My memory said −61%. My records said −78%.

Investing journal, Memory vs Records

Three things I learned

1. Memory rewrites itself around outcomes. My own notes put my loss by the end of 2023 at −61%. Money in versus money left said about −78%.
2. The most useful thing a record catches is not a number. It is the moment you grant yourself an exception.
3. So in fall 2026 I rebuilt my record system. One document a day. Nothing overwritten. One source of truth.

What happened

I got into crypto at the end of 2021. For years all I had were a few memo notes, some screenshots and trade history files. In September 2026 I lined them up by date and wrote down what they actually showed.

1. Dec 2021: I mistook a lottery for skill. A few NFT mints revealed as rare pieces and my money grew about 5x in days. A reveal is random. I remembered it as judgment. I then spread money across four or five projects, which were really one bet on the same NFT mania.
2. Jan 31, 2022: one screenshot. I put money into an NFT project whose team I could not even name. It was a rug pull. I saved a screenshot titled “Remember this RUG Pull” and promised myself to verify everything. Then I used that promise to justify going all in on the next project.
3. 2022 to 2023: the file said −61%, the math said −78%. The Fed started hiking on March 16, 2022, and the crypto market went through one collapse after another. I was not watching macro at all. My largest NFT project shut down and its value went to almost zero. My own file said −48% at the end of 2022 and −61% at the end of 2023. When I recomputed it from money in and money left, the end of 2023 was about −78%.
4. 2023 to 2024: one survivor and my first rule. I sold what was dead and moved what was left into one game project that was still alive. It moved to another chain and came back to life, and so did my portfolio. Lining up the dates, the thing that saved me was an outside event, not my judgment. It was luck. In April 2024 I wrote my first ban: no ETH restaking, no altcoin DeFi.
5. May 2024: one signature. I signed a request on a fake airdrop site and the deposit tokens in my hardware wallet were drained. When I traced it on chain two years later, my memory was off twice. I remembered mid April; it was about four weeks later. I remembered a bigger loss; the real one was about a fifth smaller. A few days after the hack I wrote a rule to retire from crypto.
6. Nov 2024: the exception. I broke both rules and went back to scalping altcoins. It worked, which made it worse. Looking back in 2026, I wrote: “I broke the rule. I made an exception that fit the moment and rationalized it.” (translated)
7. Sep 17, 2025: my first review note. It opened with: “Learn the present from the past, and change the future through the present.” (translated) It was mostly memory, with no dates or numbers checked.
8. Sep 22, 2026: I thought it was 3x. It was 10x. I chased a short without checking the leverage and got liquidated within hours. “More than losing the money, I felt ashamed that I could not control myself.” (translated)
9. Sep to Oct 2026: from notes to a system. Everything went into one Google Drive folder: a charter, thesis notes, a macro map, execution rules, records and quarterly reviews. Every decision got one tag: complied, claimed exception, or violated. A good outcome did not turn a claimed exception into compliance. An AI helped me write and asked the awkward questions, but my own words stayed as I said them.

What I take from it

Records fixed my memory. Memory did not err in one direction. Long, slow losses shrank. Sharp shocks grew. Memory leaned wherever my feelings leaned at the time.

Records showed me the loop. The 2021 reveal, the 2024 recovery, the 2026 short wins. Three times I read luck as skill, and three times the next bet got bigger.

Records made waiting count. Days when I wanted to buy and did not were logged too. My quarterly template says it plainly: not acting is a decision, and waiting is a position.

Where I could be wrong

• Records written after the fact still carry today’s interpretation, even when memory, files and guesses are labeled separately.
• Records do not stop violations. They stop hiding them. In October 2026, after building all this, I still moved all my stablecoins into ETH in one go.
• Building a record system is fun enough to feel like investing. It is not. More documents do not mean better decisions.
• My recovery was luck. This is not a story about records making money. I built them so that when luck runs out, I would not go back to zero.

Not financial advice. This is my personal record, not a recommendation to do what I did.

Written 2026-11-04 15:55 KST. BTC about $77,600, ETH about $2,300 (market price at time of writing, rounded).

Sources

• My investing records, Google Drive (2021-12 to 2026-10, private)
• Federal Reserve FOMC statement, March 16, 2022