Reviewed: November 19, 2026 · Data window: March 31 to September 30, 2026 · Sample: 17 institutional 13F reporting groups, 1,179 manager–security observations comparable across all three quarter-ends
Did the fixed 17-manager panel keep expanding the same disclosed positions after Q2? Only just over half did. Of 530 three-quarter-comparable positions whose split-adjusted share count rose from Q1 to Q2, 279 rose again from Q2 to Q3—a 52.64% continuation rate. The rate falls to 43.78% after removing Bridgewater, which supplied 198 of the panel’s 279 repeated increases. The headline result therefore describes observation counts inside this disclosed panel, not broad institutional conviction.
A repeated quarter-end increase is stronger evidence of a continuing disclosed pattern than a one-quarter appearance, but it is still not a transaction record. Form 13F does not reveal the trade date, cost basis, motive, short exposure, or the manager’s full portfolio.
Three findings
- 279 of 530 prior expanders increased again. Another 240 reported fewer shares in Q3 and 11 were unchanged, so a single quarter of expansion did not reliably establish a durable pattern.
- Bridgewater materially changed the aggregate answer. It accounted for 65.09% of prior expanders in the matched sample and 70.97% of repeated expansions. The panel-wide 52.64% rate becomes 43.78% without it; the equal-weight manager mean was 50.31% and the median 52.05%.
- Broad agreement was rare. The 279 repeated manager–security increases covered 265 unique securities. Only 13 securities were repeated expanders for at least two managers, and NVIDIA was the sole security with three: ARK, Altimeter and Appaloosa.
What happened to the Q2 expanders?
| Q1→Q2 direction | Q2→Q3 decrease | Q2→Q3 unchanged | Q2→Q3 increase | Total |
|---|---|---|---|---|
| Increase | 240 | 11 | 279 | 530 |
| Decrease | 250 | 25 | 259 | 534 |
| Unchanged | 23 | 88 | 4 | 115 |
| Total | 513 | 124 | 542 | 1,179 |
The 52.64% continuation rate is close to an even split, but the transition matrix shows why “continued buying” would be too strong. A 13F gives three snapshots. It cannot show whether a change occurred in one transaction or many, when it occurred inside the quarter, or whether an outside manager caused the change in a jointly reported position.
Manager results: denominators matter
| Manager | 3-quarter comparable | Q2 expanders | Expanded again | Continuation |
|---|---|---|---|---|
| Dalal Street | 3 | 1 | 1 | 100.0% |
| Icahn reporting group | 9 | 1 | 1 | 100.0% |
| Situational Awareness | 4 | 1 | 1 | 100.0% |
| Coatue | 53 | 8 | 5 | 62.5% |
| Altimeter | 10 | 5 | 3 | 60.0% |
| Bridgewater | 643 | 345 | 198 | 57.4% |
| Berkshire Hathaway | 28 | 7 | 4 | 57.1% |
| Light Street | 15 | 9 | 5 | 55.6% |
| ARK | 160 | 73 | 38 | 52.1% |
| Third Point | 14 | 8 | 4 | 50.0% |
| Whale Rock | 25 | 6 | 3 | 50.0% |
| Tiger Global | 36 | 6 | 2 | 33.3% |
| Greenlight / DME | 32 | 10 | 3 | 30.0% |
| Soros Fund Management | 97 | 33 | 9 | 27.3% |
| Appaloosa | 15 | 10 | 2 | 20.0% |
| Fundsmith | 27 | 3 | 0 | 0.0% |
| Pershing Square | 8 | 4 | 0 | 0.0% |
The observation-weighted panel rate and the typical-manager result answer different questions. The 52.64% rate gives every manager–security observation the same weight. The 50.31% mean and 52.05% median first calculate a rate for each manager, then give each manager equal weight. Neither makes Bridgewater’s broad filing “wrong”; the two views simply expose how panel composition changes the story.
Only NVIDIA reached three-manager persistence
| Security | CUSIP | Managers expanding in both intervals |
|---|---|---|
| NVIDIA | 67066G104 | ARK; Altimeter; Appaloosa (3) |
| Delta Air Lines | 247361702 | Berkshire Hathaway; Bridgewater (2) |
| Micron Technology | 595112103 | Coatue; Soros Fund Management (2) |
| Broadcom | 11135F101 | Coatue; Light Street (2) |
| Amazon | 023135106 | ARK; Altimeter (2) |
| Block | 852234103 | ARK; Bridgewater (2) |
| Viavi Solutions | 925550105 | Bridgewater; Whale Rock (2) |
| CRH | G25508105 | Bridgewater; Third Point (2) |
| Eli Lilly | 532457108 | ARK; Bridgewater (2) |
| Viatris | 92556V106 | Bridgewater; Greenlight / DME (2) |
| Toast | 888787108 | ARK; Bridgewater (2) |
| Etsy | 29786A106 | Bridgewater; Soros Fund Management (2) |
| Oklo | 02156V109 | ARK; Bridgewater (2) |
NVIDIA is the only case where three managers in this fixed panel expanded the same security in both intervals. That is a broader disclosed pattern than a one-manager repeat, but it is not a consensus signal: 14 of the 17 reporting groups did not meet this exact three-quarter condition for NVIDIA. For context, our earlier AMD–NVIDIA Q2 comparison showed why share counts and reported values must be separated.
Method, controls and limitations
- Periods: quarter-end holdings reported for March 31, June 30 and September 30, 2026. All 17 fixed reporting groups had a verified Q3 filing path as of November 19, 2026.
- Comparable set: we start with 1,396 clean Q1–Q2 continuing manager–security observations from our prior bridge. Matching the same manager and CUSIP to an eligible long-share Q3 row leaves 1,179. We do not convert an unmatched row into a zero or a full exit.
- Security scope: the standing clean set excludes ETFs/ETPs, options, debt, warrants, preferred shares and other non-operating-company instruments. Q3 put/call and principal-amount rows remain outside the comparison.
- Account aggregation: account slices inside the same reporting group are summed by manager and CUSIP. Pershing Square’s management entity filed a Q3 13F-NT notice while its linked combination report supplied the holdings table; the two are treated as one reporting group.
- Corporate actions: Q1–Q2 share counts retain the previously verified split adjustments. A Q2–Q3 screen tested common split ratios against inverse implied-unit-value changes and produced no unresolved hits in the 1,179-row matched set. Identifier changes and unmatched securities remain outside the calculation.
- Interpretation: “expanded again” means a larger split-adjusted quarter-end share count in both intervals. It does not establish an actual purchase date, price, motive, conviction, total exposure, current position or return.
For the row-level Q1–Q2 bridge and exclusion logic, see our share-count-versus-valuation decomposition. The manager offset analysis explains why opposite directions inside one filing can disappear in a net-only result. The Guru research hub keeps the fixed panel and related reports together.
Sources and original documents
SEC, Frequently Asked Questions About Form 13F (updated March 6, 2026); SEC Form 13F and instructions; SEC Form 13F data sets. Direct Q1 and Q2 filing indexes for the fixed panel are enumerated in the sources section of the completed Q1–Q2 source bundle article.
Direct Q3 2026 filing indexes:
Altimeter; Appaloosa; ARK; Berkshire Hathaway; Bridgewater; Coatue; Dalal Street; Greenlight / DME; Fundsmith.
Icahn reporting group; Light Street; Pershing Square 13F-NT notice and combination 13F-HR; Situational Awareness; Soros Fund Management; Third Point; Tiger Global; Whale Rock.
AHNIM STUDIO calculation from the cited filings. This article is for information and education, not investment advice. Data and assumptions can be revised if an amendment or verified corporate action changes the normalized comparison.

Leave a Reply