{"id":"2078500058308194305","url":"https://x.com/Damir_Akaza/status/2078500058308194305","text":"","author":{"name":"Damir Akaza","username":"Damir_Akaza","avatarUrl":"https://pbs.twimg.com/profile_images/2070665160948080640/cuJWnmQV_200x200.jpg"},"createdAt":"Sat Jul 18 15:20:05 +0000 2026","engagement":{"replies":8,"retweets":24,"likes":98,"views":210932},"article":{"title":"How the Best Traders Cut Losses: a 10,000-Account Study, the Math of Recovery, and Exit Rules","previewText":"Investors sell winning positions 1.5 times more readily than losing ones. The winners they sell then outperform the losers they keep by 3.4% over the following year. That comes from a study of 10,000","coverImageUrl":"https://pbs.twimg.com/media/HNggOriWEAAzBR5.jpg","content":"Investors sell winning positions 1.5 times more readily than losing ones. The winners they sell then outperform the losers they keep by 3.4% over the following year. That comes from a study of 10,000 real brokerage accounts, and I break it down below.\n\nAlso in this article: what holding a loss actually costs, in numbers (a recovery table, including time), why the brain makes this choice automatically (measured by Kahneman and Tversky), two cases worth a combined $30 billion, and the specific exit rules used by Paul Tudor Jones, Soros, Seykota, O'Neil and Druckenmiller. At the end: a five-step exit protocol you can run on your portfolio today.\n\n## 1. The study that caught everyone red-handed\n\nIn 1998, Berkeley professor Terrance Odean got access to the records of 10,000 accounts at a large discount broker, covering 1987 to 1993, and counted what clients actually did.\n\n![](https://pbs.twimg.com/media/HNghQi4WIAA6-DA.png)\n\nThe results:\n\nWhen investors had a gain available to take, they took it 14.8% of the time. When they had a loss available to take, they took it 9.8% of the time. A winning position got sold 1.5 times more readily than a losing one.\n\nThe winners they sold went on to outperform the losers they kept by an average of 3.4 percentage points over the next 12 months. People were systematically selling strength and holding weakness.\n\nThe only month the pattern broke: December. Ahead of tax season, losses got realized far more readily, because an external reason appeared and took the pain out of the decision. Which tells you the problem was never analysis. Give the finger an excuse, and it presses the sell button calmly.\n\nThe pattern is called the disposition effect, first described by Shefrin and Statman in 1985. It has since been confirmed on data from the US, Finland, China and Israel, and in professional money managers as well.\n\n## 2. The price tag: the math of recovery\n\nA loss and its recovery are not symmetric.\n\n![](https://pbs.twimg.com/media/HNghxO4WsAAvTea.png)\n\nHere is the full table, the one worth saving:\n\n![](https://pbs.twimg.com/media/HNgh2NOXEAAInL7.jpg)\n\nRead the bottom row again. After a 90% loss you need a 10x just to get back to zero. Same person, same skill set that just produced the 90% loss.\n\nNow the same thing measured in time. Say you can reliably compound at 20% a year, the level of the best funds in the world. Recovering a 10% loss takes about 7 months. A 30% loss: almost 2 years. A 50% loss: 3.8 years. A 70% loss: 6.6 years.\n\nThe takeaway is short. A 7-10% loss costs months. A 50%+ loss eats years of a strong trader's life. The entire job of risk management comes down to one line: never let a position cross from the first zone into the second.\n\n## 3. Why the brain chooses to hold\n\nKahneman and Tversky measured it. In their 1992 paper, the loss aversion coefficient came out at 2.25. Translation: losing $1,000 feels as strong as winning $2,250. Kahneman later received the Nobel Prize for this line of research.\n\n![](https://pbs.twimg.com/media/HNgiIYmW0AA1Cml.png)\n\nThe second mechanism: while a loss only exists on the screen, the brain files it as a temporary state. The moment you press sell, it becomes a final fact, and that is when the double-strength pain arrives. So holding a losing position feels like postponing pain, while Odean's data from section 1 shows that on average it grows the loss instead.\n\nKnowing the mechanism does not cure it. Odean found the effect in experienced investors too. The only thing that works is an external system of rules, which is what sections 5 and 6 are for.\n\n## 4. Two cases: $1.4 billion and $30 billion\n\nSame mechanism, different number of zeros.\n\n![](https://pbs.twimg.com/media/HNgiU-dX0AA2q6Q.png)\n\nBarings, 1995. Nick Leeson ran trading at the Singapore office of Barings, the oldest merchant bank in England: 233 years old, banker to the royal family. His Nikkei futures positions went into the red. Leeson hid the loss in internal account 88888 and started adding to the position to trade his way back. On January 17, 1995, the Kobe earthquake crushed the Nikkei. He bought more. The final bill: 827 million pounds, around $1.4 billion. On February 26, 1995, the bank was declared insolvent and sold to ING for 1 pound.\n\nArchegos, 2021. Bill Hwang's family office: roughly $20 billion in capital, concentrated positions with up to 5x leverage through swaps. In March 2021, the stocks in the book started falling. Instead of cutting, Hwang defended and added. Within days the office's capital went to zero, and the counterparty banks lost about $10 billion between them: Credit Suisse wrote off $5.5 billion, Nomura $2.9 billion. Two years later, Credit Suisse ceased to exist as an independent bank.\n\nBoth times, the first mistake cost survivable money. The multi-billion bill came from the decision to add to a losing position instead of exiting it.\n\n## 5. The rules the best actually use\n\nOnly specifics here, name by name.\n\nWilliam O'Neil (founder of Investor's Business Daily, studied every top-performing stock since the 1880s): sell any stock at a loss of 7-8% from your purchase price. No exceptions, no second opinion. His logic is visible in the table from section 2: at that level, the cost of a mistake is still measured in months.\n\nLarry Hite (Mint Investment, one of the pioneers of systematic trading): never risk more than 1% of capital on a single trade. At that size, even ten losses in a row leave the account fully operational.\n\nPaul Tudor Jones. Three of his rules are documented on film and in interviews. One: the sheet of paper above his desk. Two: only enter where the potential reward covers the risk 5 to 1, so you can be right just 20% of the time and still come out ahead. Three: the 200-day moving average as the last line of defense, below it he does not hold. His summary of the mindset: he spends his day thinking about how much he can lose, not how much he can make.\n\n![](https://pbs.twimg.com/media/HNgi5nJWEAE5pS-.png)\n\nEd Seykota, asked in Market Wizards to name the elements of good trading, gave three: cut losses, cut losses, cut losses. Follow all three and you may have a chance.\n\nGeorge Soros. October 1987: he misread the crash and took a loss in the hundreds of millions. Instead of proving himself right, he liquidated within days, accepted the damage and moved on. Quantum still finished 1987 up around 14%. His principle: it does not matter how often you are right. What matters is how much you make when you are right and how much you lose when you are wrong.\n\nStanley Druckenmiller. March 2000: fully aware the dot-coms were a bubble, he cracked under the emotional pressure of watching everyone else get rich and bought $6 billion of tech stocks near the very top. Six weeks later he was down $3 billion. Here is the part that matters: he did not sit in it. He admitted the mistake, got out, and left Quantum. Asked later what the episode taught him, he said: nothing. He already knew you cannot do that. He simply broke emotionally. Even the owner of one of the best track records in history has breakdowns. The difference is the speed of admission.\n\n## 6. The exit protocol: 5 steps\n\n1. The stop is defined before the entry, and written down. A price or a condition under which the idea is officially dead. The calm version of you makes this decision, not the version staring at red numbers.\n\n1. Risk per trade: 1-2% of capital. Position size is calculated from the stop, not from the hoped-for profit. Formula: position size = (capital × 1%) / distance to the stop.\n\n1. One question over the whole portfolio, once a week: \"If I did not own this position, would I buy it today at this price?\" A \"no\" on an open position means the only thing holding it is the refusal to realize a loss.\n\n1. Averaging down is allowed only if it was in the plan before the entry, with pre-set levels and a total risk cap. An unplanned buy into a falling position repeats the Leeson and Hwang pattern in miniature.\n\n1. The phrase \"it will come back\" showing up in your own reasoning = an exit signal. When the argument for holding is hope instead of the original thesis, the trade is already over.\n\n## The bottom line\n\nOdean's data: people hold losses and sell gains, and it costs them 3.4% a year. The recovery table: under 10%, a loss costs months, past 50% it costs years. Kahneman's number: losses hurt 2.25 times more, so \"hold\" gets chosen automatically, before any analysis happens. Barings and Archegos: adding to a losing position is what turns a working mistake into a catastrophe. The rules of O'Neil, Hite, Tudor Jones and Seykota exist for exactly one purpose: to move the exit decision outside the moment of pain.\n\nEvery trader in this article took heavy losses. The ones who kept their accounts were the ones who cut fast.\n\nOne question from the article to run over your portfolio right now: \"If I did not own this position, would I buy it today at this price?\" If the answer is no and the position is still open, you already know what is holding it."}}