Investing channels highlight their best calls and let the worst ones fade. This article explains how survivorship bias shapes what viewers see and how dated records correct for it.
Imagine a channel that discusses fifty stocks over a year. Some of those ideas will do brilliantly, some will go nowhere and some will fall hard. Twelve months later, which ones will the creator mention again? The answer is obvious, and it is the reason investing content is so difficult to evaluate honestly.
Survivorship bias is the tendency to focus on the things that made it through a selection process while forgetting the things that did not. In investing content, the selection process is the creator’s own editorial choice about what to revisit. Winners get a follow-up video titled “I told you so.” Losers get silence.
How the bias is built into the format
This is not primarily a character flaw. It is a feature of the medium. A video celebrating a 150% gain performs well. A video explaining why a 45% loss happened performs badly, and it also invites uncomfortable comments. The algorithm rewards the first and punishes the second, so over time even scrupulous creators drift towards showing the highlights.
There is a second layer. Creators talk about far more stocks than they hold. Watch lists, “stocks I am looking at”, “companies on my radar” and hypothetical valuations generate dozens of mentions that never become positions. When one of those names doubles, it is easy to retroactively frame the earlier mention as a call. When one of them halves, it was “just a watch list.”
The illusion of a great record
Put those two effects together and you get an audience that believes a creator has a far better hit rate than they actually do. The believers are not being foolish. They are responding accurately to the evidence they have been shown. The problem is that the evidence has been filtered before it reached them.
Consider what a viewer might see over a year on a typical channel: a triumphant retrospective on a semiconductor pick that rose 150%, a follow-up on an energy infrastructure name that rose 40% and a portfolio update focusing on the positions in the green. What they would not see is the quiet exit from a payments stock that fell 20%, the “long-term mistake” in a software company down 28% or the beverage stock down 45% that has not been mentioned since spring. Each of those omissions is individually defensible. Together they produce a fiction.
Why the fix has to be independent
A creator cannot correct their own survivorship bias, for the same reason a student cannot fairly grade their own exam. Even with the best intentions, the choice of what counts as a call and what counts as a passing remark will lean in their favour. The only real correction is a record kept by someone else, using fixed rules applied before the outcome is known.
That means logging every dated quote about a stock, not just the memorable ones. It means recording the price on the day, not on a flattering day nearby. It means tracking outcomes for the losers as diligently as for the winners, and it means keeping the record visible even when the creator would prefer it forgotten.
This is a large amount of unglamorous work, which is why it has rarely been done at scale. A new project called They Said Buy has taken it on for a growing set of investing YouTubers: hundreds of dated quotes across dozens of creator-stock pairs, each linked to the exact second in the source video, with the price change since the first forecast shown next to the S&P 500 over the same dates. Crucially, the pages that show a stock down 30% or 45% sit right beside the ones showing a stock up 100%.
What a complete record reveals
When you look at unfiltered timelines, a few patterns become visible that highlight reels hide.
• Dispersion is enormous. The same creator can be up 150% on one name and down 45% on another within the same year. Averages tell you little; the distribution tells you a lot.
• The market explains a great deal. Many “wins” shrink to nothing once the S&P 500’s move over the same window is subtracted.
• Language predicts outcomes less than you would hope. The most confident quotes are not reliably attached to the best results.
• Reversals are common and usually unannounced. Creators drift from “buying” to “not buying more” to “sold” without a video explaining the change.
None of this means creators are useless. It means their value lies in the reasoning they share, not in their implied batting average, and viewers should weight the content accordingly.
Guarding yourself against the bias
If you consume investing content, a few habits will protect you.
First, treat every mention as a potential call and keep your own list. When a creator says “I like this one”, write the ticker and the date down. Six months later, check it. You will quickly discover the gap between the record they present and the record they have.
Second, pay attention to what disappears. A stock that was a recurring topic and then vanishes is usually a loser. Its absence is information.
Third, distrust retrospectives. A video built around “my best calls of the year” is by construction the least representative sample of a creator’s judgement. The question worth asking is what the worst calls were, and whether they were acknowledged.
Fourth, use independent trackers where they exist. The point is not to catch anyone out. It is to see the full distribution of outcomes, which is the only basis for deciding how much weight a voice deserves.
The broader lesson
Survivorship bias is not unique to YouTube. It shapes how we think about hedge funds, start-ups, athletes and authors. What makes investing content unusual is that the correction is so easy in principle: everything was said in public, on a dated recording, about an asset with a public price. The data to build a complete record has always existed. All that was missing was someone willing to collect it and keep it in view.
Now that such records exist, the excuse for evaluating creators by their highlight reels is gone. Look at the whole timeline, losers included, and you will see a very different picture from the one on the thumbnail.
