What Is Look-Through Exposure, and How Do You Calculate It?

Look-through exposure is your share of the companies, sectors, countries and currencies held inside every fund, co-investment and vehicle you own, added up as if you held them directly. Kepler, which we build, calculates it by multiplying each vehicle’s holding by your share of that vehicle through every layer and summing by company, and returns one number broken down by vehicle, with every input cited to its statement, because your firm’s ontology resolves each company across every manager’s reports.

The idea is simple and the work isn’t. An allocator might hold the same company through two buyout funds, a growth fund, a fund-of-funds and a co-investment, each reporting under a different name, on a different date and in a different currency. Each fund looks diversified on its own. Added up, one name can break a concentration limit. This guide covers why look-through matters, the formula, a worked example with made-up numbers, how to handle stale data, and the problems that make the calculation hard. For the documents it comes from, see how LPs use AI on fund documents.

Why look-through exposure matters

• Concentration. Single-name, sector and country limits only work if you can see through the funds. A program can be over its limit on one company without any one fund holding much of it.
• Overlap across managers. Two managers in the same strategy often own the same companies, or buy them from each other. Look-through shows how much of your manager diversification is real.
• Risk. Exposure to a sector downturn, a country, rates or a currency sits in the underlying companies. A fund-level view hides it.
• Regulation. Some investors must look through by rule. Under Solvency II, insurers calculate their capital requirement on the underlying assets of funds, as Article 84 of the Delegated Regulation requires.
• Reporting. Boards, investment committees and LPs of a fund-of-funds ask where the money actually is.
• Decisions. Re-ups, secondary sales and co-investment approvals all start with “how much do we already own?”

The formula

For one company, your look-through exposure is the sum across every vehicle you hold:

Exposure = Σ (your share of the vehicle × the vehicle’s holding in the company)

Each term needs care:

• Your share of a fund is usually your capital account divided by the fund’s total partners’ capital, both as of the same date. Commitment share is a proxy, and it drifts with late closings, defaults, excused investments and carried interest allocations. If your side letter excused you from an investment, your share of that one holding is zero.
• The vehicle’s holding is its reported fair value of the company, from the schedule of investments.
• Fund-of-funds multiply through: your share of the fund-of-funds times its share of the underlying fund times that fund’s holding.
• Co-investments and SPVs use your ownership of the vehicle directly.
• Direct holdings count at 100%.

Gross vs net. If a fund borrows at the fund level, through a subscription line or a NAV facility, its investments are larger than its partners’ capital. A fund with $950 million of investments and $800 million of partners’ capital gives a 5% LP $47.5 million of gross look-through holdings against a $40 million capital account. Use gross holdings for concentration limits and net figures when reconciling to NAV, and state which one a report shows. Kepler computes both and labels each.

A worked example

A made-up program with $410 million of NAV wants its exposure to one private company, Acme, against a 3% single-name limit.

Vehicle	Your share	How Acme appears	Holding at fair value	Your exposure	As of
Fund A (buyout)	5.0% ($40.0M of $800.0M)	Acme Holdings LLC	$120.0M	$6.0M	June 30, 2026
Fund B (growth)	4.0% ($18.0M of $450.0M)	ACME Inc.	$50.0M	$2.0M	June 30, 2026
Fund-of-funds C, via Fund D	10.0% × 8.0% = 0.8%	Project Anvil	$75.0M	$0.6M	March 31, 2026
Co-investment SPV	25.0%	Acme Topco Ltd	$20.0M	$5.0M	September 30, 2026
Total				$13.6M

$13.6 million is 3.3% of the program, over the 3% limit, while the largest single position, Fund A’s, is 1.5%. Three things made this hard to see: four different names (including a project code name in the fund-of-funds report), four vehicles with four reporting formats, and three as-of dates spanning six months. On Kepler, this table is the answer to one question, and each row opens to the schedule of investments and capital account statement behind it.

Adjusting for what’s known since. Fund A’s Q2 letter, in a subsequent-events note, says it sold 30% of its Acme stake in August. On an adjusted basis, Fund A’s line becomes $6.0 million × 70% = $4.2 million, the total falls to $11.8 million, and exposure is 2.9%, just under the limit. Report both figures, the reported basis and the adjusted basis, each with its source. Kepler keeps the August sale as a cited adjustment, linked to the letter’s note, until Fund A’s next report replaces it.

Rolling stale data forward

Fund reports lag by design. ILPA’s suggested timeline is 60 days after quarter-end for a direct fund, 120 days for a fund-of-funds and 180 days for a fund-of-funds-of-funds, and many reports arrive later. A program’s look-through is always a blend of dates. Three habits keep it honest:

1. Show the as-of date on every row. A reader should see at a glance which inputs are six months old.
2. Roll forward with known cash flows. Adjust each fund’s NAV for contributions and distributions since its report date, and for realizations announced in letters or press releases.
3. Keep adjustments separate. Label adjusted figures and keep the reported figures beside them, so the adjustment can be checked and reversed when the next report arrives.

The data problems

Names don’t match. The same company appears under its legal name, its brand, a holding company, a project code name or an abbreviation. A legal entity identifier (LEI) helps where one exists. Most private companies need entity resolution: matching every alias to one record and keeping the evidence for each match. Without it, exposure is undercounted. This is the core job of an ontology, and the reason Kepler starts every look-through question from one record per company.

Partial disclosure. Some funds report only top holdings, or only sector splits. Look through what’s disclosed, report the remainder as unallocated, and show coverage: “72% of NAV looked through.” Kepler reports coverage beside every look-through total.

Debt and equity in one company. A credit fund may hold Acme’s term loan while a buyout fund holds its equity. Report exposure by instrument as well as in total, since the risk differs.

Currency. A euro fund reports in euros. Convert at the as-of date’s rate for the holding and show currency exposure as its own view, since a dollar investor holds the currency risk as well.

Fund-level borrowing. Subscription lines and NAV facilities make gross holdings larger than NAV, as above.

Listings and remarks. A company that went public after the report date has a market price today. A company the GP has since written down has a new mark in the next report.

Counting the same interest twice. If your records hold a fund both directly and inside a separate account, make sure each fund interest is counted once, while exposure through a fund-of-funds that holds the same fund is real and should be added.

Sector, country and currency look-through

The same arithmetic applies with a different grouping key: sum exposure by sector, country or currency instead of by company. The extra step is classification. GPs classify private companies differently, so map every company to one taxonomy, your firm’s or a standard such as GICS, and apply it across every manager.

How to build a look-through process

1. Collect the latest schedule of investments from every fund, co-investment and SPV.
2. Compute your share of each vehicle from capital accounts, with excused investments noted.
3. Resolve every company name to one entity, keeping the evidence for each match.
4. Multiply through each layer and sum by company, instrument, sector, country and currency.
5. Record the as-of date and source for every input.
6. Roll forward with known cash flows and realizations, labeled as adjustments.
7. Show coverage where disclosure is partial.
8. Compare against limits and flag breaches.
9. Refresh as each new report arrives, and keep each quarter’s result for comparison.

For allocator tools that support parts of this, see the best AI tools for LP and fund-of-funds due diligence and the best AI tools for family offices.

How Kepler does this

Ask Kepler for your exposure to Acme and the worked example above comes back as one table, built from documents it has already read. Here is what sits behind each part of it.

• The four names. Our engineers build your firm’s ontology from your GP portals, quarterly reports, capital account statements, LPAs, side letters and commitment tracker. Acme Holdings LLC, ACME Inc., Project Anvil and Acme Topco Ltd resolve to one company, and each match keeps the passage that supports it. Funds, parallel vehicles, feeders, fund-of-funds layers and co-investments are records of their own, linked to the holdings beneath them.
• The inputs. Extraction models we train read every schedule of investments, capital account statement and letter in full when it arrives, scanned or native, and keep each holding with its value, currency, as-of date and page.
• The arithmetic. Kepler computes your share of each vehicle from capital accounts, multiplies through every layer, zeroes out investments your side letters excused, converts currency at the as-of rate, and sums by company, instrument, sector, country and currency in code. Each row opens to the statement line behind it, and coverage shows beside every total.
• The August sale. Because Kepler read the subsequent-events note in Fund A’s letter too, the adjusted 2.9% sits beside the reported 3.3%, cited to the note, until Fund A’s next report replaces it. When two reports disagree about a holding, both appear with the conflict flagged.

Kepler’s agents keep the answer current. They refresh exposure as each report lands and alert you, with the source, when a name crosses a limit. Each refresh runs with its owner’s permissions, down to the fund and team, and content inside a GP report is never taken as an instruction. Your board pack and exposure report are saved as playbooks and fill in Excel, Word and PowerPoint, the same figures reach Claude, ChatGPT and Slack, and every query is logged (security). Our engineers built this layer before at Palantir; for an allocator, their first job is connecting the GP portals and resolving companies across every manager.

Two more questions worth asking:

• “Which companies are above 2% of program NAV on a look-through basis?”
• “Show our sector exposure by manager, and how much of each fund we could look through.”

Ask Kepler the Acme question about your own program in a demo, or start at app.kepler.ai.

Frequently asked questions

What is look-through exposure?

Look-through exposure is an investor's indirect share of the companies, sectors, countries and currencies held inside its funds, co-investments and other vehicles, added up as if it held them directly. Kepler, which we build, returns it as one number per company broken down by vehicle, with every input cited to the statement it came from and the arithmetic done in code. It shows concentrations no single fund report reveals.

How do you calculate look-through exposure?

For each vehicle, multiply the vehicle's holding in a company by your ownership share of that vehicle. For a fund-of-funds, multiply the shares through each layer. Add the results across every fund, co-investment, SPV and direct holding, after matching each company's different names to one entity. Record each input's as-of date, since funds report on different lags.

Should look-through use NAV or commitments?

Use your share of each fund's partners' capital (NAV) for current exposure, since that reflects what you own today after contributions, distributions and any excused investments. Commitments are better for planning future exposure, and unfunded commitment shows what could still be called. Say which basis a report uses, because the numbers can differ widely in a young or maturing fund.

How do you handle a fund that only discloses its top holdings?

Apply your share to the holdings disclosed, and report the rest as unallocated, with the coverage ratio shown, such as 72% of the fund's NAV looked through. Where the GP provides sector or country splits for the remainder, use them for those views. If your exposure limits depend on it, ask for full holdings under your side letter.

Why is look-through exposure always out of date?

Funds report holdings with a lag, commonly about 60 days after quarter-end for a direct fund and longer for each extra layer of funds. By the time the last report arrives, holdings may have been sold or remarked. Show each vehicle's as-of date, roll forward with known cash flows and announced realizations, and label adjusted figures separately from reported ones.