Whether to follow on and when are two different questions

Two outputs answering two different questions

The Scenario Engine returns a recommendation and a timing signal separately, and the separation is the point. Whether to put more money into a company and whether to do it in this round are different questions, and collapsing them into one answer loses the distinction a General Partner most needs.

A strong company in a round with unresolved terms is a yes on the first question and a wait on the second. A weak company in a well-priced round is the reverse. Reading the two together is what tells you whether you are deciding about the company or about the round.

The recommendation

Four values, covering the company and the fund's ability to act on it.

• Yes

  The evidence supports following on, and nothing in the fund's position stands in the way.

• Maybe

  The case is real but not clean. This is the output that most rewards reading the rationale rather than the headline.

• Wait

  The case may well be there, and something specific has not resolved yet. The timing signal names what.

• No

  Either the evidence does not support it or a constraint makes it unavailable. Again, the rationale distinguishes which.

The recommendation arrives with an allocation attached, expressed both as a dollar figure and as a share of the fund's pro-rata entitlement. Pro-rata is the natural unit because it is the amount that holds the fund's ownership steady, so a share below one is a decision to let the position dilute, and a share above one is a decision to build it. The upper end of the range reaches modestly above pro-rata, and it is reserved for the highest-conviction cases. The curve is calibrated separately by lifecycle stage, which is why a Conviction-stage firm and a Continuity-stage firm can see different allocations from the same evidence: concentration of conviction is structurally appropriate at one stage and less so at the other.

The timing signal

The timing signal reads three things: the stage and structure of the round being proposed, the size of the valuation step-up it implies, and how close the company sits to its next milestone. Four values come out.

• Now

  Nothing identifiable is due to change the picture, so waiting costs the position without buying information.

• Wait for a milestone

  A specific company milestone is close enough that its outcome would change the read.

• Wait for the next priced round

  The proposed terms leave the company's value genuinely unresolved, and a priced round would settle it.

• Reassess

  The timing picture is unclear or contested. This is an instruction to look again, not a soft no.

Waiting is not free, and the signal is not a recommendation to be cautious. A round that closes while a fund waits is a position the fund did not take. What the signal identifies is the narrower case where a specific, identifiable event would change the read, and where that event lands before the decision genuinely has to be made.

When the inputs are thin

Sometimes the module cannot see enough to answer. A company with little mark history, or a round whose structure has not been settled, leaves parts of the picture blank. In that case the module says so rather than producing a confident-looking output from thin evidence. Where the timing picture specifically is unavailable but the rest of the evidence is intact, the recommendation still stands and the gap in the timing read is surfaced alongside it.

This matters more than it sounds. The failure mode for any decision-support instrument is a plausible answer built on an incomplete input, because it is indistinguishable from a good answer at a glance.

What these outputs are not

Neither output is a prediction about the company. The recommendation does not say the company will succeed, and a no does not say it will fail. Both are readings of the evidence available at the moment the scenario runs, against a fund in a specific position, and the module holds no view about what happens after the round closes.