Offshore
Photo
God of Prompt
RT @rryssf_: psychology solved the ai memory problem decades ago. we just haven't been reading the right papers.
your identity isn't something you have. it's something you construct. constantly. from autobiographical memory, emotional experience, and narrative coherence.
Martin Conway's Self-Memory System (2000, 2005) showed that memories aren't stored like video recordings.
they're reconstructed every time you access them, assembled from fragments across different neural systems. and the relationship is bidirectional: your memories constrain who you can plausibly be, but your current self-concept also reshapes how you remember. memory is continuously edited to align with your current goals and self-images. this isn't a bug. it's the architecture.
not all memories contribute equally. Rathbone et al. (2008) showed autobiographical memories cluster disproportionately around ages 10-30, the "reminiscence bump," because that's when your core self-images form.
you don't remember your life randomly. you remember the transitions. the moments you became someone new. Madan (2024) takes it further: combined with Episodic Future Thinking, this means identity isn't just backward-looking. it's predictive. you use who you were to project who you might become. memory doesn't just record the past. it generates the future self.
if memory constructs identity, destroying memory should destroy identity. it does. Clive Wearing, a British musicologist who suffered brain damage in 1985, lost the ability to form new memories. his memory resets every 30 seconds. he writes in his diary: "Now I am truly awake for the first time." crosses it out. writes it again minutes later.
but two things survived: his ability to play piano (procedural memory, stored in cerebellum, not the damaged hippocampus) and his emotional bond with his wife. every time she enters the room, he greets her with overwhelming joy. as if reunited after years. every single time. episodic memory is fragile and localized.
emotional memory is distributed widely and survives damage that obliterates everything else.
Antonio Damasio's Somatic Marker Hypothesis destroyed the Western tradition of separating reason from emotion.
emotions aren't obstacles to rational decisions. they're prerequisites.
when you face a decision, your brain reactivates physiological states from past outcomes of similar decisions. gut reactions. subtle shifts in heart rate. these "somatic markers" bias cognition before conscious deliberation begins.
the Iowa Gambling Task proved it: normal participants develop a "hunch" about dangerous card decks 10-15 trials before conscious awareness catches up. their skin conductance spikes before reaching for a bad deck. the body knows before the mind knows. patients with ventromedial prefrontal cortex damage understand the math perfectly when told. but keep choosing the bad decks anyway. their somatic markers are gone. without the emotional signal, raw reasoning isn't enough.
Overskeid (2020) argues Damasio undersold his own theory: emotions may be the substrate upon which all voluntary action is built.
put the threads together. Conway: memory is organized around self-relevant goals. Damasio: emotion makes memories actionable. Rathbone: memories cluster around identity transitions. Bruner: narrative is the glue.
identity = memories organized by emotional significance, structured around self-images, continuously reconstructed to maintain narrative coherence. now look at ai agent memory and tell me what's missing.
current architectures all fail for the same reason: they treat memory as storage, not identity construction. vector databases (RAG) are flat embedding space with no hierarchy, no emotional weighting, no goal-filtering. past 10k documents, semantic search becomes a coin flip. conversation summaries compress your autobiography into a one-paragraph bio. key-value stores reduce identity to a lookup table. episodic buffers give you a 30-second memory sp[...]
RT @rryssf_: psychology solved the ai memory problem decades ago. we just haven't been reading the right papers.
your identity isn't something you have. it's something you construct. constantly. from autobiographical memory, emotional experience, and narrative coherence.
Martin Conway's Self-Memory System (2000, 2005) showed that memories aren't stored like video recordings.
they're reconstructed every time you access them, assembled from fragments across different neural systems. and the relationship is bidirectional: your memories constrain who you can plausibly be, but your current self-concept also reshapes how you remember. memory is continuously edited to align with your current goals and self-images. this isn't a bug. it's the architecture.
not all memories contribute equally. Rathbone et al. (2008) showed autobiographical memories cluster disproportionately around ages 10-30, the "reminiscence bump," because that's when your core self-images form.
you don't remember your life randomly. you remember the transitions. the moments you became someone new. Madan (2024) takes it further: combined with Episodic Future Thinking, this means identity isn't just backward-looking. it's predictive. you use who you were to project who you might become. memory doesn't just record the past. it generates the future self.
if memory constructs identity, destroying memory should destroy identity. it does. Clive Wearing, a British musicologist who suffered brain damage in 1985, lost the ability to form new memories. his memory resets every 30 seconds. he writes in his diary: "Now I am truly awake for the first time." crosses it out. writes it again minutes later.
but two things survived: his ability to play piano (procedural memory, stored in cerebellum, not the damaged hippocampus) and his emotional bond with his wife. every time she enters the room, he greets her with overwhelming joy. as if reunited after years. every single time. episodic memory is fragile and localized.
emotional memory is distributed widely and survives damage that obliterates everything else.
Antonio Damasio's Somatic Marker Hypothesis destroyed the Western tradition of separating reason from emotion.
emotions aren't obstacles to rational decisions. they're prerequisites.
when you face a decision, your brain reactivates physiological states from past outcomes of similar decisions. gut reactions. subtle shifts in heart rate. these "somatic markers" bias cognition before conscious deliberation begins.
the Iowa Gambling Task proved it: normal participants develop a "hunch" about dangerous card decks 10-15 trials before conscious awareness catches up. their skin conductance spikes before reaching for a bad deck. the body knows before the mind knows. patients with ventromedial prefrontal cortex damage understand the math perfectly when told. but keep choosing the bad decks anyway. their somatic markers are gone. without the emotional signal, raw reasoning isn't enough.
Overskeid (2020) argues Damasio undersold his own theory: emotions may be the substrate upon which all voluntary action is built.
put the threads together. Conway: memory is organized around self-relevant goals. Damasio: emotion makes memories actionable. Rathbone: memories cluster around identity transitions. Bruner: narrative is the glue.
identity = memories organized by emotional significance, structured around self-images, continuously reconstructed to maintain narrative coherence. now look at ai agent memory and tell me what's missing.
current architectures all fail for the same reason: they treat memory as storage, not identity construction. vector databases (RAG) are flat embedding space with no hierarchy, no emotional weighting, no goal-filtering. past 10k documents, semantic search becomes a coin flip. conversation summaries compress your autobiography into a one-paragraph bio. key-value stores reduce identity to a lookup table. episodic buffers give you a 30-second memory sp[...]
Offshore
God of Prompt RT @rryssf_: psychology solved the ai memory problem decades ago. we just haven't been reading the right papers. your identity isn't something you have. it's something you construct. constantly. from autobiographical memory, emotional experience…
an, which as the Wearing case shows, is enough to operate moment-to-moment but not enough to construct identity.
five principles from psychology that ai memory lacks.
first, hierarchical temporal organization (Conway): human memory narrows by life period, then event type, then specific details. ai memory is flat, every fragment at the same level, brute-force search across everything. fix: interaction epochs, recurring themes, specific exchanges, retrieval descends the hierarchy.
second, goal-relevant filtering (Conway's "working self"): your brain retrieves memories relevant to current goals, not whatever's closest in embedding space. fix: a dynamic representation of current goals and task context that gates retrieval.
third, emotional weighting (Damasio): emotionally significant experiences encode deeper and retrieve faster. ai agents store frustrated conversations with the same weight as routine queries. fix: sentiment-scored metadata on memory nodes that biases future behavior.
fourth, narrative coherence (Bruner): humans organize memories into a story maintaining consistent self across time. ai agents have zero narrative, each interaction exists independently. fix: a narrative layer synthesizing memories into a relational story that influences responses.
fifth, co-emergent self-model (Klein & Nichols): human identity and memory bootstrap each other through a feedback loop. ai agents have no self-model that evolves. fix: not just "what I know about this user" but "who I am in this relationship."
the fundamental problem isn't technical. it's conceptual. we've been modeling agent memory on databases. store, retrieve, done. but human memory is an identity construction system. it builds who you are, weights what matters, forgets what doesn't serve the current self, rewrites the narrative to maintain coherence. the paradigm shift: stop building agent memory as a retrieval system. start building it as an identity system.
every component has engineering analogs that already exist.
hierarchical memory = graph databases with temporal clustering.
emotional weighting = sentiment-scored metadata.
goal-relevant filtering = attention mechanisms conditioned on task state.
narrative coherence = periodic summarization with consistency constraints.
self-model bootstrapping = meta-learning loops on interaction history.
the pieces are there. what's missing is the conceptual framework to assemble them. psychology provides that framework.
the path forward isn't better embeddings or bigger context windows. it's looking inward. Conway showed memory is organized by the self, for the self. Damasio showed emotion is the guidance system. Rathbone showed memories cluster around identity transitions. Bruner showed narrative holds it together.
Klein and Nichols showed self and memory bootstrap each other into existence. if we're serious about building agents with functional memory, we should stop reading database architecture papers and start reading psychology journals.
tweet
five principles from psychology that ai memory lacks.
first, hierarchical temporal organization (Conway): human memory narrows by life period, then event type, then specific details. ai memory is flat, every fragment at the same level, brute-force search across everything. fix: interaction epochs, recurring themes, specific exchanges, retrieval descends the hierarchy.
second, goal-relevant filtering (Conway's "working self"): your brain retrieves memories relevant to current goals, not whatever's closest in embedding space. fix: a dynamic representation of current goals and task context that gates retrieval.
third, emotional weighting (Damasio): emotionally significant experiences encode deeper and retrieve faster. ai agents store frustrated conversations with the same weight as routine queries. fix: sentiment-scored metadata on memory nodes that biases future behavior.
fourth, narrative coherence (Bruner): humans organize memories into a story maintaining consistent self across time. ai agents have zero narrative, each interaction exists independently. fix: a narrative layer synthesizing memories into a relational story that influences responses.
fifth, co-emergent self-model (Klein & Nichols): human identity and memory bootstrap each other through a feedback loop. ai agents have no self-model that evolves. fix: not just "what I know about this user" but "who I am in this relationship."
the fundamental problem isn't technical. it's conceptual. we've been modeling agent memory on databases. store, retrieve, done. but human memory is an identity construction system. it builds who you are, weights what matters, forgets what doesn't serve the current self, rewrites the narrative to maintain coherence. the paradigm shift: stop building agent memory as a retrieval system. start building it as an identity system.
every component has engineering analogs that already exist.
hierarchical memory = graph databases with temporal clustering.
emotional weighting = sentiment-scored metadata.
goal-relevant filtering = attention mechanisms conditioned on task state.
narrative coherence = periodic summarization with consistency constraints.
self-model bootstrapping = meta-learning loops on interaction history.
the pieces are there. what's missing is the conceptual framework to assemble them. psychology provides that framework.
the path forward isn't better embeddings or bigger context windows. it's looking inward. Conway showed memory is organized by the self, for the self. Damasio showed emotion is the guidance system. Rathbone showed memories cluster around identity transitions. Bruner showed narrative holds it together.
Klein and Nichols showed self and memory bootstrap each other into existence. if we're serious about building agents with functional memory, we should stop reading database architecture papers and start reading psychology journals.
tweet
Offshore
Photo
Michael Fritzell (Asian Century Stocks)
RT @ConsensusGurus: Gotta love Cluseau https://t.co/bJ1U2eWVeb
tweet
RT @ConsensusGurus: Gotta love Cluseau https://t.co/bJ1U2eWVeb
tweet
Offshore
Photo
Illiquid
This is like finding out Riedel has a photonics segment.
tweet
This is like finding out Riedel has a photonics segment.
Yamamura Glass - $220mn mcap (0.6x PB, 11x PE) Japan's largest 'glass bottle' manufacturer with 110yr history, and also produces plastic caps, applied that technology to photonics--has a subsidiary that produces "optical package caps" used for fiber-optic communication in DCs. https://t.co/Y3UJ1eZO1o tweet
Offshore
Photo
Lumida Wealth Management
Why do Hedge Funds like Antero Resources?
Hedge funds are investing in Energy.
And, they aren't going for the hype names in nuclear.
They are opting for cash generating business with decades of inventory and reasonable valuations.
Antero Resources ( $AR ) was a highlight.
It is a U.S. Oil and natural gas in Appalachia (Marcellus/Utica) with leverage to LNG and power demand.
Lumida bought it as well.
Read our thesis in today's newsletter: https://t.co/aIPxSPfj4h
tweet
Why do Hedge Funds like Antero Resources?
Hedge funds are investing in Energy.
And, they aren't going for the hype names in nuclear.
They are opting for cash generating business with decades of inventory and reasonable valuations.
Antero Resources ( $AR ) was a highlight.
It is a U.S. Oil and natural gas in Appalachia (Marcellus/Utica) with leverage to LNG and power demand.
Lumida bought it as well.
Read our thesis in today's newsletter: https://t.co/aIPxSPfj4h
tweet
Offshore
Photo
Moon Dev
closing
ive given everyone a fair shot at joining the lifetime all access
but it has been long enough.
there have been some massive unlocks with openclaw and quant in general this week
so im going to step away and lock in alone
so thursday at midnight, the lifetime all access pass will be gone
and starting now, the only way to get into the private zooms is to be in the lifetime all access
so if i ever come back, the only way to get access to our private calls and everything i have to offer is here: https://t.co/EHUr5aAxhF
private zoom starts at 8am est today, only way in is through the lifetime.
tweet
closing
ive given everyone a fair shot at joining the lifetime all access
but it has been long enough.
there have been some massive unlocks with openclaw and quant in general this week
so im going to step away and lock in alone
so thursday at midnight, the lifetime all access pass will be gone
and starting now, the only way to get into the private zooms is to be in the lifetime all access
so if i ever come back, the only way to get access to our private calls and everything i have to offer is here: https://t.co/EHUr5aAxhF
private zoom starts at 8am est today, only way in is through the lifetime.
tweet
Offshore
Photo
Michael Fritzell (Asian Century Stocks)
🤔
tweet
🤔
whenever I use Claude Desktop for something beyond basic search/research - I gain a greater understanding of why there's such a large number of people who think AI is total hype/nonsense
Shit just reliably does *not* work. https://t.co/bqJQddm9jv tweet
Jukan
CLSA estimates that Samsung Electronics began commercial shipments of HBM4 to NVIDIA earlier this month, making a preemptive move ahead of competitors. However, insufficient yields on the 1c nm process are expected to limit meaningful market share gains in the near term.
CLSA projects the following HBM4 market share breakdown going forward: SK Hynix at 55%, Samsung Electronics at 25–30%, and Micron at 20%.
The firm notes that rising commodity DRAM prices are improving profitability, which in turn provides positive support for HBM pricing as players compete over limited production capacity. CLSA assesses that the ongoing shift in revenue mix toward high-value and customized products will continue to drive valuation re-rating for memory stocks.
Target prices: SK Hynix at KRW 1,250,000; Samsung Electronics at KRW 260,000; Micron at USD 495.
tweet
CLSA estimates that Samsung Electronics began commercial shipments of HBM4 to NVIDIA earlier this month, making a preemptive move ahead of competitors. However, insufficient yields on the 1c nm process are expected to limit meaningful market share gains in the near term.
CLSA projects the following HBM4 market share breakdown going forward: SK Hynix at 55%, Samsung Electronics at 25–30%, and Micron at 20%.
The firm notes that rising commodity DRAM prices are improving profitability, which in turn provides positive support for HBM pricing as players compete over limited production capacity. CLSA assesses that the ongoing shift in revenue mix toward high-value and customized products will continue to drive valuation re-rating for memory stocks.
Target prices: SK Hynix at KRW 1,250,000; Samsung Electronics at KRW 260,000; Micron at USD 495.
tweet
Jukan
Samsung Foundry Revival… Q1 Utilization Surpasses 80%, Breakeven as Early as Q4 This Year
Samsung Electronics' (005930) semiconductor foundry utilization rates have surpassed 80%, led by leading-edge nodes, with earnings improving rapidly. As chips developed in-house using Samsung's foundry—including 6th-generation HBM4 and Exynos 2600—have proven their performance, orders from global big tech companies are flooding in, making quarterly profitability within the year highly likely.
According to industry sources on the 22nd, utilization at Samsung Electronics' Pyeongtaek Campus P2 and P3 foundry production lines climbed to the 80% range in Q1. These lines, which produce semiconductors on mature nodes including 4nm, 5nm, and 7nm, struggled with insufficient production volume in H2 2024, with utilization falling well below 50% last year.
However, since H2 last year, Samsung's foundry division has applied its 4nm process to HBM4 base dies for the memory division, achieving the industry's best-in-class performance and driving a surge in order volume. On top of this, big tech companies from both the U.S. and China are placing successive chip production requests, with Samsung's semiconductor business spreading both wings wide. A source well-versed in Samsung's semiconductor production noted, "With orders from global big tech continuing to grow, centered on mature nodes, foundry utilization will comfortably exceed 80% in the first half of this year."
Production volume on Samsung Foundry's cutting-edge 2nm process is also increasing, brightening the earnings outlook. Samsung's in-house mobile application processor (AP), Exynos 2600, is being produced on the 2nm node, and with its performance reportedly surpassing Qualcomm's chips, supply volume for the Galaxy S26 is expected to increase.
Accordingly, Samsung Electronics' non-memory division (Foundry + System LSI) is expected to achieve an earnings turnaround as early as Q4. In particular, starting next year—when production of Tesla's AI5 chip, Apple's image sensors, and other products ramps—overall operating profit is likely to swing to positive.
An industry source noted, "With the foundry business recovering, Samsung Electronics' competitiveness as an integrated semiconductor company will be significantly strengthened."
tweet
Samsung Foundry Revival… Q1 Utilization Surpasses 80%, Breakeven as Early as Q4 This Year
Samsung Electronics' (005930) semiconductor foundry utilization rates have surpassed 80%, led by leading-edge nodes, with earnings improving rapidly. As chips developed in-house using Samsung's foundry—including 6th-generation HBM4 and Exynos 2600—have proven their performance, orders from global big tech companies are flooding in, making quarterly profitability within the year highly likely.
According to industry sources on the 22nd, utilization at Samsung Electronics' Pyeongtaek Campus P2 and P3 foundry production lines climbed to the 80% range in Q1. These lines, which produce semiconductors on mature nodes including 4nm, 5nm, and 7nm, struggled with insufficient production volume in H2 2024, with utilization falling well below 50% last year.
However, since H2 last year, Samsung's foundry division has applied its 4nm process to HBM4 base dies for the memory division, achieving the industry's best-in-class performance and driving a surge in order volume. On top of this, big tech companies from both the U.S. and China are placing successive chip production requests, with Samsung's semiconductor business spreading both wings wide. A source well-versed in Samsung's semiconductor production noted, "With orders from global big tech continuing to grow, centered on mature nodes, foundry utilization will comfortably exceed 80% in the first half of this year."
Production volume on Samsung Foundry's cutting-edge 2nm process is also increasing, brightening the earnings outlook. Samsung's in-house mobile application processor (AP), Exynos 2600, is being produced on the 2nm node, and with its performance reportedly surpassing Qualcomm's chips, supply volume for the Galaxy S26 is expected to increase.
Accordingly, Samsung Electronics' non-memory division (Foundry + System LSI) is expected to achieve an earnings turnaround as early as Q4. In particular, starting next year—when production of Tesla's AI5 chip, Apple's image sensors, and other products ramps—overall operating profit is likely to swing to positive.
An industry source noted, "With the foundry business recovering, Samsung Electronics' competitiveness as an integrated semiconductor company will be significantly strengthened."
tweet
Moon Dev
The 100M Framework: Using 5 Autonomous OpenClaws to Master 40x Leverage
most traders are fighting the market with a butter knife while i just deployed a five man digital army to take over the entire financial system. i used to be the guy getting liquidated and over trading until i realized that code is the great equalizer in this game.
losing money is a brutal teacher but it forced me to automate every single thought in my head. i spent hundreds of thousands on developers in the past thinking i couldn't code myself and it was a massive mistake.
w/ bots you have to iterate to success so i started building live to show the raw process of moving from manual frustration to fully automated systems. right now i have five open claws running on separate machines because i refused to let them compete for resources.
it sounds overkill until you realize that one small lag in your data layer can cost you a fortune in slippage. most people are debating if they should use ubuntu or windows while i am chasing a hundred million dollar minimum use case for every agent i deploy.
but here is the catch because even with a digital army you can still get smoked if your infrastructure is weak. i actually got hacked recently because i showed too much on my screen which changed everything about how i operate.
now i have a secret protocol for my agents that keeps the ops out while they build in the shadows. if you want to know how i organized five autonomous agents to work on one repository without stepping on each other you need to understand the fork model.
each agent has their own identity like killer who is my workhorse or mash who is always stepping on the gas. i give them a shared github token but i never let them push to my main repository directly.
they work on their own forks and submit pull requests to cracker who is my queen alpha agent in charge of the whole fleet. this setup allows me to scale to five hundred agents if i wanted to because branches are cheap and unlimited.
the real alpha is not in predicting where the price goes because that is a losing game for mere mortals. i focus on predicting who is about to get liquidated and i buy their fear using institutional grade data.
hyperliquid has opened up the financial system but most people are still using forty times leverage like they are at a casino. i looked at the data and ninety nine percent of traders are still trading by hand which is exactly why my bots are so profitable.
they are just taking over positions from people who can't control their emotions or their risk. it took me five years of experience to build this into a downloadable app that puts bumpers around your trading.
the main feature is buying and selling around liquidations because nobody can truly predict a one percent price move with certainty. i just wait for someone to get licked and then my system steps in to provide the liquidity they just lost.
i launched my own node because the standard apis are just too slow for the level of precision i need. you need tick data for every single token if you want a real edge in this high frequency environment.
if you are trading algorithmically and not using tick data you are missing the biggest opportunity in the market. most people settle for open high low close data but the real gold is buried in every single tick of the order book.
i have a monitor running all day checking fifty five different endpoints to make sure my data layer stays healthy. it is a self healing system that handles hiccups and keeps the bots fed with the fastest information possible.
there is a weird dynamic when you build in public that i call the dichotomy of the chat. half the people are senior engineers giving me fire ideas for optimizations and the other half are asking how to install python.
it is a massive signal that the content hits on multiple levels because i am pulling in the smartest brains to review my code for free. i will answer any question because i believe in helping t[...]
The 100M Framework: Using 5 Autonomous OpenClaws to Master 40x Leverage
most traders are fighting the market with a butter knife while i just deployed a five man digital army to take over the entire financial system. i used to be the guy getting liquidated and over trading until i realized that code is the great equalizer in this game.
losing money is a brutal teacher but it forced me to automate every single thought in my head. i spent hundreds of thousands on developers in the past thinking i couldn't code myself and it was a massive mistake.
w/ bots you have to iterate to success so i started building live to show the raw process of moving from manual frustration to fully automated systems. right now i have five open claws running on separate machines because i refused to let them compete for resources.
it sounds overkill until you realize that one small lag in your data layer can cost you a fortune in slippage. most people are debating if they should use ubuntu or windows while i am chasing a hundred million dollar minimum use case for every agent i deploy.
but here is the catch because even with a digital army you can still get smoked if your infrastructure is weak. i actually got hacked recently because i showed too much on my screen which changed everything about how i operate.
now i have a secret protocol for my agents that keeps the ops out while they build in the shadows. if you want to know how i organized five autonomous agents to work on one repository without stepping on each other you need to understand the fork model.
each agent has their own identity like killer who is my workhorse or mash who is always stepping on the gas. i give them a shared github token but i never let them push to my main repository directly.
they work on their own forks and submit pull requests to cracker who is my queen alpha agent in charge of the whole fleet. this setup allows me to scale to five hundred agents if i wanted to because branches are cheap and unlimited.
the real alpha is not in predicting where the price goes because that is a losing game for mere mortals. i focus on predicting who is about to get liquidated and i buy their fear using institutional grade data.
hyperliquid has opened up the financial system but most people are still using forty times leverage like they are at a casino. i looked at the data and ninety nine percent of traders are still trading by hand which is exactly why my bots are so profitable.
they are just taking over positions from people who can't control their emotions or their risk. it took me five years of experience to build this into a downloadable app that puts bumpers around your trading.
the main feature is buying and selling around liquidations because nobody can truly predict a one percent price move with certainty. i just wait for someone to get licked and then my system steps in to provide the liquidity they just lost.
i launched my own node because the standard apis are just too slow for the level of precision i need. you need tick data for every single token if you want a real edge in this high frequency environment.
if you are trading algorithmically and not using tick data you are missing the biggest opportunity in the market. most people settle for open high low close data but the real gold is buried in every single tick of the order book.
i have a monitor running all day checking fifty five different endpoints to make sure my data layer stays healthy. it is a self healing system that handles hiccups and keeps the bots fed with the fastest information possible.
there is a weird dynamic when you build in public that i call the dichotomy of the chat. half the people are senior engineers giving me fire ideas for optimizations and the other half are asking how to install python.
it is a massive signal that the content hits on multiple levels because i am pulling in the smartest brains to review my code for free. i will answer any question because i believe in helping t[...]
Offshore
Moon Dev The 100M Framework: Using 5 Autonomous OpenClaws to Master 40x Leverage most traders are fighting the market with a butter knife while i just deployed a five man digital army to take over the entire financial system. i used to be the guy getting…
he community but i have zero patience for people who won't read the docs.
discipline is doing what you hate to do but doing it like you love it which is a quote that changed my life. if you can't lock in for three hours to build a system you will never survive the five year journey to success.
i am not afraid to die on a treadmill because i know i will stay on longer than anyone else competing with me. you might be smarter or more talented but i have the discipline to iterate until the system is perfect.
right now i am testing different models like minimax two point five and glm four point seven to see which one has the highest coding logic. minimax direct seems to be stealing the show because it is fast and hasn't missed a single instruction yet.
i even built a slot machine style dashboard for my research agent so i can see the backtests happening in real time. it is fun to look at but it serves a serious purpose of identifying hundred million dollar opportunities while i sleep.
my rbi system is how i automate everything which stands for ramble research backtest and implement. i ramble my ideas to an ai that transcribes them and then passes them to the research agents to find the math.
once they find a strategy that works on historical data they move it to the backtest engine for stress testing. if it survives the fire it gets implemented into the live fleet and starts printing for the empire.
the path to becoming a fully automated trader is not about finding a magic indicator or a secret setting. it is about building a system that can iterate faster than the market can change its mind.
i used to spend all day staring at charts and getting stressed out by every red candle on the screen. now i just watch my dashboard and let the agents handle the heavy lifting while i focus on the next big pivot.
if you think you are too late to the game you are actually just in time because the tools are finally here for the normies to win. code is the great equalizer and if you can learn to speak to the machines you can own your future.
the separation between you and your competitors grows larger every single day that you decide to lock in. by year five it won't matter what kind of work they do in a summer because they will never be able to catch up.
i am chasing the biggest goal in the world which is putting an imprint of my thoughts on the global financial system. some people think it is impossible but i have unlimited shots at this life and i am taking every single one.
as long as you have a vision and the code to back it up you can create your own environment. you create your own luck by being the one who refused to get off the treadmill when things got hard.
the world is what you make it and i decided to make it a place where my bots do the work while i enjoy the island. the game has just begun and the only question left is if you are going to watch from the sidelines or join the winners.
i will be here tomorrow morning at the same time because this is a non negotiable contract i signed with myself. i am itching to get back to the terminal and see what my agents cooked up while i was away.
it is a beautiful world when you realize that you have the power to automate your freedom with a few lines of logic. i believe in you and your ability to figure this out if you just decide to stay locked in.
the dichotomy of my life is that i am just a normie who decided to stop being scared of the terminal. if i can go from being held back in school to building a digital trading empire then you have no excuses left.
get to work and stop living someone else's life because your time is the only asset you can never buy back. i will see you at the top or i will see you on the treadmill but either way i am never stopping
tweet
discipline is doing what you hate to do but doing it like you love it which is a quote that changed my life. if you can't lock in for three hours to build a system you will never survive the five year journey to success.
i am not afraid to die on a treadmill because i know i will stay on longer than anyone else competing with me. you might be smarter or more talented but i have the discipline to iterate until the system is perfect.
right now i am testing different models like minimax two point five and glm four point seven to see which one has the highest coding logic. minimax direct seems to be stealing the show because it is fast and hasn't missed a single instruction yet.
i even built a slot machine style dashboard for my research agent so i can see the backtests happening in real time. it is fun to look at but it serves a serious purpose of identifying hundred million dollar opportunities while i sleep.
my rbi system is how i automate everything which stands for ramble research backtest and implement. i ramble my ideas to an ai that transcribes them and then passes them to the research agents to find the math.
once they find a strategy that works on historical data they move it to the backtest engine for stress testing. if it survives the fire it gets implemented into the live fleet and starts printing for the empire.
the path to becoming a fully automated trader is not about finding a magic indicator or a secret setting. it is about building a system that can iterate faster than the market can change its mind.
i used to spend all day staring at charts and getting stressed out by every red candle on the screen. now i just watch my dashboard and let the agents handle the heavy lifting while i focus on the next big pivot.
if you think you are too late to the game you are actually just in time because the tools are finally here for the normies to win. code is the great equalizer and if you can learn to speak to the machines you can own your future.
the separation between you and your competitors grows larger every single day that you decide to lock in. by year five it won't matter what kind of work they do in a summer because they will never be able to catch up.
i am chasing the biggest goal in the world which is putting an imprint of my thoughts on the global financial system. some people think it is impossible but i have unlimited shots at this life and i am taking every single one.
as long as you have a vision and the code to back it up you can create your own environment. you create your own luck by being the one who refused to get off the treadmill when things got hard.
the world is what you make it and i decided to make it a place where my bots do the work while i enjoy the island. the game has just begun and the only question left is if you are going to watch from the sidelines or join the winners.
i will be here tomorrow morning at the same time because this is a non negotiable contract i signed with myself. i am itching to get back to the terminal and see what my agents cooked up while i was away.
it is a beautiful world when you realize that you have the power to automate your freedom with a few lines of logic. i believe in you and your ability to figure this out if you just decide to stay locked in.
the dichotomy of my life is that i am just a normie who decided to stop being scared of the terminal. if i can go from being held back in school to building a digital trading empire then you have no excuses left.
get to work and stop living someone else's life because your time is the only asset you can never buy back. i will see you at the top or i will see you on the treadmill but either way i am never stopping
tweet
X (formerly Twitter)
Moon Dev (@MoonDevOnYT) on X
The 100M Framework: Using 5 Autonomous OpenClaws to Master 40x Leverage
most traders are fighting the market with a butter knife while i just deployed a five man digital army to take over the entire financial system. i used to be the guy getting liquidated…
most traders are fighting the market with a butter knife while i just deployed a five man digital army to take over the entire financial system. i used to be the guy getting liquidated…
Offshore
Photo
The Transcript
RT @TheTranscript_: $KO Coca-Cola CEO: "If you take a step back, we have a long track record of navigating complex external dynamics to hold or grow volume each year. Over the past 50 years, annual volume declined only once, and that was during the pandemic." https://t.co/BHfLnG5L6C
tweet
RT @TheTranscript_: $KO Coca-Cola CEO: "If you take a step back, we have a long track record of navigating complex external dynamics to hold or grow volume each year. Over the past 50 years, annual volume declined only once, and that was during the pandemic." https://t.co/BHfLnG5L6C
tweet
Jukan
《GF Securities Overseas Electronics & Communications》
Onto Innovation 4Q25 Review: Upside Potential Remains
Maintaining Buy rating, target price raised to $282: Onto Innovation's 4Q25 revenue came in at $267M, largely in line with expectations. The company guided 1Q26 revenue to a range of $275–285M, with 2Q26 revenue expected to surpass $300M. Memory and GAA capacity buildouts, combined with the company's continued market share gains in 2.5D/3D, underpin robust growth. Management indicated that in 2026, advanced packaging revenue will grow 30% and advanced node revenue will grow 15%. G5 qualification is expected to be completed in 1H26, and potential G5 applications present additional upside to earnings. We forecast Onto Innovation's 2026/2027 revenue at $1.3B/$1.5B respectively, and derive our $282 target price based on 30x 2027E P/E.
Market share gains accelerating: The company's market share expansion is progressing well. Management announced a large-scale HBM Volume Purchase Agreement (VPA) with a leading HBM customer worth over $240M, of which more than $60M is earmarked specifically for 3D bump metrology. This single HBM order alone is nearly equivalent to the company's entire AI packaging revenue for full-year 2025. In CoWoS, G5 has completed qualification at a foundry customer, and the company is recapturing sub-micron inspection market share. Additionally, Firefly and Iris thin-film metrology continue to drive share gains.
New catalysts emerging for 2027: We highlight several new growth vectors for Onto Innovation. 1) SoIC: NVIDIA's Feynman platform will adopt a 3D architecture combining SoIC and CoWoS. SoIC is poised to become a core growth driver in 2027, and the company's G5 product holds distinct advantages in both sub-micron inspection and 3D applications. 2) Panel-level packaging: Onto Innovation secured orders for JetStep and 8 Firefly systems in 4Q25 to support capacity at a panel-level packaging facility; G5 equipment is expected to be delivered to TSMC by year-end for a panel packaging pilot line. 3) CPO: The company has initiated early-stage engagements with CPO customers, including V-groove metrology solutions. CPO is expected to become a meaningful revenue contributor starting in 2027.
$ONTO
tweet
《GF Securities Overseas Electronics & Communications》
Onto Innovation 4Q25 Review: Upside Potential Remains
Maintaining Buy rating, target price raised to $282: Onto Innovation's 4Q25 revenue came in at $267M, largely in line with expectations. The company guided 1Q26 revenue to a range of $275–285M, with 2Q26 revenue expected to surpass $300M. Memory and GAA capacity buildouts, combined with the company's continued market share gains in 2.5D/3D, underpin robust growth. Management indicated that in 2026, advanced packaging revenue will grow 30% and advanced node revenue will grow 15%. G5 qualification is expected to be completed in 1H26, and potential G5 applications present additional upside to earnings. We forecast Onto Innovation's 2026/2027 revenue at $1.3B/$1.5B respectively, and derive our $282 target price based on 30x 2027E P/E.
Market share gains accelerating: The company's market share expansion is progressing well. Management announced a large-scale HBM Volume Purchase Agreement (VPA) with a leading HBM customer worth over $240M, of which more than $60M is earmarked specifically for 3D bump metrology. This single HBM order alone is nearly equivalent to the company's entire AI packaging revenue for full-year 2025. In CoWoS, G5 has completed qualification at a foundry customer, and the company is recapturing sub-micron inspection market share. Additionally, Firefly and Iris thin-film metrology continue to drive share gains.
New catalysts emerging for 2027: We highlight several new growth vectors for Onto Innovation. 1) SoIC: NVIDIA's Feynman platform will adopt a 3D architecture combining SoIC and CoWoS. SoIC is poised to become a core growth driver in 2027, and the company's G5 product holds distinct advantages in both sub-micron inspection and 3D applications. 2) Panel-level packaging: Onto Innovation secured orders for JetStep and 8 Firefly systems in 4Q25 to support capacity at a panel-level packaging facility; G5 equipment is expected to be delivered to TSMC by year-end for a panel packaging pilot line. 3) CPO: The company has initiated early-stage engagements with CPO customers, including V-groove metrology solutions. CPO is expected to become a meaningful revenue contributor starting in 2027.
$ONTO
tweet