Matt Trainer
39.8K posts
Matt Trainer
@MattTrainer
Mad Scientist, Robotics Engineer, AI Cult Leader
Laguna Beach, CA เข้าร่วม Ağustos 2007
6.7K กำลังติดตาม16K ผู้ติดตาม
one of the things that keeps me going is diorama111's micro rc cars that he handmakes from model kits
diorama111@diorama1111
#_probox 026 ヘッドライトとフロントウインカーにLEDを組み込みました。 ヘッドライトには1608mm(0603inch)の電球色(Warm White)、ウインカーには0603mm(0201inch)の黄色を入れました。
English
My latest computer vision project - Tennis🎾
Featuring:
-Player, ball, racquet detection
-Player poses
-Exclusion zones to kill false positives
-Interpolation and smoothing for clean ball trails
All running at over 30FPS!
English
@MattTrainer For your personal/research purposes you can build it for free and do all what you want - just send us videos :). If you would like to sell, we take just 2% of sales :)
English
Our blueprint for the "Optimal Universal Robot":
❌ Legs: Too complex, too expensive. ✅ Wheels: Fast, stable, 10x cheaper.
❌ Rotary Joints: Massive torque stress, custom gears. ✅ Of the shelf parts & linear actuators: High payload, extrem simplicity, off-the-shelf parts.
English
Yep. And it’s far worse than this. AI is destroying all knowledge moats.
mrdoob@mrdoob
We seem to be witnessing the end of the proprietary software model.
English
The framing 'don't predict' is shorthand for something more specific: don't treat your directional bias as certainty.
Every system generates a bias. That's unavoidable. The question is whether you're working with probabilities or guaranteed outcomes.
The issue with optimizing for 20-40R weeks:
1. Those setups are only identifiable as 20-40R after the fact. You can't systematically pre-select them beforehand.
2. Maximizing your best weeks selects for setups that worked in a specific sample. That's not edge validation, that's sample selection.
3. Mechanical consistency over hundreds of trades is what separates systematic trading from luck. The big weeks follow from that process. They don't replace it.
The gurus aren't wrong. They're just explaining it badly.
English
Every guru: "Don't predict price"
My brother, that’s the whole point of trading
I built the ShitShow model specifically to predict where price WILL go
Not where it IS
Where it WILL BE
And my students are doing 20R, 30R, 40R weeks with it
Maybe the gurus are just bad at predicting so they tell you not to do it
English
@j_intradaytrade Honestly I didn’t believe it either, until I spent 8 weeks consuming your work. Fucking epic! 🔥
English
@DabsMalone I’m starting a robotics R&D lab. What printer you recommend? Six figureish budget
English
I am looking to invest in hard tech founders just starting out.
If you’re building in:
- robotics
- energy
- manufacturing
- aerospace
- defense
- semiconductors
- industrial software
- advanced materials
I want to meet you.
Reply here or DM me with what you are building (or want to build)
English
@mattpocockuk When it hit me that my 30 years of experience and knowledge was now available to anyone with an API key. 😞
English
I'm collecting stories from folks who've had a genuine 'this changes everything' moment with AI coding.
What was yours?
English
@paulcjh_ Nice work. Anodizing can be a mf nightmare to get right.
English
I've got a good annodizing process now, sand blaster really helps. Left image is fresh part off the CNC next to a sand blasted part, then right is that part after annodizing and dying.
Biggest issue is maintaining electrical contact when it's in the acid bath as the aluminium wire I'm using also forms a non conductive film. Titanium wire is coming tomorrow to fix this
English
