Do you track your overall portfolio P/E? Running into the arithmetic vs harmonic mean problem

Curious how you guys actually evaluate the collective valuation of your portfolio in your spreadsheets or tools.

I've been working on my portfolio valuation engine lately, and kept running into a massive distortion whenever people try to calculate an aggregate Price-to-Earnings multiple using a standard weighted average.

This is the problem:

Suppose you invest $1,000 total. You put $500 into Company A (trading at a 10x P/E) and $500 into Company B (trading at a 100x P/E).

If you take a standard weighted arithmetic average, you get:
(0.50 * 10) + (0.50 * 100) = 55x P/E.

Looking at 55x makes it feel like your portfolio is sitting at extreme bubble valuations. But when you look at the actual cash earnings your $1,000 bought:

Company A ($500 at 10x) generates $50 in annual earnings.
Company B ($500 at 100x) generates $5 in annual earnings.

Your total earnings across the $1,000 invested are $55. That translates to an actual aggregate multiple of 18.18x ($1,000 / $55), or a 5.5% earnings yield.

The standard arithmetic average overstates the valuation multiple by over 200% simply because price sits in the numerator, heavily weighting the outlier.

Because of this, I moved my models strictly to the weighted harmonic mean (summing the weighted earnings yields E/P and taking the reciprocal), which seems to be what institutional databases like Morningstar use.

For those of you holding 10 to 20+ individual value picks:

Do you bother tracking your collective portfolio P/E, or do you view whole-portfolio multiples as noise and strictly look at individual ticker DCFs?

If you do track it, do you use harmonic weighting, aggregate dollar earnings yield, or something else entirely? Also curious how you treat negative P/E turnarounds in the mix.

Curious to hear how others here handle this in their own tracking and models.