RESEARCH

Thirteen Early 13F Filers Trimmed More Than They Added. Two Names Went the Other Way

November 2, 2026

Thirteen investment managers filed 13F reports for the quarter ended September 30, 2026, by October 30. Together they trimmed 506 positions and added to 271. Two names went the other way in every filing that held them: Linde and Amphenol. This article tests five hypotheses against those thirteen filings, and reports which ones survive.

Thirteen of the 147 managers on this site's tracked roster with a CIK filed a 13F report for the quarter ended September 30, 2026, by October 30. Their filings were submitted between October 13 and October 30, and together they report $132.24 billion in total value across 1,099 rows. The same thirteen reported $138.64 billion for the prior quarter, so the group's aggregate reported value fell 4.62 percent, and nine of the thirteen report a lower total than in the prior quarter. The thirteen are Matrix Asset Advisors, Sarasin & Partners, Montrusco Bolton Investments, Sustainable Growth Advisers, PineStone Asset Management, Egerton Capital, Mondrian Investment Partners, Greenhaven Associates, Flossbach von Storch, Investment Management Associates, Pzena Investment Management, Markel Group, and Torray Funds. The other 134 had not filed when this data was collected on November 1, and the filing deadline is November 16, so most of the roster is still to come. A 13F report shows positions at quarter-end only. It does not say why a manager bought or sold, and it does not show short positions or most derivatives, so every finding below is a description of reported positions, not an explanation of intent.

Hypothesis one: early filers were net trimmers. This holds on a share-count basis. Comparing each manager's Q3 share count with its Q2 share count, position by position and ticker by ticker, 506 positions were reduced and 271 were increased, with 80 new positions and 170 unchanged. The ratio is close to two trims for every add. The pattern survives removing the least active portfolio in the group: excluding Markel Group, which left 93 of its 142 mapped positions exactly unchanged, the count is 498 reductions against 243 increases. Two cautions apply. First, these are share counts, not dollars: a trimmed position can still be worth more if the price rose, and the aggregate value change of minus 4.62 percent mixes price moves with share changes, so neither figure alone proves net selling in dollar terms. Second, a trim is not an exit. Most of the 506 reductions are smaller positions in the same stocks, and the count of full exits cannot be read from this dataset alone, because a stock absent from a Q3 filing may never have been held. The defensible claim is narrow and it is the one this article makes: among early filers, more positions got smaller than got bigger.

Hypothesis two: some names attracted unanimous adding. This holds for exactly two names, and they are the strongest single-name signals in the early sample. Linde (LIN) is held by six of the thirteen, and all six increased their share counts: Egerton Capital by 33.6 percent, PineStone Asset Management by 5.3 percent, Flossbach von Storch by 2.5 percent, Markel Group by 2.3 percent, Montrusco Bolton Investments by 1.0 percent, and Sustainable Growth Advisers by 34 shares, which is essentially flat and is counted as an increase only in the strictest sense. The six positions total $1,279.5 million. Amphenol (APH) is also held by six, and none of the six trimmed: five increased and one opened a new position. Egerton Capital more than doubled its position, up 120.0 percent to 10,708,713 shares worth $902.7 million. Matrix Asset Advisors added 115.6 percent, Montrusco Bolton Investments 95.9 percent, Torray Funds 89.3 percent, and Flossbach von Storch 67.1 percent, while Sarasin & Partners opened a new position of 43,900 shares worth $3.7 million. The six Amphenol positions total $1,860.3 million. Breadth is not the same as conviction: Matrix's Amphenol position is worth $0.6 million against Egerton's $902.7 million, so the unanimity is in direction, not in size. Walt Disney (DIS) is a weaker version of the same pattern, with three increases and one new position across four holders and no trims, and it is noted here rather than promoted to a headline, because four holders is a thinner base than six.

Hypothesis three: early filers trimmed mega-cap technology. This holds in share-count breadth, with one important exception. Microsoft is held by eleven of the thirteen, and seven trimmed against two increases, one new position, and one unchanged. Alphabet's class A shares show one increase against seven trims across nine holders. Mastercard shows no increases at all against six trims across seven holders. Meta shows two increases against five trims across eight holders, Apple one increase against five trims across seven, Amazon three against five across eleven, Visa one against five across seven, and Taiwan Semiconductor three against five across eight. Nvidia is the exception that keeps this hypothesis honest: across ten holders, three increased, four trimmed, one opened a new position, and two were unchanged. There is no Nvidia consensus in either direction among early filers. The mega-cap pattern is therefore real but uneven, and it is a statement about share counts in thirteen early filings, not about the market: the 134 managers that had not filed could trim, add, or do nothing, and the picture can change before the deadline.

Hypothesis four: the quarter's most interesting "new position" is not a new position at all. This holds twice, and it is the article's main verification finding. Pzena Investment Management's filing appears, on a CUSIP-keyed comparison, to show both a full exit from Exxon Mobil (XOM) and a brand-new XOM position. Both readings are artifacts of the same event. In the prior quarter, Pzena held 589,449 XOM shares under CUSIP 30231G102, issued as Exxon Mobil Corp. In this quarter, Pzena holds 557,484 XOM shares under CUSIP 30233Q108, issued as ExxonMobil Holdings Corp. Netted at the ticker level, that is a 5.4 percent trim, not an exit and not a purchase, while the reported value rose from $80.6 million to $90.7 million because the price rose. The same pattern appears at Torray Funds: a prior-quarter ONEOK position of 238,831 shares under CUSIP 682680103 disappears, and a new Oneok Inc Com position of 239,263 shares under CUSIP 30609A109 appears, worth $20.40 million against $20.76 million before. The share counts match to within 0.2 percent, which is what a re-registration looks like, although the filings themselves do not state a reason and this article does not assert one. The lesson is mechanical and it matters for every 13F comparison: matching on CUSIP alone manufactures fake exits and fake entries whenever an issuer re-registers, and only ticker-level netting removes them. Two genuine Pzena moves survive that netting and are reported as such: Cisco Systems (CSCO) is a real exit, with 1,281,329 shares worth $150.5 million on a prior-quarter basis and no CSCO row in the Q3 filing, and GE HealthCare (GEHC) is a real new position of 1,255,691 shares worth $82.1 million.

Hypothesis five: early filers crowded into a single new name. This fails, and the failure is informative. Across the thirteen filings there are 80 new fund-ticker positions on a mapped-ticker basis, but no new name was added by more than two managers. Thermo Fisher Scientific (TMO), Adobe (ADBE), and Micron Technology (MU) were each added by two managers; every other new position was added by one. New buying in the early sample is diffuse, not crowded. Combined with hypothesis two, the early picture is therefore lopsided in a specific way: agreement shows up in adding to stocks already held, Linde and Amphenol, rather than in piling into the same new names.

The limitation that governs all five hypotheses is the sample. Thirteen of 147 is 8.8 percent of the roster, and early filers are self-selected: managers that file in October may differ systematically from managers that file in the final days before November 16, and the remaining 134 filings can confirm, dilute, or reverse every pattern above. Seventy-one rows in the collected data have no ticker mapping, and their values are included in fund totals but excluded from all ticker-level breadth counts in this article, so the breadth figures cover mapped tickers only. Positions are as of September 30, 2026, not as of today, and a manager's position may have changed since. Nothing here is a recommendation: 13F filings are backward-looking disclosures, and the unanimity on Linde and Amphenol is a fact about thirteen filings, not a judgment about either company.

Method notes: figures come from the SEC EDGAR 13F-HR information tables for the quarter ended September 30, 2026, collected November 1, 2026, with prior-quarter comparisons against each manager's 13F-HR for the quarter ended June 30, 2026. Share counts are compared at the fund-ticker level, with put and call rows merged, so a manager holding the same ticker on two rows is counted once. Funds that file under a shared series CIK were matched by accession, never by CIK prefix alone: Matrix and PineStone file under filer CIK 1172661, Greenhaven under 1213900, and Torray under 1193125. Values are in dollars as reported, not thousands of dollars. The full verification datasheet, with every figure in this article and its source, accompanies this draft.

StockHolderQ2 sharesQ3 sharesShare changeQ3 value ($M)
LINEgerton Capital300,854401,932+33.6%190.8
LINFlossbach von Storch845,040865,769+2.5%410.9
LINMarkel Group215,600220,600+2.3%104.7
LINMontrusco Bolton Investments242,597244,948+1.0%116.3
LINPineStone Asset Management769,783810,807+5.3%384.8
LINSustainable Growth Advisers151,670151,704+0.0%72.0
APHEgerton Capital4,866,68810,708,713+120.0%902.7
APHFlossbach von Storch5,656,9679,455,421+67.1%797.1
APHMatrix Asset Advisors3,0566,590+115.6%0.6
APHMontrusco Bolton Investments889,0161,741,326+95.9%146.8
APHSarasin & Partnersnew43,900new3.7
APHTorray Funds59,235112,104+89.3%9.5
StockHoldersIncreasedDecreasedNewUnchanged
MSFT112711
GOOGL91701
MA70601
META82510
AAPL71501
AMZN113503
NVDA103412
V71501
TSM83500
LIN66000
APH65010
DIS43010

Sources and data

SEC EDGAR 13F-HR filings for the quarter ended September 30, 2026, collected November 1, 2026, from the working collection (q3_holdings.csv, q3_filings.csv). Prior-quarter comparisons use each manager's 13F-HR for the quarter ended June 30, 2026. Pzena and Torray CUSIP details are from the managers' EDGAR information tables and the Q3 watch diff records for the same accessions. Filing-level accessions for the thirteen: Matrix 0001172661-26-004502, Sarasin 0001510481-26-000005, Montrusco 0001475940-26-000009, Sustainable Growth Advisers 0001290668-26-000005, PineStone 0001172661-26-004579, Egerton 0001581811-26-000009, Mondrian 0001317724-26-000013, Greenhaven 0001213900-26-112422, Flossbach 0001575677-26-000011, Investment Management Associates 0000052024-26-000005, Pzena 0001027796-26-000065, Markel 0001096343-26-000085, Torray 0001193125-26-439683. Every figure in this article is listed with its source in the verification datasheet (CSV).

Data: SEC EDGAR Form 13F filings, collected November 1, 2026. For education only — not investment advice.

MORE RESEARCH

What the Disclosure Rules Actually Say: Ten Reader Questions, Answered From the Statutes
Ten questions readers ask about Congress trades, insider filings, and 13Fs — answered from the statutes, rules, and forms themselves.
How to Read a 13F Filing: What Every Line Means — and What It Hides
A 13F lists what a manager held on the last day of the quarter — not what it holds now. Who files, the 45-day deadline, what each column means, and the five things a 13F never shows.