Are Nasdaq IXIC Index and Bitcoin Price Movement 'coupled'? EDA before a Deep-Dive

-

 Introduction

Recently, I regularly do a day and scalping trading in Bitcoin(BTC) market. As a newbie investor, I tried to gather useful information that may help me learn more about various investment strategies. 

While visiting here and there wandering around to get more information, I found an interesting post in a Bitcoin community site, saying that the BTC price movement is currently 'coupled' with the Nasdaq IXIC index movement.

I checked if that is true during my trading that night. Interestingly, BTC price movement in one minute interval did seem roughly same as Nasdaq IXIC index one.

I became more interested in this 'coupled' movements because if there really is similarity, I can come up with very nice trading strategies by making use of it. For example, if their relationship is one following the other with a little bit of lag, I can create a python app that does an auto-trading based on that pattern.

As a person who loves to play with data, I want to do a thorough research on it. However, I cannot just dive into a pool without knowing how deep it is.

First, I need to check if the eye-observed coupled movements are really there. Also, there are preliminary questions to answer before collecting vast amount of data and 'torture' the graphic card on my computer.

For that reason, I decided to do a quick exploratory data analysis(EDA) before deep-diving into the topic, the relationship between Nasdaq IXIC index and BTC price movement.


Business Questions

1. Are Nasdaq IXIC index price change and Bitcoin price change coupled as I felt during my actual trading?

2. If so, the correlation between the two is constant or just in certain period for a short term?

3. If there is a correlation, should a further research focus on a scalping trading scale (focusing on one-minute-interval change in a limited term) or a day trading scale (focusing on daily change in a relatively long term)?


Data Collection

I scraped the required data as follows:

1. Nasdaq IXIC index price data : Yahoo Finance API

    -Daily price for the last one year (2021-03-11 ~ 2022-03-11)
    -One minute interval price for the past 5 days (2022-03-07 ~ 2022-03-11)

2. BTC price data : Binance API

    -Daily price for the last one year (2021-03-11 ~ 2022-03-11)
    -One minute interval price for the past 5 days (2022-03-07 ~ 2022-03-11)

(please refer to kaggle or github link below for the scraping code, csv files)


Tools Used

Python (pandas, matplotlib, sklearn)


Time Spent

1 day : 2022-03-11 ~ 2022-03-12


Python Code




Results


1. Nasdaq IXIC index price change and Bitcoin price change seem to be coupled, but only in a certain moment.

2. Similarity in price movements do not seem to take place for a long term or constantly. It does often for a short term like for an hour on a minute-to-minute basis.

3. Further research on their relationship should be in a scalping trading level, which means that I should collect and study minute-by-minute price change data or even more micro one than a minute basis if possible.


Summary of Results



The correlation coefficient ("r") table above indicates that both daily price change and one-minute-interval change for a year and 5 days respectively do not show a significant correlation between the Nasdaq IXIC and BTC price movement. In other words, the two's movements being coupled does not take place constantly or for a long term.



When daily price correlation looked into month by month, one can see that there are certain periods the two's movements are more similar than usual. For instance the price movements in 2021-05, 2021-12, 2022-01 and 2022-03 show a moderate positive correlation. However, the degree of similarity is still not high enough to consider them as a 'coupled' period for a research.


One-minute-interval price change in a daily basis gives out the same insight. The two's movements are couple only on a certain day, not everyday. But, unlike the daily price change case, one-minute-interval price change in a daily basis shows a significantly high correlation that is worth looking into. On 2022-03-08, the two's minute-by-minute price change movements were very similar, having over 0.7 r.  

(hour with over 0.6 r is highlighted as orange color)

When one-minute-interval price looked into hour by hour, one can see that there are hours their movements are strongly coupled even in a weakly correlated day. For example, the correlation of the two's one-minute-interval movements for an entire day was weak on 2022-03-07, showing 0.41 r. However, on the same day at 14:00, their one-minute-interval movements were similar, having over 0.6 r. This indicates that one should look into one-minute-interval price change in an hour basis (or more micro than that) in order to find and investigate the right period of the two's coupled movement. 



* The entire process and result can be found in the links provided in the "Python Code" section:









Comments

Post a Comment