WEBVTT 00:00.700 --> 00:05.280 In this video you will learn how to create exponentially weighted moving averages. 00:05.440 --> 00:09.160 So in a minute you will see how it works and what the differences are. 00:09.160 --> 00:11.200 To a simple moving averages. 00:11.290 --> 00:14.850 But first of all let's import pandas and also met plot lip. 00:15.070 --> 00:19.130 And again we are working with the S&P 500 data set. 00:19.210 --> 00:24.410 So we are importing the data from the S&P 500 since we fire. 00:24.530 --> 00:25.860 Let's have a look here. 00:26.010 --> 00:30.430 And in this video we are only working with the last 10 years. 00:30.460 --> 00:40.630 So from 2 9 2 to 18 and we are actually here overriding our variable S&P 500. 00:40.630 --> 00:42.820 So let's have a look again. 00:42.850 --> 00:50.240 So here our data frame starts at the very last day of the year 2000 and 8 and we've already learned 00:50.260 --> 00:54.030 how to create simple moving averages with the rolling method. 00:54.070 --> 01:00.910 So by defining the rolling window and this example 10 and changing the mean method panels calculates 01:00.970 --> 01:04.960 the rolling mean values which are actually equally weighted. 01:05.110 --> 01:10.390 So in each timestamp the immediately preceding 10 data points are equally weighted. 01:10.450 --> 01:12.780 And now this makes a lot of sense. 01:12.820 --> 01:18.250 However it might be the case that that we want to give more weight or a higher weight to more recent 01:18.250 --> 01:24.640 observations and that we can do this with an exponential weighting. 01:24.640 --> 01:30.460 So let's first of all calculate the simple moving average of the close column here with a rolling window 01:30.460 --> 01:31.120 of 10 01:35.050 --> 01:41.710 and actually the appropriate method to calculate exponentially away that moving averages is the E W 01:41.740 --> 01:48.130 M method and from a mathematical perspective there are plenty of options how to exponentially weight 01:48.160 --> 01:53.860 the observations and the parameters to do so span and half life. 01:53.860 --> 01:54.980 So let's have a look here. 01:57.040 --> 02:04.010 So you have the parameter span and half life and also alpha but actually I do not want to go here and 02:04.010 --> 02:11.600 to a mathematical details and therefore a quite popular way is to use a span. 02:11.600 --> 02:18.500 So here we have also the span parameter and here in the IWM method we actually pass the 10 to the span 02:18.500 --> 02:25.050 parameter and also the FAA the min periods parameter and we pass also the value of 10. 02:25.130 --> 02:31.430 So we need at least 10 observations to calculate an exponentially weighted moving average. 02:32.180 --> 02:35.810 And finally we change the mean method so let's have a look. 02:36.620 --> 02:40.540 So these are here the exponentially awaited moving averages. 02:41.180 --> 02:47.120 And now we can also create two new columns a simple moving average and exponentially weighted moving 02:47.120 --> 02:53.680 average and we actually append these new columns uh to our data frame S&P 500. 02:54.050 --> 02:58.640 So let's do this here and let's have a look. 02:58.670 --> 03:01.090 So here we have our closing prices. 03:01.120 --> 03:06.590 And then the simple moving averages and the exponentially weighted moving averages. 03:06.590 --> 03:11.610 So yeah at the very end we can see we have uh nominal values here. 03:11.990 --> 03:15.160 And now the question is if and to what extent. 03:15.230 --> 03:21.410 Simple moving average is an exponentially weighted moving average just behave in a different way. 03:21.410 --> 03:27.240 And uh the best way to discover this is actually with a graph. 03:27.320 --> 03:31.520 So we're actually plotting here the last two columns. 03:31.520 --> 03:33.670 So we are selecting the last two columns. 03:33.690 --> 03:35.620 And use the plot method. 03:35.630 --> 03:36.710 So let's have a look here. 03:39.660 --> 03:43.340 And we can see here in blue will the line for simple moving averages. 03:43.380 --> 03:49.800 And in green for exponentially moving averages and since exponentially weighted average just place more 03:49.800 --> 03:52.710 weight on more recent observations. 03:52.710 --> 03:59.070 So the green graph actually adapts faster to change as compared to the equal weighted versions of here. 03:59.070 --> 04:00.030 The blue graph. 04:00.360 --> 04:08.550 So for example we can see here that the S&P 500 declines and the it seems like that the green curve 04:08.640 --> 04:13.530 adapts to the declining prices faster than the blue curve. 04:13.530 --> 04:20.010 And also here we can see that exponentially weighted average US actually we are moving faster are adapting 04:20.010 --> 04:23.310 faster to most the recent trends. 04:23.310 --> 04:28.170 So this is the rationale behind exponentially weighted moving averages. 04:28.170 --> 04:31.440 And I hope to see you also in the next video by.