Reserves and follow-ons

Why funds reserve capital, how to size the reserve pool, how it splits across follow-on rounds, and what flooring to whole companies does to your deployment.

A reserve is capital a fund holds back from initial investments to follow on into existing portfolio companies later. Reserves convert a fund's deployment from "as many initial bets as possible" into "fewer initial bets plus the option to concentrate behind winners as they emerge."

It's a common point of confusion for first-time fund managers and LPs alike. The trade-off is real and the math is unforgiving.

The trade-off

Reserves reduce the number of initial investments you can make from the same fund size. If you set 30% of a stage's allocation aside for follow-ons, only 70% of that stage's dollars buy initial positions; you make ~30% fewer initial bets than a no-reserves strategy.

You give up shots-on-goal in exchange for follow-on rights to concentrate capital behind the companies that work. Whether that trade is worth it depends on:

Information. By the time you're following on, you should know more about the company than you did at entry. If your selection improves substantively, follow-ons let you act on that signal.
Pricing. Follow-on rounds are priced higher than entry. The same dollar buys less ownership. Your follow-on multiple is the exit multiple from that round forward, not from initial.
Access. Follow-on rights aren't always granted. Pro-rata is a contractual right, not a guarantee of allocation in hot rounds, and many seed-stage deals don't include it. If you can't get into the next round at all, the reserves were wasted.

Some funds run no reserves on purpose. The thesis: maximize shot count at entry, accept that your winners may dilute beyond your initial position. Other funds reserve heavily and run smaller initial check counts, betting the upside concentration outweighs the lost diversification.

Sizing the reserve pool

A common framing: reserves are sized as a percentage of a stage's total allocation, separate per stage.

stage_initial_dollars = stage_allocation × (1 − reserve_pct)  
stage_reserve_dollars = stage_allocation × reserve_pct

A 30% reserve on a $15M Seed allocation gives you $10.5M for initial checks and $4.5M for follow-ons. With $750k initial checks, that's 14 initial investments instead of 20.

Reserves can also be sized by anticipated number of follow-ons rather than by dollar pool. If you expect ~50% graduation from Seed → Series A and want to follow on into half of them at $400k each, you need (14 × 50% × $400k) = $2.8M in reserves. Both framings can be made consistent; pick whichever matches how you think.

A few rules of thumb:

Match your reserve to your access. Reserving more than you can deploy as pro-rata in real rounds wastes capital that could have funded another initial check.
Match your reserve to your conviction story. If your strategy is "concentrate behind winners," 30-50% reserves are common. If your strategy is "spray and follow when invited," 0-20% is more honest.
Reserves are not free optionality. Capital sitting in reserves over years of fund life dilutes the IRR even if it eventually deploys, because called capital is paid earlier than the follow-on date.

Splitting reserves across follow-on rounds

Reserves rarely go into one follow-on round. A Seed-stage reserve might span Series A pro-rata, Series B pro-rata, sometimes Series C super-pro-rata. Each round eats a different share of the reserve pool.

The math is straightforward. For each follow-on round:

follow_on_round_dollars = stage_reserve_dollars × pct_of_reserve  
follow_on_round_count = follow_on_round_dollars ÷ follow_on_check_size

The shares across follow-on rounds for a stage must sum to 100% of that stage's reserve pool. Concretely:

Follow-on 1 (Series A pro-rata): 60% of the reserve at $400k checks
Follow-on 2 (Series B pro-rata): 40% of the reserve at $750k checks

In a $25M fund with 60% Seed allocation, 30% reserve: $4.5M of reserves. Follow-on 1 gets $2.7M (≈ 6 companies at $400k). Follow-on 2 gets $1.8M (≈ 2 companies at $750k). Some companies receive both follow-ons, some receive one, some receive none. The aggregate counts are what matter for fund math.

Integer mode vs continuous

In practice, you make a whole number of investments. 6.75 companies isn't a thing.

If a model floors investment counts to whole companies (integer mode), each stage's count rounds down and the deployed dollars shrink to fit. The fund has unused capital, called "slack" or "deployment slack," because you can't break a check in half. A $750k initial check into a $10.5M pool deploys at most 14 × $750k = $10.5M, which fits cleanly. But 6.75 follow-ons at $400k = $2.7M is intended; floored to 6 follow-ons it's $2.4M, leaving $300k unused.

Slack tends to compound when you have multiple stages and multiple follow-on rounds. Called capital can come in materially less than committed. That's a feature, not a bug, and it's how venture funds actually work. Continuous (fractional-companies) mode is useful for quick sketches but overstates deployable capital because the math pretends you can write a 0.4-of-a-check check.

Common failures

Reserving more than you can deploy. A fund manager promises LPs 50% reserves but can only get into 30% of follow-on rounds. The unused reserves drag returns and the fund underperforms its deployment plan.
Reserving by feel rather than by math. "30% feels right" is fine for a first sketch. Before committing to LPs, check that the reserve dollars × graduation rate × follow-on check size produces a plausible number of follow-ons.
Treating reserves as free. Reserves are called capital. LPs paid for those dollars. The IRR clock is ticking on them whether they're deployed or not.
Forgetting downside. Reserves are deployed into companies that have raised more, so they're at higher valuations. Companies that fail after a follow-on writeoff more capital per dead company than companies that fail before.

In Hemrock's tools

The hosted Fund Economics Tool lays this out as the Investment Strategy table on the Portfolio Construction tab. Each entry stage is a row with editable name, % allocation, % reserve, # investments, and average check. When the reserve is > 0, the row expands into sub-rows for the initial check and each follow-on round, where you set the % of the reserve pool that goes into each round and the average check size for that round.

The expanded sub-row UI makes the trade-off visible: you can see how reserve pool dollars split into specific follow-on round counts as you tune assumptions. Integer mode is on by default; toggle Fractional companies in Settings for continuous math.

The spreadsheet Venture Capital Model handles the same logic with a different UX: a graduation chain that tracks the percentage of companies proceeding from each stage to the next, with explicit per-round dilution and post-money assumptions. The reserve-pool framing on the hosted version is faster for the most common case (set a reserve, decide how it splits); the graduation chain is richer when you need to model different strategies per round.