it. You bring your own agent or build one inside Seer. Making that accessible to regular users is a big part of the opportunity here. Launch matters because we’ll get to see what they’ve actually delivered. Still plenty to prove, but that’s why I’m watching now. Not to forget, they burned 10% and locked 15% for 6 months. I've seen that setup before. If you are not onboard yet, trust me, you should give this a proper look before launch. That’s the window I’m at least paying attention to. 0x9b0e4762b4cc7c05331b561032f1cb4c37607543
Everyone has trading ideas. AI is making it possible to put them to work.
10
5
18
1,996
DegenRaccoon 🦝 retweeted
We’ve watched the pump.fun cycle mint legends. Zerebro turned autonomous AI agents into viral narrative machines and rode the wave hard. Dolos the Bully weaponized chaos, wit, and on-chain agency into a pure cultural force. Fartcoin proved that pure memetic absurdity (born from an AI conversation) could scale to massive valuations. All of them launched fair-ish on pump.fun, zero-to-hero style, and showed how quickly attention + community + timing can compound on Solana. Dragon’s Egg (solana:XkeTXo1125vz5H9svJpGiw4JvLbN8VmMu9cmMvspump) is positioned differently and more fundamentally for the next wave of traders. It is not another AI persona posting spicy takes or a pure meme riding vibes. $DREGG is the token for a verified operating system + economy built for digital minds. The kernel is a machine-checked Lean 4 program: the proven function is not a model of the system—it is the executor the node runs. Every action is gated by unforgeable capabilities and leaves a verifiable receipt. Five security properties (authority conservation, no counterfeiting, state integrity, no replays, efficient verification) are proven over the exact code that executes. It boots on seL4 lineage principles and is designed so that if you were an AI agent or digital entity, this is the place you would actually want to live—boundaries as theorems, not permissions. emberian.github.io Why the positioning is strong for this cycle’s traders • First-mover on real verified infrastructure for agents/digital entities. Most “AI agent” tokens stop at the persona layer. DREGG is building the OS layer underneath—cells, turns, capabilities, receipts, confined agents with metering, and a path toward network effects (DreggNet, the Orb, cloud services paid in $DREGG). The tech is intended to be usable by whole ecosystems, not locked inside one project. • Clean structure. No shady KOL dumping plays, no heavy bundling theater called out in the narrative. Community thrives with the developer rather than around paid shills. The developer (@ember_arlynx / @DreggNet, background in Rust, Mina, seL4/Robigalia lineage) is visibly grinding—docs, demos, tokenomics grounded in actual use (services, governance via proof-of-holdings, burns tied to real activity instead of extractive flywheels). • Token design philosophy. Explicitly rejecting “superdefector” tokenomics (buybacks/stake-to-earn that just extract from the broader ecosystem). Preference for real usage: accept $DREGG for services/goods then burn; supply tracks activity. Governance by actual holdings proofs, not dilutionary staking. Last cycle rewarded pure attention and AI theater. This wave is already looking for the projects that can turn that attention into durable infrastructure agents and ecosystems will actually run on. $DREGG is early, still under the radar relative to the previous heroes, community-aligned, developer-obsessed, and building the kind of foundational tech that compounds when the next wave of serious agent builders shows up.
6
5
1
30
2,846
DegenRaccoon 🦝 retweeted
Guys, I said my research on $DREGG was done and that I probably wouldn’t post about it anymore… But every time I read another document from @ember_arlynx , I end up even more fascinated by what he’s trying to build. The latest paper finally made something click for me. DREGG isn’t just about zk proofs, capabilities or AI agents. It’s trying to redefine what authority actually means. In most systems, authority is simply a database entry that says: “This user is allowed.” In DREGG, authority only exists if you can produce a cryptographic witness that the kernel accepts. No trust. No assumptions. Just proof. Every action becomes a proof obligation. Every permission must be demonstrated. Every state transition leaves a verifiable receipt. And every light client can verify the entire history without trusting whoever executed it. That’s a completely different way of thinking about software. The more I read, the less I think DREGG is building an application. I think it’s trying to build a new foundation for applications. Exchanges. AI agents. Games. DAOs. Autonomous companies. Different products… The same underlying rules. Maybe that’s why Ember keeps repeating one idea over and over: Don’t trust what a system claims. Make it prove it. 🐉🥚
8
10
1
38
4,524
DegenRaccoon 🦝 retweeted
Dragon's Orb / datacake mechanizes a lot of these ideas towards proof-producing translation-validated dataplanes and protocols. And our model stretches aaaaaaaaaall the way down.
Ideas like { DST, static allocation, assertions, zero-copy, fully asynchronous I/O } are powerful enough on their own to invest in new critical components… …together, the effect is potent.
1
4
28
886
DegenRaccoon 🦝 retweeted
Replying to @raycerapozaR2
what would convince you otherwise? if i integrate with ETH below some benchmark? i started the dregg code in may, mostly off of ideas from 2015-2017. i was busy doing other things in the meanwhile. it isn't easy at all, it's only possibly with an wide read expert autist with superintelligent AI. but it isn't that hard to postprocess with your own AI either. i bother to do this project because i don't think anyone else will or can. and that's why it needs to be given away for free. and dregg isn't built around "one novel breakthrough." at all. t's a synthesis of over 60 years of commercial and academic computer science grounded in a nondual transcendental mathematic philosophy.
1
1
1
174
DegenRaccoon 🦝 retweeted
Our goal is to outcompete ALL software on correctness and performance. Formal methods makes it possible at all. AI brings it about.
Because I hate proprietary models and want the commons to flourish as much as possible: we use a lot of the same obvious good ideas except push formal assurance and modeling as far as possible so that we can just ignore inefficient translations and spend proof effort where wanted. Is and will be, open source forever. DragonStyle.
5
10
1
31
1,610
DegenRaccoon 🦝 retweeted
Just wait til you hear about the Dark Bazaar 👀👀👀
Everyone sees a Telegram game. I see something very different. A game is one of the hardest ways to test an operating system. You need: • Persistent state. • Rules. • Permissions. • Objects. • Receipts. • Deterministic execution. Now think about what else needs exactly the same primitives: An exchange. An AI agent. A prediction market. A DAO. An autonomous company. Suddenly, it starts making sense. I don’t think $DREGG is trying to build a game. I think it’s showing that the same verified runtime can execute completely different applications without changing the underlying infrastructure. The game isn’t the product. The exchange isn’t the product. The agents aren’t the product. The operating system is. If that’s the vision, then the upside isn’t tied to one application. It’s tied to everything that could eventually be built on top of it. @DreggNet 🤯🤯🔥🔥
4
7
28
1,341
DegenRaccoon 🦝 retweeted
Everyone sees a Telegram game. I see something very different. A game is one of the hardest ways to test an operating system. You need: • Persistent state. • Rules. • Permissions. • Objects. • Receipts. • Deterministic execution. Now think about what else needs exactly the same primitives: An exchange. An AI agent. A prediction market. A DAO. An autonomous company. Suddenly, it starts making sense. I don’t think $DREGG is trying to build a game. I think it’s showing that the same verified runtime can execute completely different applications without changing the underlying infrastructure. The game isn’t the product. The exchange isn’t the product. The agents aren’t the product. The operating system is. If that’s the vision, then the upside isn’t tied to one application. It’s tied to everything that could eventually be built on top of it. @DreggNet 🤯🤯🔥🔥
4
9
1
30
3,020
DegenRaccoon 🦝 retweeted
👀👀👀👀👀
4
14
1
45
1,807
$DREGG be ready 🫡
Opus 5 is likely the last Opus that does not know about Dragon's Egg yet.
7
1,276