Hacker News Reader

Comment by kfarr | Hacker News Reader

> As the astute among you might have noticed, if you sum up all the stocks used in the analysis it would only come to 18.5k. I removed ~15% of the overall recommendations as either they did not have stock data present in Yahoo Finance/Alpha Vantage or the price data did not match with the one given on the Mad Money website.

That seems like a big caveat. Wouldn’t lack of stock price data be an indicator of delisting?

View on HN

Yes, good catch, not taking delisted stocks into account is probably the most common newbie backtesting mistake.

Maybe so. But delistings probably wouldn't have a big impact on the 1-day results, and the analysis already concludes that Cramer's long-term returns are not good.

No, it's quite possible that delisted stocks behaved differently even on 1-day basis. For example, if I recommended only very risky stocks, the ones that stayed listed and survived likely paid off very well (let's say +40% in a year), while the ones that are no longer listed mostly went bankrupt (-100%). 1-day returns would show a very large difference on average.

In theory that could be true. But as others have pointed out, there are many other reasons for delisting besides bankruptcy. The scenario you paint is possible, but I still think it's unlikely---less than 15% of the stocks involved were omitted, and that includes not only delisted stocks but stocks omitted for other reasons, so the bankrupt firms would likely be much less than 15% of the total. And while stocks are unlimited on the upside, they can't go below zero on the downside, so there's a limit to how much those omitted losers could drag down the big outlying winners.

> Wouldn’t lack of stock price data be an indicator of delisting?

That's one reason. They may not have pricing due to licensing issues at the source, as well. Yahoo Finance is pretty good for a free resource, but it's not as comprehensive as paid sources.

But, even if it's due to delisting, delisting is not inherently a negative thing. It can happen due to mergers/acquisitions or a formerly public company going private, for instance.

It's not inherently a negative thing, but it's definitely a negative for this guys' analysis if none of that was accounted for.

Or they merged or changed a name, or were taken private. Searching for the SPAC PSAC shows nothing on Yahoo, because it’s now FFIE (Faraday Future) The poster was also measuring sell recommendations so those would be also be harmed by no data in delistings.

Yep, there's a reasonable chance this analysis suffers survivorship bias. Outliers whose stock-prices severally crashed are less likely to be represented. Hence, impactful sell recommendations are probably missing from the data, explaining why sell-side recommendations don't appear to to perform as well.

Reply on news.ycombinator.com