Blog

  • Built at 60 — The Series

    One retired engineer. One computer. Eleven parts. No guru promises.

    On 31/12/2025, my company went bankrupt. I was 59, with 35 years of civil
    engineering behind me — and nowhere to go in the morning.

    I knew zero code. Then I met AI. This series is what happened next: what I
    built, what broke, what it actually cost, and who I became along the way.

    No income claims. No “anyone can do it.” Just one true story, told in
    eleven parts, one per day.

    New here? Start at Part 1 👇

    [https://www.vetting-smart.top/i-retired-at-60-then-i-started-over/]

    Act I — The Beginning

    1. I Retired at 60. Then I Started Over.
    Retirement as a beginning, not an ending. The empty calendar, the three hard
    months, and the decision to build instead of rest.
    [https://www.vetting-smart.top/i-retired-at-60-then-i-started-over/] · Live 19/10

    2. I Knew Zero Code
    The blank screen at age 60. First contact with code and AI, and the discipline of showing up every day.
    Part 2 is live :[https://www.vetting-smart.top/i-knew-zero-code/]

    3. My First 100 Stupid Questions
    The dumbest question that unlocked everything. Asking well turned out to be
    the skill itself.
    Part 3 is live: [https://www.vetting-smart.top/my-first-100-stupid-questions/]

    Act II — The Building

    4. I Built the Machine
    From doing tasks by hand to building systems that do them — workflows, agents,
    a VPS. The engineer’s shift.
    Coming 22/10

    5. I Built a Website for $2 (None of It Mine)
    A $5 broker gift became a live website: $1 domain + $1 hosting. Not to get
    rich — the joy of working in my twilight years.
    Coming 23/10

    6. What It Actually Cost
    The “$0” myth vs the real ledger: VPS fees, domains, and the hours nobody
    counts. Line by line, honestly.
    Coming 24/10

    Act III — The Hardest Build

    7. From Zero to an EA on a VPS
    The trading journey in one arc: the bug that lied to me, 11.75 years of
    backtesting, and an EA now running live — untouched.
    Coming 25/10

    Act IV — The Human

    8. I Publish in English. It’s Not My Language.
    Every word you read was written in a language I don’t speak. The ideas are
    mine. The English is borrowed. Both are disclosed.
    Coming 26/10

    9. My Workday Starts at 1 AM
    While the city sleeps. The 1–2 AM rhythm, honestly told — discipline designed
    like a structure.
    Coming 27/10

    10. A Civil Engineer Builds Systems — Even in Markets
    Foundations, load testing, safety margins. What 35 years of engineering taught
    me about markets, machines, and myself.
    Coming 28/10

    Epilogue

    11. The Easiest Way to Support an Old Man’s Work
    The ask — earned by the story, carried by humor. It costs you nothing extra.
    It keeps me building.
    Coming 29/10

    This page updates daily as new parts go live. Want each part in your inbox?
    [Newsletter signup link] — one email per part, no spam, unsubscribe anytime.

    We vet tokens. We vet systems. We vet ourselves.

  • “Honest System Validation: A 4-Part Series”

    Most trading content shows you the wins. This series shows you the bug hunt. 🔍

    Over the next 4 days, I’m publishing everything — the full, unfiltered story of how I validated a trading system from scratch:

    Part 1  is live: I Almost Lost Money Because of a Stupid Coding Bug. 👇

    https://www.vetting-smart.top/i-almost-lost-money-because-of-a-stupid-coding-bug/

    Part 2 is live — 819 trades, 11.75 years: the complete backtest report. Every number. Including the losing year.👇

    https://www.vetting-smart.top/819-trades-11-75-years-the-complete-backtest-report/

    Part 3 is live — I trained an AI to predict gold prices. It flipped coins. Here’s the autopsy.👇

    https://www.vetting-smart.top/i-trained-an-ai-to-predict-gold-prices-it-flipped-coins-heres-the-autopsy/

    Part 4 is live— The EA goes live on a VPS. I promise not to touch it for 6 months.👇

    https://www.vetting-smart.top/my-trading-ea-went-live-today-the-hardest-rule-dont-touch-it/

    No rented cars. No guru promises. Just data, code, and every mistake I made — so you don’t have to make them.

    We vet tokens. We vet systems. We vet ourselves.

    #VettingSmart #Backtesting #AlgorithmicTrading #HonestQuant

  • Musk Describes Tesla-SpaceX Merger as Good Idea Following Ninefold AI Revenue Growth

    Musk Describes Tesla-SpaceX Merger as Good Idea Following Ninefold AI Revenue Growth
    Original illustration by Vetting Smart.

    Confirmed Facts

    Elon Musk has characterized a potential merger between Tesla and SpaceX as a “good idea.” This statement coincides with reports that revenue related to artificial intelligence (AI) has increased ninefold. If executed, the merger would bring Tesla into the same corporate group as SpaceXAI, the developer of the Grok AI model, and the social media platform X.

    Possible Implications

    Such a merger could lead to a centralized corporate structure, potentially streamlining the integration of Grok’s AI capabilities with Tesla’s hardware and automotive systems. A ninefold jump in AI revenue suggests that Musk’s AI ventures are scaling rapidly, which could influence the valuation of his public and private entities if consolidated. This move might indicate a strategic shift toward unified resource management across his social media, aerospace, and electric vehicle companies.

    Source Limitations

    The provided excerpt does not specify the exact dollar figures behind the ninefold revenue growth or the specific timeframe for this increase. It fails to clarify the legal or structural relationship between SpaceX and the entity referred to as “SpaceXAI.” Furthermore, the text does not outline a formal timeline for a merger or provide details on the regulatory hurdles associated with combining a publicly traded firm like Tesla with private enterprises.

    Source: Cointelegraph. Analysis based on the RSS excerpt.

    Read more on Vetting Smart Blog

    Published: 21/10/2026 – 04:00 (UTC)

  • OpenAI Acquires Orchestra Bio to Integrate AI Agents into Life Sciences Suite

    OpenAI Acquires Orchestra Bio to Integrate AI Agents into Life Sciences Suite
    Original illustration by Vetting Smart.

    Confirmed Facts

    OpenAI has acquired Orchestra Bio to incorporate the company’s drug-program agents into Rosalind, which serves as OpenAI’s life sciences suite. The objective of this acquisition is to utilize artificial intelligence to reduce the time required to bring new medicines to patients, a process that typically spans more than a decade.

    Possible Implications

    If these AI agents successfully automate routine drug development tasks, the pharmaceutical industry could see a marked reduction in development cycles. This move may signal a broader trend of major AI firms developing specialized vertical suites for high-stakes industries like biotechnology.

    Source Limitations

    The provided excerpt does not disclose the financial terms or the specific date of the acquisition. It fails to provide technical specifications for the Rosalind suite or detail the exact mechanisms by which the AI agents will accelerate drug development. Additionally, the text does not identify specific pharmaceutical partners or target medical fields.

    Source: Decrypt. Analysis based on the RSS excerpt.

    Read more on Vetting Smart Blog

    Published: 20/10/2026 – 22:00 (UTC)

  • My First 100 Stupid Questions

    The dumbest one unlocked everything.

    I have asked artificial intelligence roughly a hundred stupid questions. Maybe more. I stopped counting somewhere around the time I stopped feeling embarrassed.

    Here is the thing nobody tells you about learning at sixty: the hardest part isn’t the technology. It’s the shame of not knowing.

    “Explain it like I’m a villager”

    My earliest questions weren’t even about code. They were: I want to do this — where do I start? How? And then, holding up a piece of code: translate this into plain folk language so I can understand the problem.

    I didn’t ask “teach me programming.” I asked a machine to speak human to me first.

    The three questions that never got old

    Over the months, three questions became my permanent toolkit — I asked them about every piece of code I touched:

    1. What does this line actually mean?
    2. Why does it have to be there?
    3. If I remove it, what breaks in the whole system?

    The AI answered all three. Every time. It never sighed, never said “we covered this yesterday.” A human teacher would have lost patience by question thirty. The machine was on question ninety-nine with the same calm as question one.

    Ask, break, ask again

    Here is a loop I lived in for months: ask the AI, follow its answer, watch it fail, ask again differently, fail again, ask a third way — and suddenly it works.

    My favorite embarrassment: I once spent days asking the AI to fix an EA dashboard that refused to appear on the chart. Back and forth, back and forth — and I even scolded the AI for not understanding me. Then one night, lying in bed with my hands behind my head, I asked myself: why is this so weird? And it hit me: every time I updated to a new EA version, I forgot to change the EA’s name in the Tester’s “Expert” field. I had been testing the OLD version over and over — of course nothing changed.

    I went straight back to the computer. And there it was: the beautiful dashboard, working since several versions ago. I slapped my forehead. How dumb can you be.

    Somewhere inside loops like that I learned the real lesson: the problem was rarely that the AI was wrong. The problem was me — my mistake, my impatience, my question not good enough yet. Learning to ask well turned out to be the actual skill. The code was just the excuse.

    The free-tier university

    I studied on the free tier. When my questions ran out, someone taught me a trick: switch to another account and keep going. My university had many doors — when one closed for the day, I knocked on the next.

    Was it elegant? No. Did it work? Completely. The limits didn’t stop the learning; they scheduled it. The 2 a.m. sessions, the rationed questions — scarcity made every single question count.

    The only stupid question

    People my age often don’t start because they’re afraid of looking foolish. I understand — I felt it too, every time I typed something a teenager would laugh at.

    But here is what a hundred stupid questions taught me: the only stupid question is the one pride won’t let you ask.

    The machine never laughed. Not once. And because it never laughed, I never stopped asking. That — more than any code I ever copied — is what built everything.

    Next: I Built the Machine — from doing tasks to building systems that do tasks.

  • BlackRock Proposes Bitcoin Fund to Mitigate Losses Following Calamos Model

    BlackRock Proposes Bitcoin Fund to Mitigate Losses Following Calamos Model
    Original illustration by Vetting Smart.

    Confirmed Facts

    BlackRock has proposed a new Bitcoin fund designed to shield investors from significant downside risks. This initiative follows the performance of similar products from Calamos, which reportedly assisted investors in avoiding substantial losses during Bitcoin market selloffs. The proposal is framed by the performance of a rival fund that successfully navigated a 50% market crash.

    Possible Implications

    The introduction of a loss-shielding product by a major asset manager like BlackRock may indicate a growing institutional demand for structured digital asset vehicles that prioritize capital preservation. If these risk-managed funds become a standard offering, they could potentially attract conservative participants who have previously avoided the spot market due to its inherent volatility.

    Source Limitations

    The provided excerpt does not define the specific technical mechanisms, such as options-based strategies or buffer percentages, that BlackRock intends to use. It fails to identify the specific rival fund referenced in the headline or provide a definitive timeline for regulatory approval. Furthermore, the text does not quantify the exact amount of losses avoided by investors in the existing Calamos products.

    Source: CoinDesk. Analysis based on the RSS excerpt.

    Read more on Vetting Smart Blog

    Published: 20/10/2026 – 16:00 (UTC)

  • SkyAI Liquidates Solana Treasury and Initiates 201 Million USD Share Buyback

    SkyAI Liquidates Solana Treasury and Initiates 201 Million USD Share Buyback
    Original illustration by Vetting Smart.

    Confirmed Facts

    SkyAI, formerly a medical device manufacturer, has liquidated its entire Solana (SOL) treasury. The company reports holding 223.8 million USD in cash reserves. Utilizing these funds, SkyAI has launched an initiative to repurchase outstanding shares and warrants, with the buyback program valued at 201 million USD.

    Possible Implications

    The complete liquidation of SOL holdings could suggest a strategic pivot away from digital asset treasury management for this specific entity. A 201 million USD buyback may provide significant liquidity for existing shareholders and warrant holders, potentially consolidating ownership as the firm moves away from its former business model.

    Source Limitations

    The provided excerpt does not state the specific quantity of Solana liquidated or the average price at which the assets were sold. It fails to provide an explicit reason for the exit or the total profit or loss realized. Furthermore, the text does not establish a definitive timeline for the completion of the buyback.

    Source: Cointelegraph. Analysis based on the RSS excerpt.

    Read more on Vetting Smart Blog

    Published: 20/10/2026 – 04:00 (UTC)

  • I Knew Zero Code

    I mean it literally. I didn’t know what code was.

    When I say “zero,” people imagine a beginner who at least knows what a variable is. I didn’t. I had a computer at home used for reading news, listening to music, and answering emails. That was its full job description.

    Then one night I asked myself a strange question: can AI teach me something? At my age, I can’t walk into a classroom and sit next to Gen Z students. But maybe a machine would have me.

    The first question

    My first conversation was with ChatGPT. My first question wasn’t “teach me programming.” It was practical, almost rude in its directness: I want to convert a TradingView indicator to MT5 so I don’t have to watch two screens. Can that be done?

    The AI said yes — and did it.

    So I asked the next question: If that’s possible, can I enter trades automatically instead of clicking by hand?

    The AI said yes again.

    That’s how it started. Not with a course. Not with a textbook. With two questions from a man who didn’t know what code was.

    The 2 a.m. classroom

    I have no teacher and no classroom. What I have is old age: I sleep badly, wake up around 2 a.m. completely alert, and sit down to finish what yesterday left undone. The free tier helps with discipline — when your questions run out, you stop, whether you want to or not.

    Every night, the same ritual. Ask. Copy. Paste. Break something. Ask why it broke.

    Ninety-nine percent

    Let me be precise about the division of labor: the AI did 99% of the work. I copied and pasted. And my real contribution — the 1% — was asking what I call “de-dumbing” questions: What does this line of code actually mean? Why does it have to be there? If I remove it, what breaks in the whole system?

    The AI never laughed. It explained, again and again, until a 60-year-old civil engineer understood.

    Theory is theory

    I was overjoyed when my first EA was born. The Strategy Tester drew a beautiful climbing curve — $3,000 becoming $94,000 on the screen. It felt like magic.

    “My first backtest: $3,000 → $94,000. Beautiful — and a lie.”

    Then reality: forex is brutal. Win three trades, give one back. None of those early bots ever brought home a profit. Theory is theory. The market doesn’t read backtests.

    “Reality doesn’t read backtests.”

    The same instinct later built the Shitcoin Radar: hundreds of new tokens appear every minute, and nobody can tell which are real and which are spam. So I built a machine to look.

    What “zero” actually looked like

    Zero looked like a man at 2 a.m., asking a machine to explain a single line of code like he was five years old — and the machine answering, patiently, every time.

    I still can’t write code from scratch. What I learned is something else: how to ask, how to persist, and how to tell the difference between a beautiful curve on a screen and money in an account.

    The tool did the heavy lifting. I take credit for exactly one thing: I showed up.

    Next: My First 100 Stupid Questions — the dumbest question that unlocked everything.

  • Nasdaq-Listed Strive Acquires 335 Million USD in Bitcoin Following Warrant Exercise

    Nasdaq-Listed Strive Acquires 335 Million USD in Bitcoin Following Warrant Exercise
    Original illustration by Vetting Smart.

    Confirmed Facts

    A Nasdaq-listed Bitcoin holder, identified as Strive, has purchased 335 million USD in Bitcoin. This acquisition follows a capital influx of 369 million USD generated by warrant holders exercising their options. Additionally, the entity has disclosed plans for a 630 million USD sale of 15% preferred stock.

    Possible Implications

    The successful exercise of warrants by investors may indicate sustained confidence in the entity’s strategy to hold digital assets on its balance sheet. If the planned 630 million USD preferred stock sale is fully realized, it could significantly increase the firm’s liquidity, potentially allowing for further expansion of its Bitcoin holdings or other corporate initiatives.

    Source Limitations

    The provided excerpt does not specify the exact quantity of Bitcoin purchased or the average price per coin at the time of the transaction. It fails to identify the specific warrant holders or provide a definitive timeline for the proposed 630 million USD preferred stock sale. Furthermore, the text does not clarify the intended use of the remaining 34 million USD from the warrant exercise that was not immediately directed toward the Bitcoin purchase.

    Source: Decrypt. Analysis based on the RSS excerpt.

    Read more on Vetting Smart Blog

    Published: 19/10/2026 – 22:00 (UTC)

  • Crypto Vault Deposits Exceed 10 Billion USD Amid Institutional Interest and Lending Risks

    Crypto Vault Deposits Exceed 10 Billion USD Amid Institutional Interest and Lending Risks
    Original illustration by Vetting Smart.

    Confirmed Facts

    Total deposits within crypto vaults have surpassed 10 billion USD. These assets are primarily concentrated in curator-managed lending strategies, which are reportedly attracting significant institutional capital. However, these investment vehicles expose depositors to several categories of risk, specifically involving collateral, liquidity, and smart-contract vulnerabilities.

    Possible Implications

    The growth of crypto vault deposits to over 10 billion USD suggests a strengthening institutional appetite for managed yield products despite the inherent risks of decentralized infrastructure. If smart-contract or liquidity risks are realized, it could result in substantial losses for institutional participants, potentially prompting the regulatory scrutiny mentioned in the report’s title.

    Source Limitations

    The provided excerpt does not identify the specific platforms, curators, or institutional investors contributing to the 10 billion USD total. It fails to define the exact nature or origin of the regulatory scrutiny mentioned or provide a specific timeline for when the deposit milestone was reached. Furthermore, the text does not quantify the specific level of exposure across different risk categories.

    Source: CoinDesk. Analysis based on the RSS excerpt.

    Read more on Vetting Smart Blog

    Published: 19/10/2026 – 10:00 (UTC)

  • Qubic Halts Network Operations Following Exploitation of 1,400 Wallets

    Qubic Halts Network Operations Following Exploitation of 1,400 Wallets
    Original illustration by Vetting Smart.

    Confirmed Facts

    Qubic has suspended its network operations after an attacker successfully drained approximately 1,400 wallets. The exploit is attributed to the use of weak seeds. Reports indicate that 87.1% of the stolen assets remain on-chain, while the remaining portion has been frozen by the cryptocurrency exchange MEXC. Network validators are currently preparing to vote on a formal recovery plan.

    Possible Implications

    If the validator vote fails to reach consensus, the restoration of network functionality and the recovery of assets could be delayed. This incident may underscore the necessity for enhanced security standards and seed generation protocols within the ecosystem to mitigate future vulnerabilities.

    Source Limitations

    The provided excerpt does not state the total monetary value of the stolen assets. It fails to define the technical parameters of weak seeds or provide a specific timeline for the validator vote and the subsequent network restart.

    Source: Cointelegraph. Analysis based on the RSS excerpt.

    Read more on Vetting Smart Blog

    Published: 19/10/2026 – 04:00 (UTC)

  • Spot Bitcoin ETFs Record Nearly 1 Billion USD in Net Outflows Over Two-Week Period

    Spot Bitcoin ETFs Record Nearly 1 Billion USD in Net Outflows Over Two-Week Period
    Original illustration by Vetting Smart.

    Confirmed Facts

    Bitcoin exchange-traded funds (ETFs) recorded a net outflow of 228.6 million USD during the most recent week. This follows a reported 681.1 million USD drain in the preceding week. Collectively, these figures represent a total outflow of nearly 1 billion USD over a fourteen-day period. The report characterizes the performance of these products during the current month as being deep into the red.

    Possible Implications

    If the trend of multi-hundred-million-dollar weekly outflows persists, it could suggest a sustained cooling of interest in spot Bitcoin investment products. The shift toward significant outflows during this period might indicate that broader market factors are currently outweighing historical seasonal performance expectations.

    Source Limitations

    The provided excerpt does not provide a breakdown of flows by specific ETF issuers or individual fund brands. It fails to identify the exact calendar dates for the weeks mentioned or provide data regarding total assets under management. Furthermore, the text does not establish the specific macroeconomic or technical catalysts responsible for the recorded outflows.

    Source: Decrypt. Analysis based on the RSS excerpt.

    Read more on Vetting Smart Blog

    Published: 18/10/2026 – 22:00 (UTC)