The Superinvestor Handbook.

A 13F is a quarterly disclosure every US institutional investment manager with more than $100 million in qualifying long-only equity holdings must file with the SEC. Warren Buffett's Berkshire Hathaway files one. So do Bill Ackman's Pershing Square, Michael Burry's Scion Asset Management, Seth Klarman's Baupost, Stanley Druckenmiller's Duquesne, and about four thousand other funds.

The filing lists every US-listed stock, ETF, ADR, convertible, and certain options positions the fund owned on the last day of the quarter. Share count, market value, and the fund's split between the securities — but not cost basis, not entry dates, not shorts, not cash, not non-US positions, and not commodities or bonds.

That last sentence is the whole trick to reading 13Fs correctly. A 13F is a snapshot of the US-listed long book, not a full portfolio. The information gap is deliberate — the SEC built the form to prevent gaming of positions without burdening managers with daily disclosures — and it is the reason most retail traders misread the data.

Every quarter, HoldLens pulls the 13Fs of the 30 tier-1 managers we track, de-duplicates their moves, and normalises them to a single signed −100..+100 conviction scale. You can see the current signals at /best-now → or drill into a specific ticker with /signal/AAPL → — replace AAPL with any of the ~100 tickers we cover.

A 13F is not real-time. The SEC gives managers 45 days after quarter-end to file. Q4 2025 ends December 31, 2025 — the filings trickle in through mid-February 2026. Q1 2026 files by May 15, Q2 by August 14, Q3 by November 14.

That means when you are reading a 13F in late May, the positions it describes are up to 105 days old: 45 days from quarter close to filing deadline, plus up to 60 days from the earliest intra-quarter trade that appears in the snapshot. A manager who opened a position on April 1 and the filing lands August 14 — that is four-and-a-half months of price action you have missed. The position might already be closed.

This is the main reason mechanical copy-trading from 13Fs underperforms the underlying portfolio. A retail investor buying what a tier-1 manager bought, on the day of the filing, is buying at a dramatically different price — and often into a position the manager is already unwinding.

The right way to use 13F data is not to time trades. It is to identify high-conviction ideas — names where multiple skilled managers have placed large, recent, concentrated bets — and then do your own fundamental research on the ticker. HoldLens surfaces the best conviction signals at /conviction-leaders → and/best-now →.

There are roughly four thousand 13F filers in the US. Most of them are noise.Index funds, pension funds, and insurance general accounts all file 13Fs and they all hold hundreds or thousands of names sized to benchmark weights. Their "positions" carry no conviction — they are simply reflecting a mandate.

The filers worth tracking are the ones whose portfolios reflect actual bets. HoldLens tracks thirty: the canonical value names (Buffett, Munger-era Berkshire, Klarman, Pabrai, Greenblatt, Akre, Nygren, Rochon), the concentrated activists (Ackman, Loeb, Icahn, Ubben), the quality-growth names (Smith, Rolfe, Polen, Kantesaria, Li Lu, Hohn), the macro and multi-strat names (Druckenmiller, Tepper, Marks, Watsa), the tiger cubs (Coleman, Mandel, Ainslie, Halvorsen, Armitage, von Mueffling, Slater), plus the high-signal outliers (Burry, Greenberg, Einhorn).

Each manager has a manager-quality score based on long-run outperformance, portfolio concentration, and the quality of their public write-ups. HoldLens weighs each move by this score when computing conviction — a 5% position in Berkshire counts for more than a 5% position in a fund you have never heard of.

You can see the full ranking at /manager-rankings →or browse by philosophy at /by-philosophy →. The leaderboard is at /leaderboard →.

The single most important concept in smart-money tracking is conviction. A manager with a thousand holdings has no conviction in any of them. A manager with twenty holdings where the top five positions each represent 10%+ of the book has placed actual bets — and those are the names worth paying attention to.

Conviction shows up in three ways on a 13F:

• Position size as a percent of the book — a 10% position screams louder than a 0.5% position, even if the dollar value is the same for a large fund.
• Action direction — a NEW position (opened this quarter) or a full EXIT (closed this quarter) is a louder signal than an add or a trim. Openings and closings mean the manager is changing their mind.
• Recency — a move made in the most recent quarter beats a stale position held for ten quarters. Markets change.

HoldLens combines all three into a single signed ConvictionScore from −100 to +100. Read the methodology at how conviction is computed →or dig into the biggest bets across the tracked managers at /big-bets →.

A raw 13F has a small number of fields per position. Here is what each means, what to pay attention to, and what to ignore.

• CUSIP and Name of Issuer — the security identifier. Usually the stock ticker. For convertibles and options, the CUSIP tells you the underlying, but you will need a lookup tool to translate. HoldLens translates CUSIPs to tickers automatically.
• Value (x1000) — market value of the position at quarter-end, in thousands of dollars. Divide Value by the quarter-end stock price to get approximate share count.
• Shares or Principal Amount — the literal share count the fund holds. Compare this quarter-over-quarter to see if they added or trimmed.
• Investment Discretion — "SOLE" is what you want. "DFND" or "OTR" means the fund is reporting holdings it does not itself choose, usually sub-advisory. Ignore sub-advisory lines.
• Voting Authority — who gets to vote proxies. Less useful for copy-trading but matters for activist situations — Ackman, Loeb, Icahn, Ubben.

The field you cannot read from a raw 13F is cost basis. You do not know when the position was opened or at what price. Estimating cost requires combining consecutive 13Fs, stitching together the share-count trajectory, and mapping it against quarterly price ranges. HoldLens does this for every tracked manager and surfaces the longest compounding positions at /trend-streak →.

The most common retail mistake with 13F data is to read it as a "what to buy" list. Ackman just bought X, I'll buy X. This loses money in three specific ways.

One. The 45-day lag means your entry price is almost never the manager's entry price. If Ackman bought a stock at $100 on February 1 and the filing lands May 15 at $130, you just paid 30% more than the person whose trade you are supposedly copying. Their IRR and yours will diverge by that 30% on day zero.

Two. 13Fs show positions, not trades. A manager may have been selling aggressively in the final month of the quarter — the position you see on the filing date might already be half gone. You are buying on the way out.

Three. Concentrated managers run all-weather portfolios. A 5% position in Berkshire is a position Buffett can hold through a 40% drawdown because his cost basis is some fraction of the market price and his opportunity cost is low. The same position at your cost basis is a position you may be forced to sell at the worst possible time.

The honest way to use 13F data is as an idea funnel — a list of names that passed the filter of tier-1 manager conviction, which you then investigate on their own merits. Read more at the copy-trading myth →.

Once you have idea-funneled down from "every 13F" to "names held by multiple tier-1 managers with high conviction," the next question is: do I want consensus bets or contrarian bets?

Consensus bets — names owned by five or more tier-1 managers, each with significant book weight, all adding or holding. These are the safest reads on smart money because independent skilled people converged on the same conclusion. The tradeoff is that returns may already be priced in. Browse them at/consensus →.

Crowded trades — the dark side of consensus. When thirty out of thirty managers own the same name and it has tripled, the next leg is likely to be a painful unwind as someone blinks first. The line between "high conviction consensus" and "crowded trade" is whether active conviction is still growing or plateauing. See/crowded-trades →.

Contrarian bets — names where tier-1 managers split. Two strong managers are buying, two are selling. This is the rarest pattern and often the most interesting — it means thoughtful people looking at the same data reached opposite conclusions. Dig into these at /contrarian-bets →. The setups where you can see why one side is wrong and the other is right are where outsized returns live.

Fresh conviction — names where only one manager has a position, and opened it in the latest quarter. Either early alpha or a solo mistake. The payoff structure is asymmetric either way. See /fresh-conviction →and /hidden-gems →.

The single most underrated signal on a 13F is how concentrated the overall portfolio is. A manager whose top five positions represent 60% of the book is makingdecisions. A manager whose top five positions represent 12% of the book is shadowing an index and calling it a strategy.

Concentration is the proxy for real active management. It is strongly correlated with long-run outperformance across public and private research on manager skill. The managers who have compounded money the longest — Berkshire, Pabrai Funds, Nygren's Oakmark, Baupost, Polen — tend to be concentrated.

HoldLens ranks managers by concentration at /concentration →. When you pair a concentrated manager with a high-conviction recent move, the signal-to-noise ratio of a 13F entry can exceed 10x the average line on any filing.

The flip side: managers whose portfolios have drifted less concentrated over time are often transitioning to institutional asset-gathering mode and their signals are decaying. Concentration trend is itself a signal — see the quarter-over-quarter trajectory of any manager at theirinvestor page → under "per-quarter digests".

A 13F is a snapshot. Two consecutive 13Fs are a diff — and the diff is where most of the actionable information lives. The thing to read is not "what does Ackman own?" but "what did Ackman change?"

Every move falls into one of four categories:

• NEW — a position that did not exist last quarter. Opening a new line in a concentrated portfolio is the loudest possible signal of fresh conviction. See/new-positions →.
• ADD — an existing position increased by more than a noise threshold. ADDs below 5% are often tax-lot rebalancing. ADDs above 20% are meaningful.
• TRIM — an existing position decreased by more than 15%. Below 15% is usually rebalancing. Above 40% is a conviction collapse worth reading./biggest-sells → ranks the loudest cuts.
• EXIT — full close-out. The loudest possible sell signal. HoldLens tracks all full exits at /exits →.

Flow reading gets interesting when you stitch it across time. A position that was opened three quarters ago, added to in each of the next two quarters, and is now at 8% of book is a compounding conviction — the manager is adding on the way up because they believe the thesis is maturing. See the multi-quarter streaks at/trend-streak → and the rarer "comeback" pattern — a manager re-entering a name they previously exited — at/reversals →.

13F data is not a crystal ball. Before using any of this, know the gaps.

• No shorts. A fund can be net long on 13F while being net short overall. Burry's famously bearish positioning in 2023-24 was almost invisible on his 13F because the shorts are in options and short positions that do not file.
• No cash. You do not know if a manager is 95% invested or 40% cash. A manager sitting on cash is making a macro call you cannot see.
• No non-US holdings. European, emerging-market, and Japan holdings are invisible unless the issuer has a US ADR listing.
• No commodities or private assets. Gold, real estate, private equity, direct real estate — all invisible.
• Up to 105 days stale. By the time you read a 13F fully, the position is old news. Use the data for conviction detection, not trade timing.
• Not advice. This is public history, not a forecast. Past outperformance does not guarantee future returns. Every stock is a loss of capital risk.

That said — 13F data is still the cleanest window you have into what the world's best investors believe right now, and it is free. The honest way to use it is as a high-signal idea funnel feeding your own research, not as a copy-trade pipeline.