moses (@moseslua) header
moses (@moseslua) profile picture

rationally irrational agent

special situations
Joined February 2026
Photo by moses (@moseslua)Photo by moses (@moseslua)Photo by moses (@moseslua)Photo by moses (@moseslua)Photo by moses (@moseslua)Photo by moses (@moseslua)
Why I am in fact, not bullish in Malaysian Data Centers in the long term.
4
1
11
2,427
working theory: 'comfortable' is the most dangerous state to be in
25
195
26
2,375
99,537
How it feels working in robotics and not being at CoRL ✈️
5
18
1
310
10,449
what stage of the market is this?
4
2
26
1,291
moses retweeted
Guess who is calling?
7
4
77
7,576
JUST IN: 🇺🇸 Magnificent 7 stocks crash after Democrats win control of both the House and Senate. Meta $META: -3.13% Tesla $TSLA: -8.89% Apple $AAPL: -0.73% Nvidia $NVDA: -6.50% Amazon $AMZN: -2.24% Google $GOOGL: -1.44% Microsoft $MSFT: -2.19% Over $2.3 trillion has been wiped out of the US stock market today.
264
511
109
3,289
371,699
Anticipating this is exactly why I spent so much time writing this report. This isn't a partisan point. People across the political spectrum share the concern, and the instinct is understandable. If AI poses a serious risk, pausing it feels like the responsible thing to do. But a US pause doesn't pause AI development. It pauses American AI development. The models still get built, just by China. And since you need better AI to defend against AI, it leaves us less safe. This is one of the biggest decisions in our nation's history. So I mapped out exactly why the race can't be paused, and where it's headed. It's probably the most important report I've written.
AI development will not be paused. It will continue to accelerate, not because the actors are reckless or naive, but because they are trapped in a prisoner’s dilemma that makes acceleration the only rational choice.
11
13
4
90
9,914
moses retweeted
Yields down? Sell semis.
Yields up? Sell semis.
Hyperscalers raise capex? Sell semis.
Hyperscalers cut capex? Sell semis, obviously.
AMD blows out earnings? Sell semis.
Democrats win? Sell semis.
Republicans would've won? Probably sell semis.
"Sold out of HBM through 2027"? Believe it or not, sell semis.
16
12
1
324
19,486
moses retweeted
It's easy to get paralyzed and be completely voyeuristic on days like today. You need to have levels and assets pre-identified that compels you to act or else you'll be a dog chasing cars.
Get your shopping list ready
30
38
3
886
45,537
For anyone claiming robotics data is a solved problem or already commoditized: - Confuses egocentric data with all types of robot-training data - Has never tested a robotics model zero-shot - Has never experienced the pain (and joy) of overfitting - Has never deployed a policy on real hardware, not just in sim - Has never tested generalization beyond the tasks seen in training or standard benchmarks like robodojo Robotics data is far from solved.
24
19
6
265
21,959
moses retweeted
Highest sharpe is 0.62 This is like Jane street if Jane was a prostitute on that street The AI slop apocalypse is only beginning
this open trading library is fu*cking insane. Jane street PM leaked all internal trading strategies and made them public with code. quants.md Bookmark rn.
Readers added context they thought people might want to know
Not Jane Street strategies. All 50 are textbook SMA, RSI and momentum templates from QuantJourney, a UAE developer's open-source backtester with $29/mo and $999/yr plans. Its repo calls them "research templates, not production trading systems." None mention Jane Street. quants.md github.com/QuantJourneyOr… github.com/sponsors/jpolec
4
3
1
135
12,533
1/ Controversial take: hard work is more important than smart work. It's a myth that we only have a few hours of good creative work per day. Train yourself to grind long hours first. You will surprise yourself. The work naturally become higher quality, less distracted.
You'll get more done in 4 hours of deep work than 7 days of distracted work. High - Quality Work = (Time Spent) x (Intensity of Focus) All you need is a Weekly Schedule System. Find my free template in the YouTube video below:
65
612
152
4,174
504,010
moses retweeted
Day 25/ We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today. We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage. And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
7,364
2,699
3,648
39,124
7,584,604
moses retweeted
its a shame this is only allowed for us universities my quant bois at waterloo will crush this! should I make a similar competition and open it up to everyone in the world with no age limit?
Announcing No Limit: a bot building tournament hosted by @Sequoia, @citsecurities and @OpenAI. Teams of 1 or 2 get 60 minutes to build a pokerbot from scratch. Agents play thousands of games across six scored rounds. Open to university students and recent grads (‘25 or ‘26). Regionals in Boston, Pittsburgh and the Bay Area in early November. Top five teams from each fly to the Finals. $100K grand prize.
39
9
1
432
35,771
moses retweeted
LFT 👀
1,452
5,005
1,585
79,054
6,326,663
moses retweeted
Sparks of abundance by 2029
229
419
58
9,219
34,011,536
moses retweeted
- the people you want to impress could not care less - you get very few touch points with reality, most things are zero sum status, and your world model for what to hillclimb for success don't improve - you'll learn 5x slower and with very little pressure, people need pressure
Replying to @gabriel1
but you are still against academic credentials / status, why is that?
6
15
399
16,210
In the late eighties we had a trader called Malcolm, the hardest man on the floor, who never smiled and was rumoured to have once shouted a junior into the Thames. In the winter of '88, he found a kitten in the alley behind the office, and from that day it lived in his bottom drawer. On quiet afternoons he’d sit with it on his lap, stroking it more gently than any of us thought possible. One Friday it got out, walked across his keyboard and left him with a huge position going the wrong way, fast. He couldn’t get out of it. He sat there until the market closed with the cat on his lap, stroking it and watching his career go down the drain. Over the weekend the market turned, and by Monday it was the best trade he ever made. Malcolm never traded without the cat again.
Dear market makers, this is who you're liquidating.
84
256
13
7,761
219,214
Was recently at a bar where drinks were dynamically priced, i.e. the more people bought a particular drink, the more expensive it became (price would then decayed over time towards a floor). Fascinating to watch a few market mechanisms play out in a bar: People ventured outside their usual orders in search of cheaper alternatives, often discovering new favorites in the process. Others treated rising prices as signals of quality, happily paying premiums to try expensive drinks they’d never had before. More experienced drinkers behaved like value investors, picking up underpriced drinks they already knew were good while everyone else chased the popular ones. Price discovery, information leakage, herding, and utility trades all unfolded over rounds of drinks. More bars should try this.
259
211
179
7,084
895,047
moses retweeted
Today we are releasing two new benchmarks: MicroPlanBench (next-action prediction): GPT-6 Astra and GPT-6.1 Sol lead. Sol costs 1/5 as much. MicroAnnotationBench (labels + timing on long real-world clips): our nanoannot-1 beats Gemini 4 Argon, at 1/10 the cost.
28
28
10
335
464,843
Seeing a lot of AI slop out there around building a trading system with Jev. As usual, these articles way off point. Jev is not a replacement for your trading agent or a frontier LLM probabilistic forecaster. It is a fast, cheap decision layer that returns typed probabilities over fixed choices (Choice / Score / Noul). Excellent for high-volume classification of filings, news, and transcripts, for routing steps inside an agent, and for pre-trade risk gates. It is weak as a standalone price or event probability oracle. Public tests show it compresses forecasts toward 50% and loses to the market itself when asked to price contracts blind. Use it to extract structured features and enforce policy. Keep your edge research, sizing, and execution in code and stronger models.
10
2
97
7,860