Down rounds and liquidation preferences, worked through for an angel

A business angel who holds ordinary shares loses twice in a down round. The percentage shrinks, partly because earlier preferred investors receive extra shares under their anti-dilution clause. And in a modest exit the percentage is not what gets paid, because investors with a liquidation preference take their money back first. In the example below, an angel with 4.8 per cent of a company sold for EUR 9 million receives EUR 360,000 and not the EUR 430,000 the cap table suggests. At EUR 4 million the angel receives nothing.

Down rounds are not rare. On Carta's platform, more than 20 per cent of new venture rounds were down rounds in seven of the eight quarters from the second quarter of 2023 to the first quarter of 2025; in the third quarter of 2025 the share was about 17 per cent, according to Carta's State of Private Markets. That is mostly American data, but the contract terms are the same ones a German shareholder agreement uses.

What is the starting position?

The numbers are invented for illustration. Two founders hold 90,000 shares. An angel invests EUR 200,000 at a pre-money valuation of EUR 1.8 million and receives 10,000 ordinary shares at EUR 20 each: 10 per cent.

A Series A follows. A fund invests EUR 3 million at EUR 9 million pre-money, so EUR 90 per share and 33,333 new shares. The fund's shares are preferred: they carry a 1x non-participating liquidation preference and broad-based weighted average anti-dilution protection. The angel now holds 7.5 per cent of a company valued at EUR 12 million, EUR 900,000 on paper.

Then the plan slips. Two years later the company needs money and finds it only at EUR 4 million pre-money. A new investor puts in EUR 2 million at EUR 30 per share, 66,667 new shares, with a 1x non-participating preference that ranks ahead of the Series A.

What does anti-dilution protection do in the down round?

Anti-dilution protection compensates an investor whose entry price was higher than the price of the new round. There are two families. Full ratchet resets the old price to the new one. Weighted average moves it part of the way, depending on how large the new round is. The broad-based variant, which is the one in the NVCA model term sheet, computes the new price as

new price = old price × (A + B) / (A + C)

where A is the number of shares outstanding before the round, B the number of shares the new money would have bought at the old price, and C the number of shares actually issued. Here A is 133,333, B is 2,000,000 / 90 = 22,222 and C is 66,667. The Series A price falls from EUR 90 to EUR 70.

At EUR 70 the fund's EUR 3 million corresponds to 42,857 shares instead of 33,333, so it receives 9,524 additional shares. In an American corporation this happens through the conversion ratio of the preferred stock. In a German GmbH the investor is typically given the right to subscribe to the additional shares at nominal value, EUR 1 each, in a capital increase the other shareholders have agreed in advance to approve.

Under a full ratchet the fund would be treated as if it had paid EUR 30: 100,000 shares, 66,667 of them new. That is why the method in the clause matters more than the clause's existence.

Without the anti-dilution shares the angel would hold 5.0 per cent. With them it is 4.8 per cent, of a company now valued at EUR 6 million post-money: about EUR 286,000 on paper, down from 900,000. The anti-dilution shares are paid for by those who have no such protection, which means the founders and the angel.

How does the waterfall pay out in a modest exit?

A waterfall is the order in which exit proceeds are distributed. With non-participating preferences, each preferred investor chooses between two things: its preference, here one times the amount invested, or its percentage of the proceeds as if it held ordinary shares. It takes the larger.

Say the company is sold for EUR 9 million. The Series A fund compares 20.5 per cent of 9 million, EUR 1.84 million, with its preference of EUR 3 million and takes the preference. The new investor would receive 40 per cent of the remaining EUR 6 million, EUR 2.4 million, which is more than its EUR 2 million preference, so it converts. The remaining EUR 6 million is shared among 166,667 shares: new investor 2.4 million, founders 3.24 million, angel 360,000.

By the cap table, the angel's 4.8 per cent of EUR 9 million would be about EUR 430,000. The preference costs the angel roughly EUR 70,000. Still a return of 1.8 times the investment, but not the one the percentage implied.

At EUR 4 million the picture is harsher. The new investor takes EUR 2 million first, the Series A fund takes the remaining 2 million of its 3 million claim, and the ordinary shares receive nothing. An angel who reads "4.8 per cent of a company sold for four million" as EUR 190,000 is wrong by EUR 190,000.

The preference stops mattering when every preferred investor is better off converting. For the Series A fund that is when 20.5 per cent of the proceeds exceeds EUR 3 million, at an exit of about EUR 14.7 million. Above that, everyone is paid by percentage.

What would a participating preference change?

A participating preference pays twice: first the preference, then a pro rata share of what is left. If both investors held 1x participating preferences, EUR 5 million would come off the top of the 9 million exit and the remaining 4 million would be split by ownership. The angel would receive 4.8 per cent of 4 million, about EUR 191,000, less than the EUR 200,000 invested, in a sale at one and a half times the last valuation. This is why "1x non-participating" is considered the fair standard and why anything else in a term sheet deserves a question.

What should an angel check in the documents?

Four things decide the outcome, and none of them is on the cap table that the founders send round.

First, the method of anti-dilution in the existing shareholder agreement: full ratchet, narrow-based or broad-based weighted average, and whether it applies to every later round or lapses after a period. Second, the multiple, the participation and the ranking of each preference: whether the newest round ranks first or all preferred investors rank equally. Third, whether the angel's own shares carry a preference. Many early angel rounds are done in ordinary shares, and a convertible loan usually converts into the share class of the next round, which is one argument for it. Fourth, pay-to-play clauses, under which an investor who does not take part in the down round loses protection.

In a GmbH these terms sit in the shareholder agreement and the articles, not in the commercial register's list of shareholders. The list shows nominal amounts. It says nothing about who is paid first. This is not legal advice, and the wording of the actual clause decides the individual case.

A cap table that only shows percentages therefore answers half the question. The other half is a small model of the waterfall: the preferences in their order, and the proceeds to each holder at three or four exit values. An hour with the shareholder agreement and a spreadsheet is enough for one company, and the result is the number an angel should carry in the portfolio for a company that has been through a down round.