- Data for testing
- Point A
- Failed Point A
- Point C
- Pivot range
- W-pivot range -new
- Points A and C through the pivot
- Closing method -new
- Dynamics of Points A and C
- Significant time frames
- Entry and exit
According to The Logical Trader, the ACD values are determined based on the volatility measurements of a particular stock, commodity or financial derivative (see page 14). Fig. 9.1 shows the change in the 30-day average range of H-L and A value obtained for the 30-day calculation period in the case of the 20-minute opening range for SBER stocks. A value is calculated based on the maximum profit for the 30-day period, as described in chapter 2. As can be seen in Fig. 9.1, no correlation between A value and H-L is observed. The correlation coefficient between two data series is 0.04, which indicates the absence of correlation. Thus, A value is independent of H-L and volatility.
Are A and C values and their range stable over time? (It is better to consider the range of A and C values, such as 0-10,5-15, etc., to avoid random deviations of Points A and C). To verify this, calculations were made, the results of which are presented in Fig. 9.1, fig. 9.2 and fig. 9.3.
The black line in Fig. 9.1 shows the evolution of A value in the case of a 30-day calculation period and a 20-minute opening range for SBER stocks. This calculation period is quite short, and A value varies greatly. If A value changed smoothly, then we could predict it the next day using interpolation. But in reality, interpolation will not provide a good forecast for such a short calculation period.
Fig. 9.2 shows how A value and its range change for the current day, if calculated for different periods from 30 to 2400 days, in the case of a 20-minute opening range. As the period increases, A value becomes more stable. For most securities, a period extension of over 1000 days no longer changes A value. Thus, 1000 days seems to be a sufficient period of history to calculate Point A. As can be seen in Fig. 9.2, the range of A value is more stable than A value. Sometimes A value goes beyond its range because of random deviations.
According to The Logical Trader, A and C values are the same for stocks (see page 16). However, the calculation of Points A and C based on the maximum profit for the 1000-day period shows that this is not so. You can see this in fig. 9.3, which shows the change in Point A, its range, and point C for different stocks and futures. Points A and C are not correlated. The range of A values smooths out random deviations of Point A, which justifies the use of the range of A values along with Point A. Most of the time, A and C values are stable, but sometimes change abruptly. The fact that adding only one new day and removing one (from the tail) leads to such drastic changes in the 1000-day period is surprising. Based on common sense, smooth changes could be expected.
To find it, we will calculate the average daily profit from Point A, calculated for different time periods and updated at different intervals.
First, let's calculate the average daily profit for the last 600 days of history. Initially, point A is calculated on the basis of a 20-day period, starting from mark I, as shown in graph (a) below, that is, from 619 to 600 days from mark II. Profit from Point A obtained in this way is calculated for the 600th day back from mark II. Then the 20-day time period is shifted day by day until it reaches mark II, and each time the profit from the new Point A is calculated. The profit for all 600 days is added up. Finally, the average daily profit is calculated by dividing the total profit for all 600 days by 600. After that, point A is calculated based on a 21-day period starting from mark I, and the average daily profit from point A based on a 21-day period is calculated as described above. The process is repeated until the time period for calculating point A reaches the beginning of the history, for example, 1822 days, as shown in graph (b) below.
For practical use, we determine the profit from Point A, calculated for the previous n days, for the next day, which has not yet begun. In the calculations, Point A was updated daily and every 120 days, or approximately every six months. In the latter case, Point A was updated 5 times (600/120=5).
The reason why I chose update of Point A every six months is as follows. The actual values of Point A (C) for stocks and commodities from a specific list can be obtained by a paid subscription at mbfcc.com . I did not find any data on how often A and C values are updated in this subscription. Someone on the forum claimed that they were updated every six months, and he was surprised why it was so rare. So let's check if six months is enough to update Points A and C.
Fig. 9.4 shows the average daily profit from Point A, for the calculation of which different periods of time were used. As you can see from Fig. 9.4, for small time periods, the average daily profit varies greatly. This is due to the fact that for small periods A value also varies greatly, and the forecast of A value the next day is often poor. The longer the time period for calculating point A, the more stable the average daily profit. Therefore, it makes sense to use at least a 1000-day interval to calculate point A. The daily average profit from point A updated daily and every 120 days is not much different (see the black and red lines in Fig. 9.4). Therefore, there is no point in frequently updating point A. Updating once in 120 days or half a year is enough.
Now look at Fig. 9.5, which shows the accumulated profit by days from Point A, calculated for 1000 days daily and every 120 days. Both lines are very close to each other. This once again proves that recalculation of Point A once every six months is enough.
All of the above in this chapter also applies to point C.