We track the quarterly filings of 33 large investors: Buffett's Berkshire, Griffin's Citadel, Dalio's Bridgewater, Ackman's Pershing Square, Cathie Wood's ARK, Cohen's Point72, Renaissance, Millennium, D. E. Shaw, Tiger, Coatue, the Gates Foundation.
Here is what they bought, what they dumped, roughly what they paid, and whether you can still get a better price than they did.
What they are buying
Biggest increases between the December and March quarters, by dollars committed:
| Ticker | Was | Now | Shares |
|---|---|---|---|
| $GOOG | $39.0bn | $46.7bn | +30% |
| $DAL | $666m | $3.3bn | +390% |
| $CAT | $9.7bn | $12.1bn | +2% |
| $CRWV | $2.2bn | $4.6bn | +92% |
| $TSM | $16.9bn | $19.2bn | +2% |
| $MU | $5.0bn | $7.1bn | +20% |
Delta at +390% is the standout, going from a rounding error to a $3.3 billion position in one quarter. CoreWeave nearly doubled. And they added a fifth more Micron.
What they are dumping
| Ticker | Was | Now | Shares |
|---|---|---|---|
| $MSFT | $36.0bn | $23.8bn | -14% |
| $AMZN | $28.7bn | $23.4bn | -9% |
| $NVDA | $42.4bn | $37.5bn | -5% |
| $WMT | $12.3bn | $7.9bn | -42% |
| $V | $10.6bn | $6.4bn | -28% |
| $BAC | $33.8bn | $29.8bn | -1% |
The rotation in one line: out of mega-cap software and consumer, into search, airlines, industrials and AI infrastructure.
One thing to be precise about. A trim is not a dump. Nvidia came down 5% and is still a $37.5 billion position, the second largest in the group. Microsoft losing $12.2 billion is the biggest single move here, and Microsoft remains an enormous holding.
The brand-new positions
Names that did not exist in their books a quarter earlier, largest first: $VSNT ($522m), $FPS ($280m), $INDV ($276m), $MANE ($238m), $EQPT ($185m), $PAYP ($166m).
This is where the real risk-taking is. A $522 million position built from zero is a decision, not a rebalance.
Can you still get their price?
This is the fun part. We know roughly what they paid, because we know which quarter they bought in. Here is what the stock actually traded at during that quarter, against where it is now.
| Ticker | Their quarter's range | Rough average | Now | You are paying |
|---|---|---|---|---|
| $WMT | $109.81 to $134.69 | $122.77 | $115.27 | 6.1% LESS |
| $GOOG | $271.54 to $350.15 | $314.16 | $343.54 | 9.4% more |
| $V | $294.32 to $358.62 | $321.55 | $364.15 | 13.2% more |
| $MSFT | $356.28 to $489.70 | $418.44 | $495.40 | 18.4% more |
| $AMZN | $196.00 to $248.94 | $220.17 | $262.65 | 19.3% more |
| $CRWV | $67.15 to $114.45 | $86.78 | $105.26 | 21.3% more |
| $NVDA | $164.27 to $197.63 | $183.46 | $225.16 | 22.7% more |
| $CAT | $575.31 to $789.81 | $692.61 | $856.57 | 23.7% more |
| $TSM | $311.70 to $390.21 | $344.13 | $426.35 | 23.9% more |
| $BAC | $46.12 to $57.55 | $51.60 | $64.49 | 25.0% more |
| $DAL | $55.28 to $76.39 | $67.33 | $89.35 | 32.7% more |
| $MU | $294.86 to $471.34 | $391.72 | $971.66 | 148.0% more |
Eleven of twelve are more expensive than the institutions paid. So the honest answer to "am I too late" is, on this list, mostly yes.
Micron is the one that got away. They added a fifth more at around $392 and it is now $971.66. Whatever they saw, the market has since agreed enthusiastically.
Exactly one name is cheaper than their entry: Walmart, 6.1% below. Enjoy that with appropriate caution, because they were selling it. That is not a discount on a stock they liked. It is a discount on a stock they cut by 42%.
⚠ The average is the quarter's mean close, not their actual fill. A fund buying across three months lands somewhere inside that range, and the range is the honest part: Micron's quarter ran $294.86 to $471.34, so "around $392" could really be anywhere in there.
The catch, briefly
These filings describe the quarter ending 31 March and arrive 45 days later. That is the rule, not slowness: 64% of the filings we hold landed on the legal deadline exactly.
We ingest within a day of a filing hitting EDGAR, so this is as current as the form allows. But that still means a position opened in early January stayed invisible until mid-May.
And about one position in six is replaced each quarter. Of 25,139 positions held in December, 4,312 were gone by March and 4,169 new ones took their place. They are not selling, they are rotating, so a name you read about today has a real chance of already being gone.
One gap, stated rather than hidden. We hold the June quarter for only 6 of the 33 filers so far. The rest are complete through March, which is why every figure above uses March.
The trick: three forms that beat a 13F
If the lag bothers you, stop using 13Fs for timing. Three other forms are far faster and free.
| Form | Trigger | Speed |
|---|---|---|
| Form 4 | An insider buys or sells | 2 business days |
| Schedule 13D | Crossing 5%, intending to influence | days |
| Schedule 13G | Crossing 5%, passively | days to weeks |
These are event-driven. They arrive because something happened, not because a quarter ended.
Live example from this week. SpaceX listed on 12 June, so its first 13F-visible quarter is months away. But between 11 and 14 August, four Schedule 13Gs disclosed the entire ownership structure at once: Musk at 48.4%, Alphabet at 7.2% through Google LLC and XXVI Holdings, Antonio Gracias at 6.5%, Founders Fund around 2.2%.
Alphabet owning 7.2% of SpaceX is in nobody's 13F. It arrived days after the trigger, on a form most people never read.
So: 13Fs for context, Form 4 and 13D/G for timing.
Our smart money board carries these filers with their period dates attached, each filer's page reads positions across quarters, and the investors page is the same data by name.
This is not advice. It is what 33 large investors reported, roughly what it cost them, and what it would cost you today.