Gundlach is being vocal about how the 3% yield level for the 10y is a key level that could send stocks down. The issue is that he is using a variation of the 'Fed model' to predict stocks, which historically had serious deficiencies compared to other valuation tools like simple PE ratios Cliff Asness does a great job of explaining why one can't just look a nominal bond yields when assesing where stocks are headed in this paper http://faculty.som.yale.edu/jakethomas/otherpapers/Asness.pdf That paper should be required reading to anyone investing these days. There is so much nonsense floating around in terms of comparing stocks to bonds or talking about discounted rates of stock cash flows and the paper just debunks it all. Here are some selected quotes
Most everyone is talking their book, the only utility is the marks they set which shape market perceptions to a degree and incite orderflow when those triggers are approached, more often than anything those triggers become outliers where price has or yield has a hard time maintaining, which essentially becomes support or resistance based on which side price is on relative to those triggers.
So there are excess returns to be earned in period of high interest rates likely exactly because of the errors that investors make that Asness describe
The Fed model has ZERO correlation to 10y returns of equity markets. The supposed efficacy of the model is solely based on the 1980-2010 bond bubble which happened to occur hand in hand with one of the biggest bull markets in history. Correlation does not imply causation!
There's a lot of wiggle room for yields to rise before they impact prices. Here's a pretty solid and digestible blog post on it: https://brooklyninvestor.blogspot.com/2016/11/bonds-down-stocks-up.html