Some Graphics and Figures

OK, I couldn’t sleep this morning so I’ve been playing with some data from @asoong and some ideas from @Etienne

I’m playing with methods to visualise $DPI and the underlying tokens… The goal is to help me understand DPI, and try and get some figures to build into future posts / blogs etc. All the figures below are based not eh same data set and valid. Each is useful in understanding different aspects of the data and telling different stories. However, some are better suited to our purpose than others. Even so I think it’s useful to create and share them so we can see what works in different situations / for different people.

I’m not the worlds greatest at generating pretty graphics (I just use a spreadsheet to create static graphics). I’ll happily let others see what they can do to improve them. I think @Kiba was doing something for a web UI showing DPI value over time, but I couldn’t find it.

My background is a Scientist, with an interest in finance, so any corrections on financial terminology are welcomed. I’ve not reviewed my data in great detail, I think there is some minor errors in the calcs, but overall I think it’s good enough for now.

OK, on to the fun part:

All data is 10th September 2020 to 16th October 2020, UNI joined on the 18th Sept.

The first figure shows the relative performance for DPI and the components . All values are normalised to 100 on the day they joined.

Unsurprisingly this shows that DPI is approximately the average of the group with less volatility than any single member (market cap weightings means that DPI should be closer to the larger componets (Lend, Maker, SNX…)

The second figure shows the volatility vs DPI for each component. It compares the change in an individual token to the change for DPI over the same period. E.g. if DPI drops 10% over the period, but $XXX drops 20%, this figure would show that $XXX had a ~10% drop vs DPI. (or possibly closer to 11% relative…?) [If anyone can give me the correct financial terminology for this I would appreciate it]

This figure is very close to the 1st. However, by removing the total market cap volatility (and emotions for the Hodler) it emphasises the relative volatility of the components and so the beauty of diversification.

Figure 3 shows the contribution of each underlying token in $USD to a single DPI. This is based on the price, and the weighting of each in the Set. Other than a general trend down :slightly_frowning_face: , and the addition of UNI, I don’t think there is much jumping off the page.

Figure 4 shows the same information with the components stacked. Compared to Figure 3 it emphasises the DPI value more (~$120 to ~$80). However, I think it is harder to see what the individual components are doing.

Figure 5, is the percentage contribution to a particular days DPI price. Basically, figure 4, divided by the $DPI price every day. As with figure 2, looking at % contribution removes the emotion associated with seeing the DPI price dropping. So this focuses attention on the difference between Tokens over time.

Figure 6 is the % contribution without stacking (unlike figure 5). I think this is better at seeing individual changes (e.g REP kicking on the 12th October), However, it’s not clear that UNI starts on the

Figure 6 is a simple pie chart for the 16th October 2020. This makes it very easy to see the relative contributions from each component. However, there is no historical trends to show whats happening.

Pie

Some final comments, This is a very rough draft. I’m sure there are typos in the text. More importantly the figures can be improved in a number ways: All the titles and labels could be clearer, I’ve abbreviated some tokens in ways that the protocols would not do themselves (which could annoy some of the people we want to build bridges with), and figures 1 and 2 use a different colour set to the rest. If I did this at work, it would be returned to me covered in red ink… But for now “Done is better than perfect”

Some questions:

  • What do the figures tell you about $DPI?
  • Which figures do you find most useful (and in thinking about what)?
  • Can you suggest other graphics to help illustrate this data set better?

Google sheet with data please have a play with it and create some new graphics(should be read only so take a copy)

Regards

OA

9 Likes

Awesome work @overanalyser. Thanks a lot.
This kind of data is great and we definitely need some of these charts to be displayed in Quarterly reports to DPI holders (will start a IIP later once we’ve reached the first product milestones).

I think that we could extrapolate a bit from this chart :

  • In a global down trend, some DeFi tokens are mooning because X and Y good reasons - this is the case of Maker and REP here
  • In an uptrend, all tokens are mooning
    => DPI amortizes the downtrend and follow-up the uptrend

Should be validated with uptrend and downtrend data i guess. Let’s wait for an uptrend :smiley:

3 Likes

@Etienne

Makes sense, I’m sure we / someone could look at the historical data. However, for DPI, I’m wary that going back 6 months would capture the DeFi bull run and we could be accused of cherry picking data.

However, doing a back test just looking at the selected tokens and seeing if we could pull out any patterns (without calculating a “theoretical DPI”) could be interesting [or another indication that I need to get a hobby].

3 Likes

I put together a spreadsheet today to calculate DPI volatility with a view to showing how indexes fair against constituent tokens in terms of risk vs reward. Big caveat of DPI only being around for a month so far, but I think it’s a great way of showing the main benefit of holding an index, risk adjusted returns.

I’ve not done anything like this before so would be good to have it looked over and see what you think.

4 Likes

Sounds good. I would suggest that you upload it to google sheets and share a read only link, then a few of us can have a look.

I’m not expert on financial modelling and volatility calculations, but there are lots of smart people in the community with different strengths.

1 Like

Will do, probably be tomorrow now it’s been a bugger converting it to google and I’ve had enough for this evening :joy:

2 Likes

ok, I forgot to add ETH to figure 1.

During the period examined, ETH was pretty stable between 94 and 109% of the initial value. So it looks like DeFi and DPI were slightly uncorrelated with ETH in terms of USD. Yopu could claim that Maker was the most correlated with ETH for the short period examined.

1 Like

Thanks @overanalyser for the nice charts. As a visual type of person I very much enjoy seeing how those individual tokens compare with $DPI and $ETH in general