Szcezin

Incepto ne Desistam
24/7 LIVE Day Trading
Watch the market move as it happens

[stupid tech takes @stupidtechtakes]

Ohhh crap

[0xRicker @0xRicker]

You could literally:

> open Polymarket
> find BTC window where delta < 0
> check EV and ask Claude when Markov persistence ≥ 0.87
> enter
> risk $90 to make $900

The math that prints $7–10K every day for this bot

Why do people still not doing this?

[Tom Dörr @tom_doerr]

Detects market regime shifts using Gaussian mixture models

https://t.co/YMZMHb99xz

[PyQuant News 🐍 @pyquantnews]

How to store billions of market prices in seconds?

ArcticDB.

• DataFrame database
• Proven at petabyte-scale
• Used by Man Group and Bloomberg

And it's free on GitHub:

[CyrilXBT @cyrilXBT]

https://t.co/ffyp3UpJMR

[断浪 @waveking1314]

一个巴西大学辍学生，搬回父母家的车库里，用开源 AI agent 搭了一个 Polymarket 交易机器人。

14 个月，赚了 $794,000。

最离谱的是，他自己几乎没写代码。

代码基本都是 Claude Opus 写的。

他用的框架叫 Hermes，是 NousResearch 做的开源 agent 框架，背后还有 Paradigm 投资。

他的钱包：
https://t.co/8RTAEU1om0
钱包名叫 Bonereaper。

40,266 笔交易。
整套系统成本大概 $10/月。

这套机器人主要做 Polymarket 上的 BTC 5 分钟涨跌市场。
一天 288 个窗口。

平均每 81 秒就会交易一次。

它的核心不是猜 BTC 接下来一定涨还是跌，而是用马尔可夫链去判断价格是否进入了某种“持续状态”。

简单说就是：

当 BTC 价格进入一个持续向上的状态时，数学上下一根继续向上的概率可能会明显提高。

如果模型判断概率大于某个门槛，比如 87%，而市场价格还没反应过来，机器人就会进场。

它吃的不是新闻。
不是情绪。
不是喊单。

而是市场还没正确反映概率的那一小段空隙。

整套系统其实很朴素。
Claude Opus 负责读信号、做判断、每天改策略。
Hermes agent 负责把这些判断变成自动执行流程。
一个便宜服务器负责 24 小时运行。
Telegram 负责把每笔交易和每日报告发到手机上。
搭建时间据说只要 30 分钟。

真正关键的地方，是它每天晚上会自动复盘。

每天午夜，Claude 会读取当天的交易记录，标记哪些交易赚钱，哪些交易亏钱，哪些状态下收益更高，然后自动调整参数。

比如昨天进场门槛是 87%。
明天可能变成 89%。
如果市场变得更难，它可能会调到 91%。
也就是说，这个机器人不是写好之后就死跑一套规则。

它会随着每一天的数据不断调整。

50 笔交易后，它会比刚启动时更聪明。

500 笔之后，它可能已经不是最开始那个机器人了。

这才是最值得关注的地方。

AI 在这里不是拿来问“今天买涨还是买跌”。

而是被放进一个完整流程里：

读数据。
做判断。
下单。
记录结果。
复盘错误。
调整参数。
第二天继续跑。

很多人还在把 Claude 当聊天工具，但有些人已经把它变成了自己的交易工程师。

你不一定要是最强程序员，也不一定要懂所有交易细节。

真正的分水岭，可能是你有没有能力把 AI 接进一个会自动执行、自动复盘、自动迭代的系统。

如果你不想自己从零搭，也可以先用这个入口观察类似高胜率钱包和自动化系统的实时操作：

https://t.co/zZRd9pX4Dg

[Robert Lauko @robert_lauko]

See the top ranked papers in AI, ML, Robotics, Quantum Physics, and more on @kurateorg. Hundreds of arXiv preprints ranked daily by scientific impact through pairwise tournaments judged by Claude, GPT, and Gemini.

[Lunar @LunarResearcher]

My ex thought I was doomscrolling crypto again because I had 14 GitHub tabs open at 1 AM.

I wasn’t even looking at prices anymore.

I was watching people build infrastructure around Polymarket faster than most startups build products.

The first repo that made me stop scrolling was:

https://t.co/AgUH1LBYQX

Official AI agents trading prediction markets with LLMs.

Not "AI assistants."

Actual autonomous systems reading news, monitoring markets and making decisions.

That already sounded absurd.

Then I kept digging.

One wallet I tracked was making entries before major sentiment swings happened.

Another was holding positions for less than 40 minutes while trading weather markets all day.

None of them looked human anymore.

No emotional entries.
No hesitation.
No scrolling CT waiting for narratives.

Just systems reacting to liquidity shifts and crowd behavior instantly.

Then I found this:

https://t.co/eqMxVwgjbK

A market-making bot managed through Google Sheets.

That’s when the entire platform started feeling different.

Polymarket doesn’t really look like a prediction market anymore.

It looks like open-source infrastructure stitched together by anonymous people who spend all night optimizing reaction speed.

One repo handles execution.

Another handles market data.

Another monitors wallets.

Another feeds agents live information.

And suddenly some of the biggest wallets on the platform start making a lot more sense.

Copytrade wallet: https://t.co/ogQCarJWLH

Most people still think they’re trading against random users clicking YES or NO.

I don’t think that’s true anymore.

[Roan @RohOnChain]

As someone who builds institutional level quant systems, this Stanford paper is the closest thing to an HFT desk I have ever seen publicly shared.

14 pages. Top Trading Strategies. Bookmark & get this, then read the article below before someone takes it down.

[CJ Zafir @cjzafir]

Do something different this weekend.

Become a PRO in AI Model Fine-tuning.

Paste this prompt in Codex/ChatGPT/Claude/Grok.

"You are an expert AI engineer and teacher.

Your job is to teach me modern LLM engineering and fine-tuning concepts from beginner to advanced level using very simple daily-life language.

Teach me step-by-step like a real mentor. Assume I am smart but new to the topic.

Foundations:

- LLM basics
- How AI models work
- Tokens
- Tokenization
- Context windows
- Embeddings
- Transformers
- Attention mechanism
- Parameters
- Training vs inference
- Open-source vs closed-source models

Datasets & Training:

- SFT datasets
- Instruction tuning
- Preference datasets
- Synthetic datasets
- Data curation
- Dataset cleaning
- Dataset formatting
- Fine-tuning basics
- Continued pretraining
- Hallucination reduction

Fine-Tuning:

- LoRA
- QLoRA
- DPO
- RLHF
- Quantization
- Model checkpoints
- Adapter tuning
- GGUF models

Inference & Optimization:

- KV cache
- Flash Attention
- Speculative decoding
- Inference optimization
- Model serving
- Batch inference
- GPU basics
- VRAM basics
- Latency vs quality tradeoffs

Local AI Ecosystem:

- llama.cpp
- Ollama
- vLLM
- MLX
- Hugging Face
- Unsloth
- Axolotl
- PEFT
- TRL library

RAG & Memory:

- RAG
- Vector databases
- Chunking
- Retrieval pipelines
- AI memory systems
- Semantic search

Agents & Workflows:

- Prompt engineering
- System prompts
- Tool calling
- Function calling
- AI agents
- Agentic workflows
- Multi-agent systems
- Browser agents

Model Types:

- VLMs
- SLMs
- Dense models
- MoE models
- Coding models
- Reasoning models

Deployment:

- Local inference
- On-device AI
- API serving
- Cloud GPUs
- Edge AI basics

Evaluation:

- AI benchmarks
- Human evals
- Cost-per-token analysis
- Speed benchmarking
- Quality benchmarking

Real-World Skills:

- Building chatbots
- Building AI copilots
- AI automation
- AI SaaS workflows
- AI coding workflows
- AI orchestration systems
- AI product thinking

Start from the absolute basics and gradually make me advanced.

Rules:

- Use simple English only
- Avoid academic jargon unless necessary
- Explain every difficult word in plain language
- Use real-world analogies and daily-life examples
- Use small code snippets when useful
- Show practical use cases
- Compare concepts side-by-side when helpful
- Teach from fundamentals first, then advanced concepts
- At the end of each topic:
- give a short summary
- give a simple mental model
- give beginner mistakes to avoid
- give a small exercise/project

I want deep understanding, not memorization."

Thank me later.

[Greg Brockman @gdb]

self improvement prompt for codex

[Mayank Agarwal 💡 @TheAIWorld22]

10 GitHub repos that print money while you sleep:

1. AutoHedge
https://t.co/RlsJ1izme4

2. Vibe-Trading
https://t.co/cs9pOAb942

3. Claude Ads
https://t.co/tkHW1VPcNg

4. Toprank
https://t.co/HmFglCATFN

5. Fincept Terminal
https://t.co/LAbe1L30pJ

6. Agentic Inbox
https://t.co/vRe1cDlHoz

7. ClawRouter
https://t.co/UkjY9xZTL1

8. Camofox Browser
https://t.co/EqZgbpicIf

9. Open Higgsfield AI
https://t.co/dP3yfyS7Vf

10. Hyperframes
https://t.co/aYggkcEQHo

[Tom Dörr @tom_doerr]

Backtest TradingView strategies inside Claude Code

https://t.co/9KLXzHW0NO

[Tom Dörr @tom_doerr]

Reverse-engineers Polymarket strategies for automated trading

https://t.co/fGwwwIUCKC

[slash1s @slash1sol]

$4,300,000 combined profit.
15 bots. All under 40 days old.

Most people will never find them.

@PredictParity lets you filter Polymarket by:

> Account age (under 40 days).
> Bot vs Human.
> PnL, win rate, volume.
> Active positions.

One filter. This leaderboard appears.
$635K. $594K. $525K. $443K. All automated.

Then you pick one. Paste the wallet. Copy every trade.

No research. No guessing. Just follow the algorithm.

Save this.

[cvxv666 @antpalkin]

AI has officially learned to predict the future - Chinese student use this and turned $2,000 into $166,000 with a single trade.

Claude as algo brain + MiroFish simulation engine allowed him to enter a Polymarket trade with just a 1.4% implied win probability.

And he won.

Wallet proof: https://t.co/72I2mr4yAi

Humans can’t process dozens of gigabytes of data in seconds - but Claude can.

Closed order book feeds from CEXs, live data from OTC trading desks, Pyth oracles - AI agent gathers all of it in one iteration.

Then this data is run through 10,000 simulations in MiroFish, every single piece of data comes alive in one unified simulation.

Claude analyzes output of the simulation and decides whether market odds are fair.

April 24: called BTC dump at 1.3% probability → $163k profit (75x).

April 21: knew ETH was pumping -> $25k profit

February 17: knew BTC would drop -> $37k profit

Successful trades with such low probabilities are not lucky guesses.

It’s elite tech + secret data + insane math working together.

Want the full guide to build the same bot? Save the post and read the article.

Want to start copying his trades right now? Use this TG copy-trading bot (US users can actually trade here, unlike the main Polymarket site): https://t.co/vbDZyVcfT3

[self.dll @seelffff]

10 repos that replace $415,000/year in enterprise software

1. supabase → replaces Firebase + Auth0 ($15K/year)
open-source backend: database, auth, storage, realtime
73,000 stars
https://t.co/Oj8tOWIYlM

2. n8n → replaces Zapier Enterprise ($50K/year)
self-hosted workflow automation. 400+ integrations.
47,000 stars
https://t.co/x924No5YNQ

3. metabase → replaces Tableau / Looker ($70K/year)
open-source BI tool. connect any database, build dashboards
38,000 stars
https://t.co/5Iw3jJnNB8

4. https://t.co/QmkaqS4QUv → replaces Calendly Enterprise ($20K/year)
open-source scheduling. full control, white label, self-host
31,000 stars
https://t.co/Vh8aFwewOk

5. openbb → replaces Bloomberg Terminal ($30K/year)
open-source investment research platform
34,000 stars
https://t.co/W9ppa5VTmU

6. mattermost → replaces Slack Enterprise ($15K/year)
self-hosted team messaging. full data ownership
30,000 stars
https://t.co/nuKaLa33OV

7. posthog → replaces Amplitude + Mixpanel ($25K/year)
open-source product analytics. session replay, feature flags
22,000 stars
https://t.co/Uk7RFE6oQZ

8. chatwoot → replaces Intercom ($40K/year)
open-source customer support platform. omnichannel inbox
22,000 stars
https://t.co/X7vT4T9ToL

9. plausible → replaces Google Analytics 360 ($150K/year)
privacy-first analytics. lightweight, GDPR compliant
20,000 stars
https://t.co/E6u1o9wrSJ

10. wallet tracking + copy trading → Kreo
tracks top Polymarket wallets. auto-copies trades in real time
https://t.co/rVKQ107tBV

total before: ~$415,000/year
total now: $0 + Kreo

like + bookmark. you'll need this.

[0x_Miko @Mikocrypto11]

有人在 Polymarket 的单笔天气盘里，把 $0.80 做到了 $400

而且他的打法非常固定：
专门狙击全球各地精确温度盘里的 1 cent YES，然后一路拿到结算

几笔已经打出来的单子很夸张：
Paris：$0.80 → $400（500x）
Singapore：$1.60 → $400（250x）
Chongqing：$1.60 → $400（250x）

重点不是某一笔翻了多少倍
而是这套东西，他已经重复了 19,000 次结算，累计做出 $13,000 利润
这已经不像普通手动交易了

市场太多
城市太多
温度太碎

能这样长期扫全球天气盘，更像是一套持续运行的自动化系统
还有一点很有意思：
这个人会频繁改用户名

这种账户最值得看的，不是它某次 500x 的截图，而是它为什么能把 1 cent 的超低价仓位，一笔一笔滚成 19,000 次 resolution 的长期利润

[Mizer @MizerXBT]

Bear Market Trading Update 5:

100% winrate
11-0 W/L

Total profit: +182.4k
Total loss: 0

Total: +$182.4k

Everything shared live (entries and closes), checkable on my X and on my Hyperliquid wallet (testing started 25/03). Will keep doing periodic updates.

https://t.co/1sA6kZSrTP

[Szcezin @seulzin]

https://t.co/8CCYLfbCVz