Q3 2026: The Quarter I Mostly Held

I did not do much in Q3, which is probably the most accurate summary of the quarter.

A lot of the decisions that mattered had already happened before July. I had spent Q1 and Q2 building Microsoft and Meta, so by the time Q3 started there really was not much left to do other than keep checking whether the reasons I owned them were still intact.

That sounds simple, but it is surprisingly easy to create work for yourself once a position starts moving. A stock is up a lot, the valuation changes, another idea starts looking interesting, and suddenly there are ten possible things you could do even though none of them are obviously better than just leaving the original decision alone.

Most of the time, that is what I did.

The Q3 scorecard

The quarter ended with $318,738 invested across accounts. The active Fidelity accounts finished at $235,557.

Using the reconstructed June 30 active balance, the dated Q3 external cash flows and the September 30 closing balance, the active portfolio returned about 5.2% in Q3 after adjusting for cash flows. VOO returned about 2.3% over the same quarter including its September distribution.

Item	Q3 result
Ending invested assets	$318,738
Active Fidelity ending value	$235,557
Active Fidelity Q3 return	~+5.2% flow-adjusted
VOO Q3 total return	~+2.3%
401(k) net transactions	$8,201.97
401(k) gains / losses	+$525.97

The exact invested-asset number is a little lower than the rough $320K snapshot I had on September 29. Nothing changed in the portfolio after the BRBR / FXAIX deployment; September 30 just moved the value around.

The 401(k) numbers are another useful reminder of why I keep contributions and returns separate. Money going into the account is good, obviously, but it is not investment performance.

The win was mostly leaving good decisions alone

If I graded Q3 by whichever stocks happened to go up the most, this section would be easy and not especially useful.

The better process example was Microsoft and Meta. I accumulated both mostly across Q1 and Q2, got to Q3 with meaningful positions in businesses I still liked, and then mostly let them sit there while the underlying theses continued to play out.

That ended up being more important than it sounds. Once a stock works, there is always another reasonable-sounding question available. Should I trim it? Is the multiple too high now? Should I rotate into something cheaper? Is there another company with more upside?

Sometimes one of those questions deserves an answer. In Q3, I mostly did not find a better answer than leaving the positions alone.

I still reviewed the businesses and the valuations. I just did not need every new piece of information to turn into a trade, which is probably the cleanest example this quarter of what I want Mostly Nothing to be about.

GOOG was not a bad investment. The process could have been better.

My clearest process mistake was probably the September Roth rotation from Microsoft into Alphabet, although even calling it a mistake needs some context because I still like the investment.

I bought GOOG around $326.83. My current base case gets to roughly $619 per share in five years, which works out to about 13.6% annualized from the actual purchase price. If Alphabet compounds anywhere near that, great. The issue is the process that got me there, not whether a good outcome would feel nice.

What I would change is the way I framed the decision beforehand. Microsoft was already a meaningful position, Alphabet was smaller, and adding GOOG improved the balance between the big compounders. That was a real reason to make the rotation.

The weak spot was the opportunity-cost comparison. I did not push it far enough.

The more useful question was not simply whether GOOG looked attractive. It was whether that roughly 13.6% base-case return was better enough than keeping MSFT, buying the index, or adding somewhere else to justify moving the Roth capital.

I got to “this is attractive” a little faster than “this is the best use of the money.”

That distinction is worth keeping even if GOOG works out perfectly well. A good result would not retroactively make the comparison more rigorous than it was at the time.

One decision I am not grading yet

BRBR entered the portfolio on September 29, when I put $5,000 into BellRing around $7.63 and sent the other $5,000 from the matured CD into FXAIX.

There is not much point pretending I know whether that was a good investment two days later.

The sizing is easier to judge than the investment outcome. The upside looked interesting enough to own, while the uncertainty around the margin recovery was still too high for BRBR to become a major portfolio bet right away.

At roughly 1.6% of the September 29 portfolio snapshot, it is large enough to matter if the thesis works and small enough that I have room to be wrong without one idea changing the character of the whole portfolio.

That feels like a reasonable place to start. Q4 should give me a lot more information about whether the actual thesis deserves to survive.

The portfolio-level change was more gradual

The bigger Q3 change was not really about one ticker. My index allocation has been drifting lower, and it happened slowly enough that there was never one obvious moment where I felt like I had changed the portfolio.

Some of that is a good problem to have. Active positions appreciated. Some came from deliberately building individual names, and some came from adding small asymmetry positions that I thought deserved a place.

Taken one at a time, most of those decisions still make sense to me. The combined effect is what I need to pay more attention to.

That is probably the most useful portfolio-level lesson from the quarter. Allocation drift does not usually happen because you wake up one morning and decide to take a lot more active risk. It happens because a series of reasonable decisions slowly changes the weights.

An arbitrary rule that exactly 50% of the portfolio always needs to be indexed feels too mechanical for what I am trying to do.

What matters more is keeping the index sleeve large enough that it remains the natural alternative every time another stock asks for capital. Going into Q4, that probably means new money leans more toward the index unless something else looks clearly better after the full expected-return and sizing work.

The quarter ended with two useful sizing examples

The September 29 BRBR/FXAIX split was one version of that. I had done enough work on BellRing to think the upside could be meaningful if margins recover, but I still split the money because I was not confident enough to make the whole $10,000 one company-specific bet.

Then literally the next day Micron reported one of the most ridiculous quarters I have seen. Q4 non-GAAP EPS was $33.42 and Q1 guidance was $38.15, but the more important part for me was that management pushed the tight supply/demand setup through 2028 and added enough contractual visibility that I raised my normalized FY2029 earnings assumptions.

My view of the business improved, but I still did not add to the position because MU was already large.

I like that contrast. BellRing looked cheap enough to tempt me into going bigger, while Micron delivered enough good news to tempt me into adding again. In both cases, the better answer was probably to let position size matter as much as conviction.

Q4 probably starts with the boring answer

My default destination for new capital in Q4 is likely to be the index, mostly because I already own plenty of businesses I like and the index sleeve could use some rebuilding.

That does not mean I suddenly prefer indexing to stock picking. It just means the next individual-stock purchase should probably need to look meaningfully better than the simple alternative of adding to the broad market.

Capital destination	What would make it interesting
Broad index	Default while the index sleeve rebuilds
Existing compounders	Expected return improves enough to justify adding, with room in the position
AI infrastructure	Better expected return without making correlated exposure uncomfortably large
Asymmetry	Better evidence behind the thesis; a lower quote by itself is not enough
Cash	Fine if nothing else looks especially compelling

That is not meant to be a rigid ranking for the next three months. It is just the starting point, which is useful because otherwise every interesting company can start feeling like it deserves money.

What I actually want to watch in asymmetry

AMPX, INFQ, RKLB and now BRBR can all move enough in a week to make it feel like something important happened. Sometimes something did, but often the stock price is just moving faster than the underlying evidence.

For those positions I care much more about contracts, commercialization, production, financing, dilution, customer validation and competitive changes than I do about whether the quote is up or down 15% in a week.

That uncertainty is exactly why the positions stay small. Better to wait for the business evidence to improve or deteriorate than let the chart do the thinking for me.

Earnings season is where the work should happen

The other Q4 priority is to let company results do more of the talking.

Earnings are the cleanest recurring checkpoint I have for whether the underwriting is actually moving: revenue or unit growth, margins, capital allocation, balance-sheet changes, and whether something that looked temporary is starting to look structural.

Micron’s September 30 report was a good example. The headline EPS was absurd, but the thing that changed my model was the duration: tighter conditions through 2028, more committed output, more contractual visibility and a clearer cash-return setup.

That is the kind of information I want to react to. A stock being down 7% on a random Wednesday is much less useful.

So Q4 is less about trying to anticipate every move and more about using earnings to keep asking the same question: did anything important actually change?

What carries into Q4

The main takeaway from Q3 is that doing nothing can be a perfectly good portfolio decision when the original work is still holding up. Microsoft and Meta did not need constant intervention from me just because the stocks were moving.

GOOG gave me a different lesson: a good expected return can still deserve a better opportunity-cost comparison before I move capital. And the portfolio as a whole reminded me that allocation can drift even when the individual decisions still look reasonable.

Heading into Q4, the plan is to make the index a stronger default again, keep the asymmetry positions tied to actual thesis changes, and let earnings season tell me where the underwriting needs to move.

If something clearly deserves the next dollar, great. If not, I already have plenty invested.