The index research brought to you this time is Tokenview’s unique exchange index data. First, we will introduce the definition of the index, and then this article will bring research conclusions worthy of in-depth attention.
Exchange deposit and withdrawal
image description
Exchange deposit amount and price
image description
image description
The number and price of exchange withdrawals
We all know that exchanges’ turnover rate (Volume) data brush volume and exaggeration are relatively common, and their credibility is low.The data on the chain shows its own advantages here: because there are thresholds for data acquisition and statistics, and the cost and difficulty of falsification are relatively high. At present, the on-chain indicators of exchanges are relatively real.
Therefore, through these indicators, we can most intuitively see the actual activity of trading users, that is, investors in the secondary market.
Comparing the active trend changes of different currency prices and investors, we found some interesting conclusions.
The correlation between the number of transactions and the price is weak
From the perspective of BTC only, the trend relationship between the number of recharges and withdrawals and the price shows less information. Among them, the trend of the number of recharges and prices can be observed to a certain extent, and the trend of the number of withdrawals and prices is not. Obvious follow-up phenomenon.
Therefore, such indicators really have little reference significance for investment.
Amount indicators are more referable
Exchange recharge amount
Taking Bitcoin as an example, we selected the price from January 1, 2018 to September 25, 2019 for comparison with the exchange recharge amount indicator. See the figure below for details.
fallfallTrends, such as around February 2018, around December, and around January 2019 on the picture.
Therefore, we extracted all the days when the index was greater than 50,000 and the corresponding BTC price, and compared the price and rate of return after 1 day, 3 days and 7 days after the threshold was exceeded. The details are as follows:
The total number of days when the indicator exceeds the threshold is 23 days, so it can be seen from the above table that in more than half of the cases, the price of BTC will drop after 1 day, 3 days, and 7 days, and the probability of falling after 3 days and 7 days is a little higher. The highest single drop was 25.28%, which occurred after 7 days.
If you take this indicator as a short-selling signal and place a short order every time the signal is generated, the cumulative returns of closing positions after 1 day, 3 days and 7 days are all positive.secondary title
Exchange withdrawal amount and price
Only from the results observed in the figure, the withdrawal amount and price do not have obvious signal characteristics like the recharge amount, but an interesting phenomenon of the withdrawal amount is that we can see that it is within the range of the lowest price of BTC (18 from December 2019 to May 2019), the withdrawal amount has obviousrise。
So we conducted a more in-depth study, selected the BTC price of $6500 as the threshold, and compared the average withdrawal amount when the BTC price was lower than $6500 (inclusive) and higher than $6500. The results are as follows:
The total number of days when BTC price exceeds $6500 is 383 days, and the total number of days when it is less than or equal to $6500 is 250 days.
We selected the original value of the indicator and compared it with the results of the two types of data processing. The results show that in the three cases, the average withdrawal amount is higher when the BTC price is lower than $6500.
Therefore, the cash withdrawal amount indicator shows a certain degree of correlation with the priceon the contraryThe trend, especially when the price is low in the near future, is strong, in other words,The cash withdrawal amount indicator may have some reference value for the bottom-hunting stage.
*Considering the surge in indicators in late July 2019, this part of the data may affect the analysis results of the final model, so we have done some smoothing data processing.
