Hello people,

I recently wrote a paper analyzing the impact of paydays on bitcoin's price. I haven't been checking lately whether this still holds true but the pattern should still be consistent. So my finding should come useful to those who trade. I think I would've appreciated this tip when I was trading. The below is an abridged version of the paper, which I wrote so that a layman can understand. Good luck.

Abstract

This paper investigates whether there exist any significant and recurring impact of paydays on the price formation of bitcoin. Using bitcoin’s data- obtained from Coinmarket.com- from 7/6/13 to 3/28/18 in UTC time, this paper finds that Mondays experienced consistent positive returns at a significant level, suggesting that biweekly paydays, which fall on every Fridays for large portions of our population, caused the higher than the average return on Mondays. The reasoning for why higher than the average return was experienced on Mondays- not Fridays, the day of the biweekly paydays- is provided for the rest of the paper.

Introduction

There is an ongoing debate on whether bitcoin’s current market value and eager support from a portion of our population is the result of a rational market behavior. A purpose of this research is to shed some light on this issue, the other being to provide a useful tool for investment and trading strategy in cryptocurrency market, by analyzing the impact of paydays on bitcoin’s price formation.

A result of this paper’s research shows that Mondays (in UTC Time) had consistent positive returns throughout bitcoin’s history, from 2013 to 2018. This pattern is more prominent during bitcoin’s bullish periods but, even during bearish periods, bitcoin still performed better than the most other days of the week. The recurring and persistent pattern suggests that there is a common source of the cause behind this pattern, which I hypothesize that the source is the biweekly paydays.

The main thesis, {T1}, of this paper, on what the possible cause of that pattern may be, is that some portion of our population exercises in biweekly purchasing of bitcoin due to their biweekly paycheck, which are received on Fridays. The intuitive reasoning behind the hypothesis is as follows: first, I derived from my personal observation of myself and others that, in general, increase in one’s wealth conduces one to spend more, which then I generalized this observation to the general population. This generalized observation will be then termed into a postulate, {p1}, to help form the thesis.

Using the postulate, {p1}, we can deduce that the weekly concurrent increases in wealth of biweekly-payday-workers, who form a significant part of the general population, would have a noticeable effect on markets on weekly basis, including bitcoin market. This reasoning sounds plausible, however, since the result shows that Mondays and Tuesdays had experienced greater demands when the paychecks are received on Fridays, we must first find a plausible explanation for the lag in days of their purchases for the main thesis to hold.

Thus, I have formed hypotheses for why Monday, not Fridays or weekends, had experienced greater demands: one, {H1.1} people generally designate Fridays and weekends for leisure activities, thereby prompting the population to purchase on the following Mondays- which is the first designated work day after the wealth increase; two, {H1.2} the population do purchase bitcoin on Fridays but the observable effect of their demand lags by couple days, and the reason being that Coinbase, the platform which most Americans purchase bitcoin with fiat, lags by a few days on “restocking” its holdings of bitcoin, thereby lagging in the observable impact.

I will first test the hypotheses- {H1.1} and {H1.2}- under the assumption that the main thesis, {T1}, is true. Depending on which hypothesis holds, the strength of its support for the main thesis will differ. Next, a support for the thesis will be established by further analysis.

Argument Outline

Postulate {p1}: Increase in one’s wealth conduces one to spend more.

Main thesis {T1}: Since {p1}, some portion of the population purchase bitcoin every other week upon receiving their biweekly paycheck, which are received on Fridays.

Observation: Bitcoin consistently experiences higher than the average return on Mondays.

The Hypotheses to Explain Why Mondays, not Fridays:

{H1.1}: People generally designate Fridays and weekends for leisure activities, thereby prompting the population to purchase on the following Mondays- which is the first designated work day after the wealth increase for biweekly-paycheck-workers.

{H1.2}: The population do purchase bitcoin on Fridays but the observable effect of their demand lags by couple days, and the reason being that Coinbase, the platform which most Americans use to purchase bitcoin with fiat, lags by a few days on “restocking” its holdings of bitcoin, thereby lagging in the observable impact on bitcoin’s price.

Reasoning

There are various ways to test the hypotheses, thus a sensible course of action would be to minimize the steps needed to arrive at the conclusion. I believe that the following method for testing the hypotheses minimizes the steps:

First, we can divide bitcoin’s history of price formation into various periods, which then we can compare the characteristics of the Monday increases between the various periods. The periods were divided based on bear and bull market patterns, resulting in 5 distinct periods. There are few variables that we can use for comparison. The first is the delivery period of bitcoin. Coinbase takes 2-8 business days (not including the day of the purchase) to deliver purchased bitcoins to the purchasers; this delivery period will be henceforth called as the waiting period.

If many people purchase on the same day, then many people will receive their bitcoin following that waiting period. Then, one can expect at least some portion of that group will decide to sell bitcoins on the day they are received, leading to an increase in supply after the waiting period. Thus, if there exists a consistent and significant price hike (increase in demand), then one can expect a consistent price fall (increase in supply) following the waiting period.

To find clues to what the length of the waiting period might have been, an auto-regression analysis was conducted, which shows the relationships that form between days. Say, you can run an auto-regression analysis on bitcoin's price and find out whether bitcoin's price on any given day generally had moved in the same direction or the opposite direction with the following days of your choosing, such as a general relationship between one day to the following day. The results of my auto-regression analysis along with my analysis on bitcoin's price patterns of days of the week discredited the hypothesis, {H1.2}. If you wonder how I arrived at the conclusion, I utilized my personal experience on how long the respective waiting periods were for the periods I had chosen and then checked whether my experience coincides with the auto-regression results, which I found to coincide precisely. Thus, {H1.2} was discredited.

To find support for the hypothesis, {H1.1}, I conducted an analysis on bitcoin's volume pattern by days of the week. The results showed that weekends generally recorded low volume activity than other weekdays. Monday recorded below average as well. With these findings, we can deduce that the general population generally does not participate in market activities such as purchasing bitcoin with fiat, in other words, it is a human tendency to not purchase bitcoin on weekends, perhaps to participate in our favorite weekend activities. Therefore, this lends support for the reason why Mondays experienced higher than the average return- not Fridays, the day that many people receive their paychecks- because biweekly payday people also tend to not participate in bitcoin's market activities, including purchasing bitcoin, because it is a human tendency. This finding alone gives us an explanation for the reason for the lag in days; a support for {H1.1} was found.

Finally, since we now have a more likely explanation for why Mondays, not Fridays, experience higher than the average returns, we must now find support for the main thesis, {T1}, that the increase in price hikes on Mondays are caused by biweekly payday group, not some other group. Although I had relied on a postulate, {p1}, to form the thesis, to provide an actual support for it, I relied on the results of the previous analysis on bitcoin's volume activity.

In the analysis result on volume activity, I had found that Mondays did not particularly record high volume activities on average compared to all the other days. More precisely, Mondays generally recorded lower than the average. Thus, this reveals a trait of the general population, that the general population does not particularly favor Mondays to participate in market activity, including making bitcoin purchases or selling. Since our general tendency is to not participate in market activities on Mondays, for some group to consistently purchase them on Mondays despite our general tendency to avoid participating on Mondays, this group's consistent action must be driven by certain urges of human nature: the postulate, {p1}, having more money makes us want to buy more.

The basic human urges are universal. We are not much different from another. We dislike similar things; we like similar things in general. Any deviation from our general tendencies indicates there are other tendencies of ours that are in play. Given more imagination or information, we can find what is driving that deviation, which is what I did to come to my conclusion.

This sums up the basic argument. If you want to see the actual analysis results, just send me an email and I will send you the complete paper.

Email: [email protected]