Raul Dias
1.3K posts
Hearing a lot of opinions from people that this keyboard was not designed for. I hope you all know how little I care.
This is MY dream keyboard, not yours.
Theo - t3.gg@theo
I spent a lot of time bugging many companies about my keyboard problems. Framework listened. They built my dream keyboard. I am so excited.
English
日本からこんにちは🇯🇵👋
世界は謎だらけです🤔
ブラジル人に質問があります🇧🇷🙋
どうしてブラジル人はポルトガル語を話すの？
ブラジル語ってないんですか？😳
日本語
The Head of Claude Code at Anthropic said he hasn’t written code by hand in months.
In 2 days he shipped 49 full features. All written 100% by AI.
He just dropped a 30 min talk on exactly how he does it.
Worth more than any $500 vibe coding course. Bookmark it:
English
use cmux, wiring codex and claude code together is a total no-brainer
just drop this exact prompt into your agent to set the foundation
"please run cmux -h, identify your surface id and one for claude, then create a suitable message protocol using xml identifiers, letting the claude know how to message back. codify this in AGENTS.md for future agent collaboration."
Numman Ali@nummanali
Codex <---> Claude Code comms with cmux New way of working is letting Codex and Claude communicate between themselves while I co-ordinate as needed. If you already have cmux installed this is a no-brainer. Send this prompt: "Please run "cmux -h", identify your surface ID and one for Claude, then create a suitable message protocol using xml identifiers, letting the Claude know how to message back. Codify this in AGENTS.md for future agent collaboration." Adjust to your liking, if you want to go more advanced ask it to add guidance for creating a new cmux tab and starting the agent directly. Then extrapolate to multiple agents ie OpenCode, Gemini etc Oh, and cmux cli can do a shit tonne more: • control all of tmux (panes, splits, hooks, buffers, copy-mode) • manage windows, workspaces, tabs, and surfaces • send keys and read screens in any pane • drive a full browser (click, type, wait, snapshot, screenshot) • show status, progress, logs, and notifications in the sidebar • live-reload markdown viewer • auto-targets your current workspace via env vars • progress bars — 0.0–1.0 progress with labels, great for long agent runs • logs — structured per-workspace log stream with levels + sources • notifications — native notify w/ title, body, subtitle • markdown viewer — open any .md in a live-reload panel • claude-hook — wire Claude session-start/stop/notification into the UI • env auto-wiring — every command defaults to the current workspace + surface
English
KDE just fixed a 21 YEAR old problem.
Plasma 6.7 is finally getting:
per-screen virtual desktops.
What this means:
• each monitor gets its own workspace
• switch desktops on one screen only
• keep the other screen static
• no more global switching across all monitors
Details:
• originally requested in 2005
• one of KDE’s oldest bugs
• now marked as fixed
• works on Wayland only
This is HUGE for multi-monitor setups.
Took two decades.
Better late than never.
English
I have finally found the most beautiful computer. New old stock from 1980. Never been turned on.
English
Essa foi a forma mais criativa de educar a população sobre impostos.
Português
Seriously, I know Netflix cannot stop ruining your childhood but let the stonk stonk please. I need stonkmaxxing if I am going to escape the permanent underclass.
English
Raul Dias retweetledi
@davepl1968 @ThePrimeagen You do realize you are talking about the same Ben, right? :)
youtu.be/l7rce6IQDWs?si…
YouTube
English
@AndrzejAwek @VivekVRao1 What I meant is that some assets have a greater oscillation during the day that does not reflect in the closing price. So depending what you look when calculating the realized vol, gives you different values.
English
@_rauldias_ @VivekVRao1 Ain't realized variance fixing this?
English
I tested a number of OHLC vol estimators to predict future close-to-close volatility. One that does well is linearly weighted Garman-Klass volatility github.com/vivek-v-rao/OH…
Vivek V Rao@VivekVRao1
I agree but have found that a realized vol computed using linearly declining weights outperforms unweighted historical volatility in predicting future volatility. To illustrate, for a 3-day weighted vol you would compute sqrt(252*(3*r[t-1]^2 + 2*r[t-2]^2 + r[t-3]^2)/6) By using linearly declining weights you can use more than 10 days while still giving more weight to more recent returns. Exponentially weighted volatility as in RiskMetrics is an alternative.
English