Part-time Crypto researcher|
SWE in singapore, EX @tiktok @microsoft|
Major in computer eng|
Minor in psychology|
NO BS|
PolyMarket/Hyperliquid Player|
love Sex
I start to think about long MSFT, 1 of the tech companies I've worked for, here are the reasons I am bullish on it(at least in the following 1 or 2 years), even if I was bearish on it in the past.
1. LLM models are not as critical as it was in 24/25 anymore, msft was not in favored partly because they pay peanuts, don't have top-tier AI talent to build LLM model, highly rely on OpenAI, and spend tons of money on it, far more expensive than developing by themselves. Moreover, high-quality opensource models are emerging, not only narrowed the perf gap to top-tier models but also making it costs MUCH less to build a brand-new great model and AI product.
2. Currently AI products need adoption, especially in these 2 areas that are profitable: a) Agentic coding b) Business Workflows. MSFT has native advantages in content distribution and compliance business products, in particular copilot+M365 suite+teams integration. Github Copilot is catching up with claude(claude code is engineering stuff, not model actually). In CY26H1 earning call we see mass adoption and subscriptions of M365 copilot.
3. AI agent becomes an engineering thing, like agents, skills, MCP, orchestrations. If you look back the last decade, AI nowadays is much like what workloads were moving to cloud. We're talking about microservices, scaling, caching, searching.... MSFT is still actually on the table, it's still relevant.
@Autonomous_Chad Project dev/owner comes from binance lab. HYPE in Chinese community, low circulation at TGE, you can't claim all at TGE, ready to crime at launch
It's easy for them to manipulate price and fuck perp players.
THAT'S HOW BSC ECOSYSYEM WORKS.
I tried something similar months ago:
https://t.co/VFVtpbya3i
Back then I found there're some noises (at stamp 1 the price spikes and at stamp2 the price dumps), even though the chainlink follows the synthetic price, but the orderbook does not price anything.
So I pivoted to AI/ML.
I'm not quant but curious if anyone can have edge simply by using basic orderbook signal(like OBI/DEPTH) with IF/ELSE.
Today I tried to vibe code a btc-5m trading bot, I think the 1st milestone is to simulate a oracle, much faster than chainlink.
I did complete one with claude, by aggregating multiple CEXs, 5~6 seconds faster than chainlink.
Overall it looks good, Next step: improve signals.
agree, back to 60s & 70s, the semiconductor, 1st computer, spaceship, operating system, nuclear power plants, aircraft, internet, were built by Americans.
Due to globalization, jobs are outsourced for profits and supplychain specialization.
It hasn't much to do with H1B workers, even without them I still think American can build all stuff and it's only a matter of time.
H1B workers(including me) come to US just for money(paid 10x in their own country).
@TheBrancaShow China only respects these kinds of labor:
1. US/EU workers (generally white, from developed countries)
2. Asian workers in korea/japan/hongkong/singapore/taiwan.
China don't give a fuck to India and India workers.
SO China will NEVER EVER import Indian labor.
decentralization means decisions can't be made arbitrarily, entities in a 'decentralized organization' should be equvalent, individual peers that have the same power and rights on every single stuff, regardless how many of them.
I used to think the more decentralized something is, the safer it automatically becomes.
But lately I’ve been realizing… decentralized doesn’t mean unhackable.
The smart contract can be decentralized, but there are still so many layers around it keys, front ends, oracles, signers, infrastructure, and of course, humans…..
And with more exploits happening, I keep coming back to one question
If something goes wrong, where does accountability actually begin?
Maybe that’s where good regulation can have a role 👀
Not to control DeFi or slow down innovation, but to help create clearer standards around security, transparency, and accountability? 👀
Random thought ~
https://t.co/9Z1RtoQrA0
Kimi will be the winner - just look at what he did in ITALY GP - from P19 to P1, what a monster.
If he keeps top 5 starting grid I am so sure he will likely win. He's going mad now and no one can stop.
It has been around a month since I used a new strategy for poly's 5m crypto up/down market, with LightGBM + RNN+LSTM, offering 2 types of signal to predict trend on a 1 minute basis, but the result does not look good.
is my direction totally wrong or am I missing something?