How this tool reaches its conclusions

Methodology.

How many companies should an early-stage / EIS portfolio hold? There is no single magic number. The honest answer depends on the shape of the return distribution, whether you are passively indexing the market or backing an active value-add manager, and the quality and breadth of your deal access.

Past performance is not a reliable indicator of future results. Capital is at risk; you may get back less than you invest. This is not investment advice.

The honest headline

Across every credible source, the evidence converges on the same practical range and the same framing. Adding positions reliably raises your typical (median) outcome and shrinks your dispersion and loss probability [A1]·[A2]·[A7]. A higher expected multiple is only implied when the underlying return power-law exponent α is below 2 [A7] — a condition that is contested[A2]·[C4] and not asserted by this tool.

Conventional EIS funds hold 8–10 companies — the zone where you are most likely to miss the winners and where one failure can sink the portfolio. The evidence supports a floor of around 30 and a target closer to 50: at that level you materially reduce the probability of loss and raise your typical outcome. The benefit is lower risk and a higher typical result — not a guaranteed higher multiple, and not a substitute for the quality of the underlying deals.

Why the curve exists (the power-law spine)

Power-law returns behave in three regimes [A7]·[B3], and the regime dictates what adding positions does:

α > 3 (finite mean and variance): random selection — portfolio size affects neither expected mean nor median.
2 < α ≤ 3 (finite mean, infinite variance): more positions raises the expected median, not the mean.
α ≤ 2 (both unbounded): more positions raises both median and mean.

AngelList estimates winning seed returns sit at α < 2[A7]; Correlation's data implies α > 2[A2]. The safe, regime-agnostic claim — true in both the 2<α≤3 and α<2 worlds — is the one this tool uses: adding positions raises your typical (median) outcome and shrinks your dispersion / loss probability. A higher expected multiple requires α<2 and is not safe to assert [C1]·[C4].

The size curve

The methodology synthesises the evidence into five bands. Your audit result places your portfolio in one of them.

0–9 holdings

High concentration risk

With fewer than 10 holdings, a single failure can determine the outcome of the whole portfolio, and you are statistically more likely to miss the small number of companies that drive returns. Increasing the number of holdings is the most direct way to reduce this risk.

10–29 holdings

Below the evidence-based floor

Single-name risk still dominates at this level. The evidence supports building toward a floor of around 30 holdings before diversification meaningfully reduces your probability of loss.

30–49 holdings

Responsible floor met

You have reached the responsible floor. At this level you are likely to hold some of the small number of companies that produce the majority of returns. Moving toward ~50 holdings continues to reduce risk.

50–79 holdings

Risk-optimised range

This is the evidence-based range where diversification delivers the most benefit relative to effort: a materially lower probability of loss and a higher typical (median) outcome.

80+ holdings

Highly diversified

Diversification benefits continue at this level but with diminishing marginal returns. Beyond this point, additional holdings add little, and for an actively managed portfolio can dilute the manager's contribution.

Supporting waypoints

At 8 holdings, the probability of a fund-level loss is approximately 28%; at 100 holdings it falls to approximately 6% [A2].
Above 50 investments, the median IRR (p.a.) was approximately 11.9%, versus approximately 2.9% at or below 50 (10,665 real LP portfolios) [A1].
In UK angel data, approximately 9% of exits produced approximately 80% of all positive cash flows; the remaining 56% of exits returned less than capital [A5].
Around 90 holdings, approximately 90% of investors were "in the money"; the median converges to the mean near 100 holdings [A1]·[A2].
UK VC dispersion remains the highest of any private asset class — upper-quartile TVPI of 2.15, with only 8% of UK funds reaching TVPI ≥3 [A4].

Deal-source quality — the second dimension

Holdings count is necessary but not sufficient. The indexing result holds only if the investor can sample the true market distribution; diversifying within a weak or adversely-selected funnel will not replicate market returns [C5]. For that reason the audit scores deal-source quality as an independent dimension. A strong size score paired with a weak deal-source score does not produce an unqualified "good" verdict.

What this tool does NOT claim

It does not promise or imply a specific forward return. The benefit surfaced by the tool is lower risk and a higher typical (median) outcome — not a guaranteed higher multiple [C1]·[C4].
It is not a substitute for the quality of the underlying deals. A larger number of holdings drawn from a narrow funnel will not behave like the same number drawn from broad, top-tier access [C5].
Over-diversification can erode returns for an active value-add manager — finite partner attention dilutes the value-add that drives returns [C2]·[C3].
It is not investment advice or a personal recommendation. It is an evidence-based diagnostic.

How this tool handles the SyndicateRoom 3.7× / 4.7× figures

The 3.7× / 4.7× multiples are genuine SyndicateRoom-published research findings. SWG attributes them to SyndicateRoom only — never to EISA — and presents them as illustrative corroboration rather than as a headline or a forward return claim. Wherever they appear they carry three caveats from SyndicateRoom's own text (seven-year holding period; the rate of return slows once the portfolio reaches around 30 investments; part of the higher figure is attributed to larger / later-stage rounds) and the past-performance disclaimer above. Full attribution rules are documented on the Sources page [SR].

See the full citation pack at /sources (Lists A, B and C). SWG operates exclusively under the FPO 2005 exemption reliance and is not FCA-authorised to provide investment advice. This is not investment advice or a personal recommendation.