Cognitive bias in cryptocurrency naming
Why naming a cryptocurrency ‘NAS’ instead of ‘XZC’ generates familiarity and boosts purchase interest in the first week after exchange listing
In the bestseller Thinking Fast and Slow, Daniel Kahneman takes a full chapter to explore how the way concepts, ideas, pictures or text are presented to us and the effort it takes for us to understand them has an effect on our attitude and emotions towards them. Thus, hard to see images or text fonts are usually related to negative attitudes or displeasure when compared to clear, outlined images or fonts. He calls this phenomenon as the pleasure of cognitive ease.
In the same line, he explains that
easily pronounced words evoke a favorable attitude. Companies with pronounceable names do better than others for the first week after the stock is issued, though the effect disappears over time. Stocks with pronounceable trading symbols (like KAR or LUNMOO) outperform those with tongue-twisting tickers like PXG or RDO — and they appear to retain a small advantage over some time. A study conducted in Switzerland found that investors believe that stocks with fluent names like Emmi, Swissfirst and Comet will earn higher returns than those with clunky labels like Geberit and Ypsomed.
The main idea behind this hypothesis is that the readability of the name of a market instrument has a causal relationship to how much we will remember it. At the same time, remembering something leads to familiarity and, unless the memory expresses the contrary, we often like or prefer things that we remember. This, in turn, would lead to an increase of interest to purchase or prefer the asset when given the option to buy one.
readability → memorability → familiarity → likability → purchase interest
Zajonc called this phenomena as the mere exposure effect. […] [He] argued that the effect of repetition on liking is a profoundly important biological fact. To survive in a frequently dangerous world, an organism should react cautiously to a novel stimulus, with withdrawal and fear. Survival prospects are poor for an animal that is not suspicious of novelty. However, it is also adaptive for the initial caution to fade if the stimulus is actually safe. The mere exposure effect occurs […] because the repeated exposure of a stimulus is followed by nothing bad. Such stimulus will eventually become a safety signal, and safety is good.
I assume that advertisement and marketing nurtures as well from the exposure effect and perhaps explains the saying “there’s not such a thing as bad publicity”. This is, doesn’t matter what they say about our product, it only matters that people talk about it, which leads to familiarity, which leads to likability and so on until causing a bigger purchase interest.
In order to test whether the ticker readability effects the purchase interest of a cryptocurrency, I conducted a small research project to find out if this is really the case. You can find the Jupyter notebook in the end of the article.
Data
The main hypothesis behind this research project is that a readable and memorable name for a cryptocurrency should influence the price of the asset the first week of release or listing on an exchange.
In our case, we are going to work with a more or less homogeneous group of cryptocurrencies listed in the Binance exchange that fulfill these requirements:
- The ticker name (the short name) of the cryptocurrency should be 3-letter long. E.g. ETH, TRX, ZEC…
- The cryptocurrency should be traded against Bitcoin. For instance, the pairs ETH-BTC, TRX-BTC, etc.
114 cryptocurrencies fulfill these criteria.
Now we will categorize each cryptocurrency as “readable” or “not readable”. We are going to use two rules:
- In one case we will consider as readable a cryptocurrency with a 3-letter ticker with at least one vowel and unreadable if the cryptocurrency ticker is only formed by consonants. For instance, ETH would be readable and ZRX would be unreadable.
- In the second case we will consider as readable all cryptocurrencies with one vowel in the middle-position and unreadable all the rest. For instance: BAT would be readable but BSV not.
We will then retrieve all the 1-hour candlesticks for the first week after the cryptocurrency was listed on the Binance exchange.
Tickers for the first rule classification:
WITH VOWEL:
ADA, ADX, AGI, AMB, ARK, ARN, AST, BAT, EDO, ELF, ENG, ENJ, EOS, ETC, ETH, EVX, FET, FUN, GAS, GTO, HOT, ICN, ICX, INS, KEY, LUN, MCO, MDA, MOD, NAS, NAV, NEO, OAX, OMG, ONG, ONT, OST, PAX, POA, POE, REN, REP, REQ, SKY, SUB, SYS, VEN, VET, VIA, VIB, WAN, XEM, ZEC, ZEN, ZIL
WITHOUT VOWEL:
BCC, BCD, BCH, BCN, BLZ, BNB, BNT, BQX, BRD, BSV, BTG, BTS, BTT, CDT, CMT, CND, CVC, DCR, DGD, DLT, DNT, GNT, GRS, GVT, GXS, HSR, KMD, KNC, LRC, LSK, LTC, MFT, MTH, MTL, NXS, PHB, PHX, PPT, QKC, QLC, QSP, RCN, RDN, RLC, RPX, RVN, SNM, SNT, TNB, TNT, TRX, WPR, WTC, XLM, XMR, XRP, XVG, XZC, ZRXAnalysis
For the analysis, we will calculate the percentual returns for each cryptocurrency and we will aggregate all the returns for each of the categories specified above. Finally, we will compare the distributions and we will perform a t-test to statistically assess if the two samples can be considered to be drawn from different distributions. Ideally, we expect that the readable cryptos will have higher returns than the nonreadable ones due to the described mere exposure effect.
Results
First let’s look at the distribution of returns for the tickers with/without a vowel in the middle:
It seems that there are some differences, but are these differences statistically significant? Let’s run a t-test for all the data ranging from 1 day to 7 days after exchange listing:
Surprisingly, the mean return for the readable class (with middle vowel) is in fact higher than the unreadable class. The difference seems to be bigger at start but it seems that decay after time. Interestingly, p-values are all low and one point is statistically significant at alpha=0.05 (5%).
Let’s look now at the results for the vowel vs no-vowel ticker classification:
The mean difference seems to be smaller at first sight but we can see at least three points with statistical significance! After all, the hypothesis seems to be true at least to certain extent.
Take-home message
While correlation does not always imply causation, we can see that cryptocurrencies with readable symbol or ticker names display in fact a higher mean return (significant at different points at alpha=0.05) the first days after exchange listing. This suggests that it may be plausible that when people are given the choice to buy an asset (specially during the altcoin boom which took place in 2017 when most of the cryptos analyzed here were listed), people may tend to buy more the cryptocurrencies with the most readable and memorable names.
Jupyter notebook
This project is part of our research at CryptoDatum.io, a cryptocurrency data API that aims to provide plug-and-play datasets to train machine learning algorithms. If you are a machine learning practitioner, get your free API key and play with them yourself at https://cryptodatum.io
