Risk & return data for DPI vs ETH & BTC

I wanted to share some risk & return data I ran for DPI. Thanks, @DarkForestCapital for giving me a nudge to do this.

Arguably, performance history of DPI is relatively short, from September 15th. So all the data needs to be taken with a grain of salt, given the small sample size.

Here is the correlation of DPI returns with BTC, ETH and a hypothetical 50% BTC / 50% ETH portfolio. Correlation measures the degree to which two assets move together. Correlation of 1 means that two assets move together, in lockstep, in the same direction. Correlation of -1 means that they move in opposite directions but with the same magnitude of the move. We can see DPI is more correlated with ETH than BTC, however the correlation is not particularly strong. Over the period of the analysis, DPI was uncorrelated with the S&P 500 , if we adjust for weekends and holidays when the market was closed.

DPI correlation with BTC 0.41 DPI correlation with S&P 500 -0.07
DPI correlation with ETH 0.62
DPI correlation with 50/50 0.57

Correlation is a pretty simple metric. It simply measures the tendency of assets to move together. Correlation doesn’t capture the magnitude of the moves. Which is why we need to look at the beta of DPI to BTC, ETH and a hypothetical 50% BTC / 50% ETH portfolio. Beta shows how much, for a 1% gain in BTC, is DPI likely to move. See, correlation tells us that each time BTC goes up, there’s a moderate chance that DPI goes up with it. Beta, on the other hand, tells us that each time BTC moves by 1% (up or down), DPI is likely to move (up or down) by 0.21%.

DPI beta to BTC 0.21
DPI beta to ETH 0.41
DPI beta to 50/50 0.31
DPI beta to S&P 500 -0.01

We can then look for upside and downside capture of DPI vs BTC and DPI vs ETH. Upside capture means for every 1% that BTC goes UP, how much of that 1%, on average, does DPI capture. 49% for BTC and 86% for ETH.
Downside capture means for every 1% that BTC goes DOWN, how much of that 1%, on average, does DPI capture. 109% for BTC and 130% for ETH.

Upside capture of BTC 48.84%
Downside capture of BTC 109.41%
Upside capture of ETH 85.67%
Downside capture of ETH 129.67%

Here’s standard deviation, which is the primary measure of volatility. Standard deviation is the statistical measure of market volatility, measuring how widely prices are dispersed from the average price. For DPI, annualised volatility based on the roughly last 90 days of daily prices is about 100%, roughly double that of BTC. The only thing this really tells us, is that DPI has so far been a very risky and volatile investment. The daily number of 5.65%, for example, shows that 68.2% percent of the time, DPI’s daily return will be either +5.65% or -5.65% (which is 1 std deviation) away from its average. Annual volatility of the S&P 500 over the last 5 years is about 15% per year.

Standard deviation Daily Annualised
BTC 2.88% 55.00%
ETH 3.73% 71.29%
50 BTC/50 ETH 3.05% 58.18%
DPI 5.65% 108.01%

We can also look at simple performance numbers.

MTD November October Since Inception * (cumulative)
DeFi Pulse Index (DPI) 9.4% 66.8% -38.0% -8.8%
BTC 21.7% 42.3% 27.9% 123.8%
ETH 7.7% 58.7% 7.3% 75.2%
50% BTC / 50% ETH portfolio 14.7% 50.5% 17.6% 99.5%

And risk-return plots for DPI vs. constituents as well as DPI vs ETH and BTC.


This one looks quite wrong actually, compared to TradFi investments. We should expect to see higher returns, assuming higher risk and this is the opposite. BTc presents lowest risk AND highest return option, while DPI is highest risk AND lowest return. Obviously, based on the very limited data that we have so far.



This is fantastic data @verto0912! You’re quickly becoming a beast in the community. Would be really cool if we could create a living dashboard from some of these numbers. Maybe something JD and his team could help with?

thanks @puniaviision, I want to add some data by looking at our competitors and the relationship of their products with BTC and ETH. At that point maybe we can chat about it on one of the calls and see if someone has the skills/time to fully/partially automate it.