Can You Trade on Trending Crypto Tokens?

The idea to implement the Trending Tokens feature seemed quite obvious and simple: we just needed to replicate the pipeline of a similar feature, which we recently described here as Social Trends, and apply it to the subset of words that also serve as slugs and tickers for cryptocurrency tokens. The only new addition was the sentiment ratio (positive/negative/neutral) for visual convenience.

Just a banner for a nice preview. The data is real, btw.

When the feature was delivered it was really exciting to check tokens, that got there and then look for their price to find something like this:

NMR token on top of Trading Tokens

And checking the price showed something like this:

NMR Token price at the same moment

But after we saw a dozen of such examples, we noticed a pattern: the token had already surged significantly, so it might be too late to enter it at the moment it appears in Trending Tokens. At some point, we even joked that this feature could be renamed to “Tokens you’ve just missed”. However, colleagues who are more experienced in trading suggested that it might also be a sign of peak attention, which often precedes a price drop. We tried to make this tool as much newcomer friendly as possible and chances are such users aren’t familiar with stuff like top spotting and short selling even though we have a short (pun intended) introduction to it at our Academy.

So, we needed to check both hypotheses and prove that it really provides value for our customers. And that’s how the following idea emerged.

We decided that user journey for that feature would be something like this:

• at some moment a user checks the Trending Token page
• then the user notices some token, reads the summary of reasons for its appearance in trends, checks the price movement and takes 10–15 minutes more to decide and act (buy or sell).
• …
• PROFIT!

All we had to do is to simulate such behavior with code and data. So we did the following:

• took 1000 random timestamps to simulate 1000 random users opening the Trending Tokens page
• checked which tokens were trending at that moment
• bought those of them, which were rising in price
• added 3 pairs of Take Profit = Stop Losses: for 10%, 20% and 30% respectively
• …

Not too surprising for those who’ve been in crypto for a while

More than half of the long trades ended up with a negative PnL. Lesson learned: taking a long position on coins that appeared in Trending Tokens doesn’t seem to work.

Ok, why not to short then? And here is where things got interesting.

After couple of iterations with the shorting setup we noticed that the following conditions worked:

• excluding BTC, ETH, and stablecoins because they are obviously in a league of their own;
• selecting a token from TOP-3 of Trending Tokens;
• available for marginal trading on some crypto exchange;
• the price graph is currently green (which means that its current price is higher than 7 days ago);

For tokens, satisfying these conditions, the following strategy was applied:

• staying in the short position for 96 hours
• Stop Loss = Take Profit = 30%

How good was it? Well, 72.1% of the resulting PnLs were positive. This means that 720 out of 1000 simulated users could make some money with that strategy.

How much money? On average +3,95%. Not too bad for ~4 days.

If we count couple of clicks as work, of course

No, really! Check out these graphs with distributions of PnLs.

Boxplot for stats OGs

Histogram for data science zoomers

Of course, we had many more ideas on how this performance can be improved. One of the most obvious ones is to include sentiment scores in the decision-making process. However, we deliberately stopped here. Our main goal for that research was never to design an optimal strategy to trade with insights from Trending Tokens. All we wanted to prove is that this feature is useful and can provide value to our clients in their trading decisions.

That’s why it’s important to add some caveats and limitations besides usual hedging phrases like this article was written in educational purposes and it’s not a financial advice so don’t forget to do your own research:

1. We performed the experiment during a rather flat period of the market and used only a relatively short period of it. At the moment we’re writing this research, the market seems to be turning bullish, which means that the particular strategy may not be working anymore.
2. Some of the tokens may have low liquidity, which could make the strategy unsuitable for them.
3. Additionally, the cost of shorting can be high depending on the platform and market conditions, which could impact returns.
4. Since we have tested several combinations of token selection criteria and strategy parameters, our results might be prone to multiple comparisons problem.

Nonetheless, we believe that this simple experiment demonstrated the potential value of the Trending Tokens for traders. However, it’s obviously not a magic bullet. Successful trading requires careful analysis, risk management, and a deep understanding of market conditions. The Santiment team is always working hard to help you with the latter. So, try our new features and let us know what other applications you could find for them in your crypto analytical practice.