Losses are cheaper to learn from when they are someone else's. This is the practice of keeping a written record of both kinds so the same error cannot arrive twice wearing a different name.
13 min read · Level 1
Experience is usually described as the thing that makes investors better, and that is only half true: experience without a written record mostly produces confidence, because the losses get reinterpreted and the lucky wins get remembered as judgement. What actually improves is an accurate account of what you believed before you knew, which almost nobody keeps, because keeping one means regularly reading evidence that you were less right than you now feel. The reason to build the library out of other people's losses first is not modesty - it is throughput. You will personally experience perhaps a dozen investment mistakes worth learning from in a lifetime, which is far too small a sample to learn from, and the entire history of documented financial failure is available for the cost of reading it.
A mistake library is a written, growing record of specific errors - yours and other people's - organised by the mechanism that caused them rather than by the asset they happened in. It is the practical counterweight to the fact that investing gives terrible feedback: results arrive years late, mixed with luck, and in small enough numbers that you cannot learn from them by pattern recognition alone.
Each entry is short. What was decided, what was believed at the time, what the mechanism of the loss turned out to be, and what would have to be true for the same mechanism to reach you. Organising by mechanism is the part that does the work: 'lost money in property' teaches nothing, while 'the position could not be exited and an unrelated cash need forced the sale' is a category you will meet again in an entirely different asset.
Because the five preceding lessons are not five separate risks - they are five mechanisms, and mechanisms repeat across asset classes while stories never do. The leverage that wipes out a margin account is structurally the same as the leverage inside a fund you bought for its name. The illiquidity that traps a private fund is the same mechanism as the property you cannot sell in a downturn.
And because memory is an unreliable narrator about your own decisions. After an outcome is known, people reliably remember having been less certain than they were, and remember reasons they did not actually hold at the time. A note written before the outcome is the only version of your thinking that the outcome cannot edit.
Educational, not personalised financial advice. Nothing in the Investor Academy is a recommendation to buy, sell or hold anything — what is right for you depends on your own circumstances and the rules where you live.
What one difference is worth, stated plainly. A portfolio compounding at 8% a year for 20 years multiplies by 1.08 to the twentieth power, which is 4.66 times the starting amount.
Now suppose everything is identical except that in year 10 one mechanism - a concentrated position, a forced sale, a fraud - permanently removes 40% of the portfolio. The remaining capital still compounds at 8% for the whole period, so the ending figure is 4.66 multiplied by 0.60, which is 2.80 times the starting amount.
The growth rate never changed. Twenty years of decisions were the same. The single event costs 40% of the final result and keeps costing it for every year that follows, which is why avoiding one category of loss can be worth more than improving the returns on everything else.
The mistake in keeping a record is writing only about losses. A library of losses teaches you to be afraid, which is not the same as teaching you to be accurate - and it quietly drops every case where a decision was sound and the outcome was poor anyway. Both belong in the record, labelled separately.
The other mistake is filing by asset. Entries organised as 'crypto', 'property', 'startups' produce the conclusion that a category is dangerous, which is both wrong and useless. Entries organised as 'I could not get out', 'I sized it by conviction rather than by consequence', 'I checked the person rather than the structure' produce a warning you will actually recognise the next time it arrives.
You do not have enough personal losses to learn from, and the ones you have are emotionally expensive to reread. Public failures, documented collapses and the losses friends are willing to describe are the cheapest material available, and they cover mechanisms you have not personally met yet.
Give each entry a mechanism heading that could apply in any asset class: could not exit, borrowed against it, too large a share of everything, trusted the introducer, bought the narrative at any price, no independent custody. The heading is the reusable part.
Record what was known at the time and what was assumed. 'The valuation was too high' is hindsight. 'The price required a 40% market share and nobody in that industry had held more than 15%' is a note you can apply again.
Label each entry as good decision / bad outcome, bad decision / good outcome, or aligned. The two off-diagonal categories are where the learning is - and they are the two everyone leaves out, because one feels like an excuse and the other feels like modesty.
For anything new, record the reasoning and the falsifier at the moment of commitment. This is the only entry the eventual result cannot rewrite, and it is what makes an honest review possible later.
Set a fixed date - annually is enough - and read the whole library before making any large new commitment. The entry you need is never the one you happen to remember; it is the one whose mechanism matches what is in front of you, which is why it has to be read rather than recalled.
Continuously, and specifically before committing to anything materially larger than usual, or anything in a category you have not invested in before.
This is not a substitute for understanding the assets themselves. A rich collection of failure patterns will stop you repeating known mistakes and will do nothing to help you judge whether a particular business is any good.
Five reusable headings come straight out of the preceding lessons, and each one has occurred in many different asset classes: the returns were not produced by the stated activity (no independent custody, no losing months); a price fall alone could force the sale (borrowed money, a maintenance requirement); the exit did not exist when it was needed (lock-ups, no secondary buyer, a redemption gate); one outcome carried too much of the total (a single company, sector, employer or currency); and the price already assumed the story (a correct narrative bought at a level that required it to be several times better).
None of the five names an asset class, which is precisely why each of them will turn up again somewhere you were not expecting it.
A common shape in portfolio reviews is an investor who studies every position that lost money and none that made money. The winners are examined only for confirmation that the original reasoning was sound. Where a position was bought for one reason and rose for an entirely unrelated one, that goes unnoticed - and the reasoning gets credited with a result it did not produce, then reused.
A process that only investigates losses will keep a flawed method for as long as that method keeps getting lucky.
Create a free account to read the rest and apply it to your own investing.