Is it feasible for RSI index to judge bitcoin price trend?

Time:2020-6-3

1. Principle of RSI index

The relative strength index (RSI) was created by wells wider in June 1978. It is a kind of technical index to judge the internal strength of asset prices and predict the future direction of price changes by calculating the comparison of market buying and selling forces based on the changes of asset prices in a specific period.

RSI is to analyze the market’s intention and strength of buying and selling by comparing the average closing up and closing down in a period of time, so as to make the future market trend. In other words, it is a calculated value between 0-100, which is used to indicate the prosperity of the market.

Calculation method:

Is it feasible for RSI index to judge bitcoin price trend?

Among them:

  • Rising average: the average of the number of rises in a certain period of time;
  • Average number of falls: the average number of falls in the same period of time.

For example, the n-day average is the average of the number of rises in a certain period of N days, and the n-day average is the average of the number of falls in the same period of time.

If calculating the 14 day RSI, it is necessary to find out the average number of rises and falls (14 for each) in the previous 15 days (excluding the same day), as follows:

Is it feasible for RSI index to judge bitcoin price trend?

Average increase:
(19.29 + 11.38 + 43.34 + 206.29 + 46.57 + 352.38 + 320.19)/14 = 999.44/14 = 111.05

Average decline:
(315.23 + 53.53 + 35.63 + 470.27 + 359.51 + 408.29 + 89.03)/14 = 1731.49/14 = 192.39

14 November 30 RSI:
[111.05 ÷ (111.05 + 192.39)] × 100 = 39.60

2. Basic usage of RSI on the 14th

Because the price of digital assets fluctuates violently, the following methods are generally adopted in such markets:

  • When RSI is higher than 80, assets can be regarded as overbought, which is the time to sell.
  • When the RSI is below 20, the asset can be considered oversold, which is the time to buy.

We use Vitu.AI The installed TA lib package calculates the RSI.

The open source code is as follows:
Is it feasible for RSI index to judge bitcoin price trend?
Is it feasible for RSI index to judge bitcoin price trend?

3. Specific analysis of transaction

Using the data of the past two years (2017-09-01 to 2019-11-30) to test the basic method of RSI on the 14th, although the cumulative yield is as high as 136.16%, we can see that in the past two years, there have been five transactions, as shown in the following table.

Is it feasible for RSI index to judge bitcoin price trend?

If these five transactions are found in the price discount line of BTC, the profit of the five transactions is shown in the figure below. The profit part is green and the loss is red

Is it feasible for RSI index to judge bitcoin price trend?

Roughly speaking, we can find that only the first and fifth signals obtained through the 14 day RSI basic method are profitable, and the three transactions in the middle are all in the state of loss.

Take a closer look at the timing of entry / exit:

Both the first and fifth transactions coincide with the bull market of BTC. RSI is actually more suitable for short-term shocks. However, the time of entry and exit is not very accurate, only a few gains have been made, which wastes the soaring market;

Is it feasible for RSI index to judge bitcoin price trend?

The second and third transactions were in a volatile market, and the entry point was ok (there was a rise the next day after buying), but the exit signal of RSI was obviously inaccurate, which led to the loss of both transactions;

Is it feasible for RSI index to judge bitcoin price trend?

The fourth transaction was also in a volatile market. RSI’s admission signal was not accurate. The buyer was in the middle of the price decline process and had not reached the bottom yet. Although the time of appearance was still OK (the next day after the sale), the early purchase still led to the loss of this transaction.

Is it feasible for RSI index to judge bitcoin price trend?

In general, the RSI’s back test performance on the 14th was not good, and this time the profit was also the result of the blind cat’s encounter with the dead mouse.

4. Why can’t the basic usage of RSI be used in actual combat?

RSI overbought and oversold theory has not given specific operation methods;

RSI is a lagging indicator. Even if there is a signal, it is too late for the signal to come out;

RSI oversold and overbought signals only tell you that the price may rebound or callback, and they don’t tell you where the rebound or callback will go, or even that the rebound or callback may turn into a trend reversal. In this case, there is no way to choose the entry point.

In summary, this study shows that the basic theory of 14 day RSI alone cannot be used in actual combat.

However, we are not saying that RSI is not effective at all, but we need more time to do relevant research:

  • Whether there are golden fork and dead fork effects in long period RSI and short period RSI
  • Deviation of RSI and reversal of RSI
  • RSI resistance and support
  • RSI combined with MACD