Sector Analysis: Which Category of Coins is Best?

Do certain groups or sectors of coins perform similarly? Here, we look at the extent to which coins can (or should) be categorized into specific sectors, how these sectors have been performing, and potential drivers of total sector growth.

First, let’s look at how these sectors are defined. We categorized over 200 coins into five different sectors. A brief description of these can be found below along with a table showing some examples of tokens in each category.

To gain a handle on how each of these sectors have been performing recently, in Figure 1 we simulated the average returns over the last four months if you held all coins within each sector equally. 

Figure 2 shows the average performance of the top 10 coins in each of these sectors. We see that all sectors have of course performed well recently, with NFTs edging out for first place with about a 14x return on average by the end of March.

Interestingly, we see some similar patterns in these lines — not all sectors are totally independent. We can see that DeFi and Payment sectors perform similarly, the same goes forNFTs and Infrastructure coins. We can quantify this by measuring the average correlation between the top 10 coins in these sectors, as shown in the bar plots. For example, NFTs are highly correlated with each other, as shown by the far left (green) bar in the NFT barplots. NFT coins are also highly correlated with Infrastructure coins, as expected, but they are not correlated with Defi or Payment tokens on average. It is worth noting that Data tokens seem to be somewhat correlated with everything.

But what are some potential drivers of sector movement? Perhaps news articles about these assets are driving prices. 

In Figure 3, we see that news articles about these sectors were initially high in late 2020, corresponding with the early stages of the current bull market. This was followed by a lull in the number of articles in late January, then a subsequent increase again in February. We see that February is when most of these sectors really had significant increases, perhaps because of the number of new articles.

However, when we take a closer look at the general relationship between the number of articles and price change of tokens, we find a significant relationship — at first. 

The left plot in Figure 4 shows the daily percent change in tokens within these sectors, on average, compared to the total number of news articles. At first glance, there appears to be no relationship. However, the right plot shows the absolute value of price change plotted against news articles. We do find evidence for a relationship here — but what does this mean? We suggest that news does have an effect on price, but it can be either positive or negative. Simply looking at the number of news articles may not be so revealing, rather, the content is what matters.

To look carefully at potential drivers of sectors, we zoom in on two of these groups: DeFi and NFTs. We chose to turn our attention to these two sectors because, 1) they are the top two performing sectors recently, 2) they show clearly separate behavior, and 3) they are everyone’s favorite these days.

For those that are unfamiliar with NFTs and DeFi protocols, in short, NFTs are unique digital assets, typically art, that an individual can create, purchase, and/or own, while DeFi (decentralized finance) protocols allow for typically centralized financial services (e.g., loans, savings accounts) to be performed without a middleman. 

As seen in Figure 5, DeFi tokens saw a surge in price in February, before NFT tokens saw their sharp increase in early March. Since reaching peak average returns in late February, DeFi tokens have seemed to flatten out while NFTs have been gaining value through late March.

To try and explain what might be driving these price changes, we look to an obvious source — Twitter activity. There has been a lot of hype around these two sectors lately. In fact, a chapter of this report is entirely dedicated to NFT hype. But can this hype be used to explain price growth?

At first glance in Figure 6, it seems weird that there are so many more tweets about DeFi tokens than NFT tokens given that NFTs have had such massive returns lately. However, there is a simple explanation for this; when people are tweeting about NFTs, they are usually posting about their art that is on a specific platform, not the token associated with that platform. However, tweets about DeFi inevitably tag the token associated with the specific DeFi protocol. Nonetheless, despite differences in the magnitude of tweets, we do see a compelling pattern; DeFi tweets started increasing in February, nearly tripling January tweet counts, aligning with when DeFi tokens saw their strongest returns. Analogously, NFT tokens saw their peak in tweet volume in mid-March, corresponding with their strong increase.

Perhaps it's not the quantity of the tweets, but rather the content of the tweets. Of course, lots of tweets can be a bad thing for a token’s price if people are saying bad things. Sentiment is calculated using The TIE’s natural language processing algorithms that analyze millions of tweets per day to calculate an overall sentiment score about different assets. Here, we show the sentiment of those tweets across a one-month average as dotted lines. If the dotted lines are high, that means sentiment is high.

This plot shows compelling evidence that sentiment plays a huge role in the performance of these sectors as a whole. Both price increases in DeFi and NFTs were immediately followed by peaks in their monthly-averaged sentiment scores. This suggests that these coins are strongly driven by the hype around them.

Is it the sector as a whole that is moving, or individual coins within the sector? Let’s take a closer look at what exactly is going on within these sectors.

We see that not all coins are doing exactly the same thing within each of these sectors. In fact, we limit this analysis to just the top 10 coins by market cap within each sector, because the lower market cap coins are often entirely uncorrelated with the rest of the sector. 

An obvious question is, are coins within a sector more correlated with each other than between sectors? We know from the bar plots in Figure 2 that this should be true for NFT and DeFi sectors, as the the left (green) barplot for NFTs is highest, showing that NFTs tokens are most correlated with each other, while the pink bar is highest for DeFi, showing that DeFi tokens are also most correlated with each other. We can also visualize this concept with a helpful tool in any data scientist’s toolbelt — a correlation matrix. 

This 20x20 matrix shows the correlations between each pair of the top 10 tokens by market cap in the DeFi and NFT sectors. Dark shading indicates high correlation (i.e., the two tokens’ prices tend to move together). We see a dark square in the top left and bottom right sections of the figure. This suggests that there is high correlation within each sector. However, in the other corners of the plot, we don’t find such dark shading. This suggests that DeFi tokens are not typically correlated with NFT tokens. 

It is interesting that UNI and LUNA seem to be strongly correlated with everything. In fact, LUNA is more strongly correlated with NFT tokens than its fellow DeFi tokens. There are a few potential reasons for this; in the case of Uniswap, it is the coin with the highest market cap out of all coins analyzed. Thus, it is likely to represent the performance of the entire crypto market at large, not just the DeFi Sector. LUNA, on the other hand, simply had a very late surge in price compared to the other DeFi tokens, as shown in Figure 6 (light green line). This relatively late increase corresponded strongly to the increase most NFTs saw in mid-March.

Is it more lucrative to try to trade the individual coins within these sectors, or to treat these sectors like indexes, and hold each coin within sectors equally? In Figure 10, we show the performance of a really simple trading strategy (of course, we do not recommend this strategy). This is just to illustrate the pros and cons of trading sectors vs. individual coins.

The strategy is based on what we see in Figure 7, with price increases in the sectors generally followed by peaks in the sentiment. Here however, instead of considering the average sentiment across the entire sector, we consider the sentiment of individual tokens. If the sentiment is above a specific threshold, we buy and hold that token for one week — the approximate delay between sentiment and peak returns in Figure 7. 

We find this simple strategy yields returns lower than holding all the coins of any individual sector. However, these strategies also had a better Sharpe Ratio. (The Sharpe Ratio is effectively a measure of how consistently there were positive returns.) For example, as shown before, the performance of DeFi as a whole in March (red line in Figure 5) has been relatively flat, whereas this strategy shows consistent returns throughout March. This can be explained by the nature of the strategy; if you trade into a coin only when sentiment is high, it is more likely that the price of that coin will increase in the near future. But if you only hold the asset for a week, you are likely to miss out on other large increases that cannot be associated with sentiment analysis. Thus, holding everything all the time, you get everything, the ups and the downs. Conversely, by trading on this sentiment strategy you get mostly ups, but definitely not all of them.

This highlights that there are actually benefits in treating these assets as entire sectors rather than individual coins. When trading individual coins, you have a higher probability to miss out on key movements the entire sector makes.  

As a last figure, we look at what would happen if you were to perfectly time swapping from holding the top 10 DeFi tokens to the top 10 NFT tokens. The best time to swap was in mid-February. Although DeFi tokens netted a return of about 10x and NFTs 14x, making this single swap would have resulted in an impressive 50x return in four months.

These results show that there are potential benefits to treating coins as sectors rather than individual tokens. Price change in an entire sector is a bit easier to predict based on the Twitter activity around the component assets. However, hype about individual tokens is a bit harder to confidently relate to price.