Should you pick your reserve percentage or derive it

The usual ordering

Most fund models begin with a reserve percentage. Someone picks a round number, often 40 or 50 percent because that is what the last fund did or what the deck template suggested, and the rest of the construction is fitted around it. Portfolio count gets adjusted until the cheque size looks sensible. Ownership targets get adjusted until the returns model works.

The reserve, in other words, is treated as a preference. It is actually a residual, and the difference shows up everywhere downstream.

A chosen number is a modelling artefact

Anubhav Srivastava, who built Tactyc after watching managers construct portfolios in spreadsheets, describes the common failure as over-reserving upfront and producing a model that is conservative in a way nobody chose. The reserve gets set generously because generosity feels prudent, and then it silently constrains the initial cheque for the next decade.

His alternative is to compute rather than assume: work out what pro-rata in the next round would actually cost, and weight it by the rate at which portfolio companies reach that round. A fund whose companies graduate at one rate needs a different reserve from a fund whose companies graduate at another, and neither number is knowable by picking a percentage first. The reserve becomes an output of the construction rather than a parameter of it.

A named pool creates pressure to spend it

The second argument is about behaviour rather than modelling. Charles Hudson, whose firm reserved about a quarter of its fund for follow-on, has come to think that having dollars earmarked for a purpose encourages deploying them for that purpose. A reserve labelled follow-on capital does not sit neutrally; it creates a mild standing expectation that follow-ons will happen.

His preference is to keep dollars in competition with each other, so that a follow-on has to win against the alternative use of the same money, which is usually a new position in a company the fund does not own yet. That is a genuinely different question from whether the existing company is doing well, and a fixed reserve percentage prevents it from ever being asked.

This is the strongest practical case for treating the reserve as an output. A residual is available for whatever use is best. A budget tends to get spent.

The assumption underneath is usually wrong

There is a third reason, and it is the quietest. The Blue Future Partners survey found managers expecting ownership at exit to sit close to ownership at entry, assuming an average of 18.5 percent dilution across the whole life of an investment. The author is direct that this is not realistic, and that stakes get diluted substantially regardless of reserves held to defend them.

A reserve set to hold ownership flat is therefore sized against an outcome that mostly does not happen. That does not make reserves pointless, but it does mean the number was chosen to buy something other than what it actually buys, which is a slower rate of dilution in a subset of companies rather than a stable position across the portfolio.

What follows from treating it as an output

The mechanics are unremarkable. Decide the stage the fund enters at, which sets what a meaningful position costs. Decide the ownership target, which sets the cheque. Decide the portfolio count. Those three determine the capital committed to initial positions, and the reserve is what is left.

What changes is not the arithmetic but what the number now means. A reserve derived this way can be explained: it is this size because the fund writes cheques of that size into that many companies at that stage. If an LP pushes on it, the conversation moves to the construction decisions underneath, which are the ones worth defending. A reserve chosen as a round number has nothing underneath it to move to.

It also makes the trade-off visible in the right direction. Raising the reserve is no longer a prudent-sounding adjustment; it is a decision to write smaller cheques, back fewer companies, or take less ownership. Those consequences were always there. Choosing the reserve first is what hides them.

Why the size was never the deciding factor anyway

Laura Thompson's scenario work at Sapphire holds the reserve percentage constant and varies only where the capital lands, producing net outcomes from 2.5x to 5.0x against a 4.0x base case. The size of the reserve did not move across those scenarios. Only the selection did.

Which is the case for caring less about the percentage than the industry does. It is a term in an equation, and treating it as the headline decision gives it a prominence the evidence does not support.

How the model reads it

The Colibrí Architecture model already treats the reserve as a derived term rather than a standalone choice. The reserve share determines the capital available for initial positions; that figure divided by the target portfolio count gives the implied initial cheque; and whether that cheque is coherent with the fund's declared stage is one of the dimensions Portfolio Efficiency reads.

A General Partner who raises the reserve percentage and changes nothing else will see the reading move, not because the model disapproves of reserves, but because the fund now intends to write smaller cheques than its stated stage supports. That is the consequence surfacing where it belongs, at the point the decision is made rather than in year four.

Sources

• Portfolio Construction for VCs, on the EUVC podcastAnubhav Srivastava, CEO of TactycOn managers over-reserving upfront, and risk-weighting the next round's pro-rata by the graduation rate rather than assuming the full amount.
• Lessons from 600+ Investments, Founder Profiles that Win, Reserve Strategies that Drive Returns, and the Hidden Potential in Consumer StandoutsCharles Hudson on The Full Ratchet, episode 488On earmarked reserve dollars encouraging their own deployment, and keeping dollars in competition with each other instead.
• Dirty Secret: Venture Reserves are Not Always a Good ThingLaura Thompson, Sapphire, May 2022The scenario work separating the size of a reserve from its effect on net returns.
• An LP take on VC portfolio constructionRodrigo Ferreira, Blue Future Partners, published on OpenVCThe dilution managers assume when they set a reserve, and why the assumption tends not to survive.