The arithmetic
The loss ratio
How many of your investments will return nothing, why that number is higher than feels tolerable, and how to hold it without changing your strategy.
A large share of angel investments return nothing at all. Not a disappointing multiple — nothing. The company winds down, the assets are sold for less than the debts, and the shares are written off.
The exact proportion depends on stage, sector and market conditions, and it is higher at pre-seed than at Series A for obvious reasons. What is consistent is that it is high enough to be emotionally difficult, and that most new angels underestimate it substantially.
The point of thinking about the loss ratio explicitly is not pessimism. It is that a portfolio designed around the true rate behaves completely differently from one designed around a hopeful rate, and the difference shows up in cheque size, portfolio size and how you react to the third failure in a row.
How losses actually arrive
They arrive slowly and they arrive late. A company that will fail rarely does so in year one, when it still has money from the round you joined. It fails in year three or four, after a bridge, after a pivot, after a period of going quiet.
This has a practical consequence for how a portfolio feels. The first two years are mostly good news — companies raise up rounds, marks increase, the portfolio looks strong. The losses concentrate in years three to six, which is when many angels conclude they were wrong about the whole enterprise. In fact the timing is simply what the failure pattern looks like.
It also means paper marks in the early years are close to meaningless. A portfolio marked at 2× in year two that has had no exits and no failures has not demonstrated anything at all.
| Period | What typically happens | How the portfolio looks |
|---|---|---|
| Years 1-2 | Up rounds, a few quiet companies | Encouraging — marks rising, no losses yet |
| Years 3-4 | First failures, first bridges | Discouraging — losses cluster here |
| Years 5-7 | More failures, first meaningful exits | Mixed and confusing |
| Years 8-12 | The winners resolve | Decided by one or two positions |
A constructed pattern to show typical timing, not observed data. Real portfolios vary widely.
Behavioural consequences
A run of failures in years three and four is the point at which most angels change strategy, and the changes are usually wrong. The common ones: raising the bar so far that portfolio size collapses, moving later to reduce risk without adjusting return expectations, and selling winners early to bank something.
The last is the most damaging. Selling a compounding winner at 5× to offset a run of losses removes the position that was going to pay for the portfolio, and the losses it was offsetting were already accounted for in the design.
The defence is to have written the expected distribution down in advance. An angel who wrote "I expect roughly half of these to return nothing" in year one experiences year four very differently from one who did not.
Reducing losses without reducing returns
Some losses are avoidable, and diligence is how. Unassigned IP, a broken cap table, a preference stack that makes any realistic exit worthless, a co-founder relationship visibly failing — these are visible in advance and each one is a loss you did not have to take.
Most losses are not avoidable, because they come from things that had not happened when you invested. Trying to eliminate them means backing safer companies, which reduces the chance of holding an outlier, which is where the entire return comes from. The cure is worse than the disease.
The honest position is to accept a high loss ratio as the cost of access to the top of the distribution, and to spend your diligence effort on the specific, checkable failures rather than on trying to lower the rate in general.
In short
What to take away
- Write down your expected loss ratio before you start, and read it again in year four.
- Losses cluster in years three to six — early portfolio marks tell you very little.
- Use diligence to prevent the specific, visible failures, not to lower the overall rate.
- Do not respond to a run of losses by selling a winner. That is the position paying for them.
From the decoder