Angel investing returns are lumpy, not average. The honest summary from the UK data is this: the typical individual investment loses money, a small number of winners carry the whole portfolio, and the strongest hard UK figures are now well over a decade old. The mean looks fine. The median is grim. Both are true at once, and holding both in your head is the start of reading the numbers properly.
This is the pillar page for that picture. It pulls the realistic returns story into one place: the shape, the dated UK figures, the gap between what angels expect and what they get, and a worked illustration of how the loss distribution falls across a book. The live beginner posts teach the mechanism. This page is the numbers and the arithmetic on top of them.
The mean is fine. The median is grim. Both are true at the same time.
The shape of angel returns, plainly
Angel returns follow a power law: a few investments return many times the money, most return little or nothing, and the winners are large enough to outweigh all the losses combined. The mean return can be healthy while the median investment is a loss. That is not a paradox, it is the defining feature of the asset class.
Put plainly, you are not trying to be right on average. You are trying to own one or two of the rare outcomes that pay for everything else. A book where every company does fine and nothing fails is, counter-intuitively, a book that probably missed the tail. The losers are the cost of buying enough chances at a winner.
The mechanism behind this is covered in the live explainer how angel investors make and lose money: the power law. This page assumes you have the idea and goes to the figures.
The UK numbers, dated honestly
The strongest realised UK data comes from one study, and it is old. Wiltbank's Siding with the Angels (NESTA, May 2009) tracked 158 UK angels across 406 exits, with data running to late 2008. It found a mean return of 2.2x over 3.6 years, roughly a 22% internal rate of return.
That headline hides the spread. In the same data, 56% of exited investments returned less than the capital invested, most of those a total or near-total loss. Only 44% of exits showed a gain at all, and about 9% of exits returned 10x or more. That thin slice of big winners produced the bulk of the entire return. The earlier UK evidence agrees on direction: Mason and Harrison's 2002 study of informal venture capital found 34% of investments were a total loss. For the full loss-rate breakdown, see the real loss ratio. The point to sit with is the date. The British Business Bank's 2025 work has a first-ever angel section, but it reports activity and sentiment, not realised returns. Nothing newer has replaced the 2009 figures.
Hope versus outcome
The most recent UK angel study reports expectations, not results, and they are rosier than the realised record. The ERC's A Nation of Angels (Hart, Wright and Fu, January 2015), drawing on networks representing around 8,000 angels, found angels expected 24% of investments to return under 1x, 44% to return one to five times, 19% six to ten times, and 13% over ten times.
Read those expectations next to the 2009 realised data and the gap jumps out. Angels in 2015 expected roughly a third of their bets to be strong multiple winners. The 2009 outcomes had 56% of exits losing money and under a tenth returning 10x or more. Expectations are not lies. They are the optimism the asset class runs on, and they are exactly why an angel book built on expected returns tends to disappoint against the realised ones. When you model your own book, the honest input is the dated outcome, not the survey hope.
What a realistic book actually does
To make the arithmetic concrete, apply the NESTA loss shares to a notional book of 40 names. This is an illustration drawn from the 2009 distribution, not a forecast and not a target, and it prescribes no number of names to hold. It simply shows what the published percentages look like spread across a round number.
Take the 2009 exit shares at face value. Roughly 56% below capital puts about 22 of the 40 underwater, most of those near-total losses. The 44% showing a gain is about 18 names, and the 9% returning 10x or more is around three or four. In this illustration, those three or four are doing nearly all the work. Strip them out and the book is a loss. That is the power law landing on a real count: most names return little, a tiny group carries everything.
This is also why diversification and the loss rate travel together. If the return lives in a thin tail, a book needs enough separate shots to have a fair chance of catching one, which is the logic behind diversification, how many startups make a portfolio and portfolio construction. The number that suits you is yours to draw. The maths only shows why too few names and a power-law return sit badly together.
Reading your own book, and a note on what this isn't
Two things distort the picture while a book is young. The first is timing: early portfolios look like losses because the failures surface before the winners exit, the pattern set out in the J-curve. A book three years in can show red ink and still be on track. The second is relief: SEIS and EIS loss relief softens the cost of a failure, but relief is a cushion, not a return, and it never changes the underlying loss rate.
One thing this page is not: a recommendation. It does not tell you to build a book of 40, to hold any particular number of names, or to commit capital at all. It describes what named studies found, most of them old, and what the arithmetic implies. Early-stage companies are among the most likely investments you can make to lose the lot. Confirm any figure at its source, read the position in HMRC's guidance on the venture capital schemes, and take FCA-regulated advice before you put money in.
Frequently asked questions
Do angel investors actually make money?
On average, the best UK data shows a positive mean: NESTA's 2009 study found 2.2x over about 3.6 years across 158 UK angels. But most individual investments lose money. In that same data, 56% of exits returned less than the capital put in, and the overall gain came from a small tail of big winners. The mean is positive while the median investment is a loss. These figures are dated to 2009 and have not been replaced by newer UK realised data.
What return do angels get?
The most-cited UK figure is a mean of 2.2x over roughly 3.6 years, from NESTA's Siding with the Angels in 2009, which is about a 22% internal rate of return. That is a mean, not a typical outcome: the same study found 56% of exits lost money and about 9% returned 10x or more. The number is old, so treat it as the strongest available read rather than a current one, and confirm it at the source.
Why do most angel investments lose money?
Early-stage failure is the base rate. Most young companies do not reach a profitable exit, so most individual angel investments return less than the money put in. The model does not rely on most bets working. It relies on a few large wins outweighing the many losses, which is the power law. NESTA's 2009 UK data showed 56% of exits below capital and a small tail of 10x-plus winners carrying the total return.
Is there newer UK data on angel returns?
Not for realised returns. The most recent UK angel study, the ERC's A Nation of Angels in 2015, reports what angels expect to make, not what they made, and those expectations are rosier than the 2009 realised losses. The British Business Bank's 2025 work covers angel activity and sentiment, not realised returns or loss ratios. So the strongest realised UK figures remain NESTA's from 2009.
How many angel investments do I need to make money?
This page will not set a number for you, because that is an investment decision and depends on your circumstances. What the data shows is that returns concentrate in a thin tail, so a book needs enough separate bets to have a fair chance of holding one of the rare winners. The research describes that shape; the number that fits your situation is yours to decide. This is general information, not financial advice, and you should take FCA-regulated advice before committing capital.