Yesterday I found this post by NYU professor Aswath Damodaron, which tried to reconcile the divergence between Mark Cuban saying diversification is for idiots and John Bogle (essentially) saying that diversification is all that matters, advocating owning the entire market. Like with all aspects of investing there is no single correct answer, simply a topic for each investor to explore and sort out for themselves.
Here I offer my approach for you to agree with or disagree with as you sort this out for yourself.
In terms of equities and using individual stocks or narrow ETFs, I am a big believer in not letting one position that goes wrong seriously damage the entire portfolio. As an exaggerated example, a 20% position in a stock that then cuts in half could be a serious problem depending on what the rest of the market is doing.
I've disclosed many times that our portfolio typically has 30-35 holdings, with most of the positions being targeted at 2-3% of the portfolio. There are several layers to this preference.
From a single-holding risk standpoint, if a three-percenter cuts half unexpectedly, then the drag obviously is only 150 basis points, which can be overcome elsewhere in the portfolio (not necessarily with a 1,000 basis point hit from one stock). In a portfolio of 30-35 stocks, it is plausible that in a given year one of the stocks held will go up 50-100%. This can be an "accident," but either way it's plausible but realize the one you might think would be up that much won't be the one that ultimately goes up that much. A two-percenter going up 50% adds 100 basis points the portfolio, if you pair that with a 3% yield for the portfolio then you already have a meaningful chunk of your return for an "average" year in the market.
From a smoothing-out-the-ride standpoint, I want enough holdings that I can take various types of attributes which potentially creates a zigzag effect in the portfolio. The importance here is that if your base case assumptions turn out to be incorrect, then some exposure to things that don't tie into your base case will minimize the consequence for being wrong.
Another element to this is that, as long-time readers will know, I think it is very important to build in various themes and countries into the portfolio. The objective in picking themes and countries is to build in what are hopefully some obvious tailwinds into the portfolio. As a totally made-up example, if we know trillions must be spent on wool from Laos, then you might want some exposure to wool from Laos.
Going too many holdings makes it difficult for a "winner" to matter, and going with too few holdings means a "loser" (and there will be losers) could damage the entire portfolio, as outlined above.