p2p Quantitative risk analysis

It's a nice thought experiment, but the problem is that in reality, originators are certainly correlated as regards their default risk. Hence the central limit theorem may not apply and the overall analysis is moot. Just imagine a recessionary tightening of credit on the European periphery: most originators would be at risk.

For P2P as asset class, we are in a time equivalent of the mid-to-late 1920s stock market boom. Central bank rates are at zero, but these loans yield 10%-15% with relatively short durations and few defaults. Of course, your analysis acknowledges this, since you assume an originator default rate of 5%-10% p.a. instead of the current <1%. However, I think it's likely defaults will be clustered, something that a MC simulation of uncorrelated event series won't capture.

The rest of my thoughts aren't really related to your article but here goes:

In the long run, I think P2P returns are more likely to resemble volatility selling strategies (experiencing steady, consistent growth over many years, punctuated by sudden violent drawdowns), like those of the late optionsellers.com and the formerly popular strategy of buying and holding inverse volatility instruments like XIV.

If you look at it in this way, then we can consider that the "implicit volatility" in P2P is at all-time-low levels while investors continue to sell volatility (i.e. invest more in the asset class). This should make anyone at least a bit uneasy when you consider how this strategy worked for Nick Leeson in the 1990s.