SEIS Fund Performance: Reading the Numbers Honestly
Reliefs flatter, fees drag, survivorship hides. How to read any performance table honestly.
In this article
Every SEIS fund brochure contains a performance number, and almost none of them can be compared with any other. Reliefs flatter returns, fees drag them, survivorship hides the failures, and immature portfolios are marked at hopeful valuations rather than exits. None of this requires bad faith; it is simply what early-stage performance reporting looks like before you interrogate it. This analysis sets out how SEIS fund performance is actually constructed, and the questions that make any table honest.
Why SEIS fund performance is hard to measure
Three structural problems, before anyone chooses a flattering font. First, immaturity: SEIS portfolios take five to ten years to resolve, so most published numbers describe unrealised valuations, not cash returned. Second, small samples: a typical fund cohort holds ten to thirty companies, so one outlier drags any average anywhere. Third, the relief question: returns can be quoted before reliefs, after initial relief, or after initial plus loss relief at an assumed tax rate, and the same portfolio can look like 0.9x, 1.4x or 1.8x depending on which convention the marketing chose. A table without its convention stated is not information yet.
The four adjustments between headline and reality
Gross to net of fees. SEIS funds commonly charge initial fees, annual management fees (often front-loaded or capped for relief efficiency), and sometimes performance fees on exits. Across a five-to-eight-year life these can consume a meaningful slice of the subscription. Ask for the net-of-all-fees figure per pound subscribed; if the answer is a gross figure “because fees vary”, assume the variance is not in your favour.
Valuation to cash. Unrealised holdings are marked, typically at the last funding round price. A portfolio marked at 1.6x with no exits has returned nothing yet, and down rounds revise history without notice. The clean metrics: DPI (cash actually distributed per pound in) alongside TVPI (cash plus marks), and the share of the portfolio value that sits in realised versus paper.
Cohort to universe. Funds report the vintages that worked; funds that failed stop reporting; new funds inherit a track record from a “team” rather than a vehicle. Ask for every vintage the manager has raised, including the ones not in the deck, and treat a two-vintage manager’s “since inception IRR” as weather, not climate.
Pre-relief to post-relief, separated. Reliefs are real money and belong in an investor’s own arithmetic, at the investor’s own rates and capacity, per the calculator. What they do not belong in is the fund’s skill claim: 50% initial relief arrives whether the manager picks well or badly. A defensible presentation shows investment performance pre-relief, then the tax layer separately.
The base rates any claim must beat
The context for any SEIS performance figure is the underlying asset class: most seed companies fail or return roughly their capital, a minority exit modestly, and a small number produce nearly all the returns, the power-law arithmetic in the success rates analysis. A manager claiming smooth positive marks across a young portfolio is describing valuation policy, not a repeal of the distribution. Equally, headline failure counts mislead in the other direction: a fund with 60% write-offs can still perform well if one holding returns twenty times. Judge the shape, not the casualty list.
Ten questions that sort funds quickly
What is DPI, per vintage? What convention does every quoted figure use (gross or net, pre or post relief, realised or marked)? How are unrealised holdings valued, and when were marks last revised downward? What are all fees across the fund life, expressed per pound subscribed? How fast is capital deployed, since relief timing follows deployment into each company, not subscription? How many vintages has this team raised, and where are the missing ones?
What is the concentration of value (top three holdings as a share of portfolio value)? How does the fund handle failed companies’ loss relief paperwork? What happened to the vintage raised in the manager’s worst year? And who audits or reviews the valuations? A manager comfortable with all ten is a different proposition from one comfortable with the brochure. The structural comparison with direct investing sits in funds versus direct.
The honest frame for a decision
The defensible reasons to use an SEIS fund are diversification and administration: fifteen holdings instead of two, and someone else chasing the SEIS3 certificates. The undefensible reason is a performance table taken at face value. Price the fund as the asset class (power-law outcomes, long illiquidity, per the suitability analysis), verify the reporting conventions, and let the reliefs do what they actually do: modify your downside at your marginal rate, regardless of whose logo is on the deck.
Common questions
What is a good SEIS fund return?
There is no benchmark number, because conventions differ. The discipline is comparing like with like: net of all fees, pre-relief, separating cash returned (DPI) from paper marks (TVPI), per vintage. A fund showing real DPI across multiple vintages is rare and informative; a young fund showing smooth marked-up TVPI is normal and uninformative.
Should tax relief be counted in fund performance?
In your own arithmetic, yes: relief at your rates really does change your net position. In the manager’s skill claim, no: initial relief arrives regardless of picking ability. Insist on seeing investment performance and the tax layer separately.
Why do SEIS fund figures look better than SEIS statistics suggest?
Mostly selection and marking: surviving funds report, failed ones stop; unrealised holdings are valued at the last round’s price; and post-relief conventions flatter the line. Adjust for those three and fund figures reconcile with the asset class base rates.
Sources
HMRC venture capital scheme statistics; British Business Bank equity tracker research; industry reporting conventions (DPI/TVPI) per Invest Europe guidelines. Analysis framework, not a review of any named fund; correct as at 17 July 2026 and consistent with our rules and limits reference. Education, not investment advice.
