Python: Exponential Moving Average (EMA) Mathematics and Stock Indicators

Key Takeaways

what's going on everybody Welcome to the next uh video in our mathematics and stock indicators with Python tutorial Series where we left off we were doing uh just a simple moving average here and now I want to show you guys how to do an exponential moving average so moving right along uh we're just going to Define exp moving average again it's going to have a values and a window don't forget your colon then we do weights equals numpy X for exponential mp. Lin space and within here a negative one with a decimal point and a zero with a decimal point and then a window um just like before I think I'll leave out the explanation of lens space there we just want to keep this quick and basic uh weights uh is going to actually be a divided equals and what this work how this works is is the easier way to explain this is you'll probably see a plus equals a lot and so plus equals just basically means weights the original version of Weights plus one and then it redefines that variable so at the same point as the negative equals it's just like weights negative equals what weights. sum um and that's that's basically how that's working uh so now the next thing we want to do is now we're going to do a equal mp. convolve and values weights again and then this creates basically an array and then the way we can specify um what part of that array we want is like this so everything up to the length of value so how many values do we have this goes to the length of those values subsequent quently um we've got a and then to the window length equals a window and then again we just return a and if you'll notice we actually uh use the convolve function with its default so we didn't say valid we could do it either way it's not as big of a deal with exponential moving average so we'll just leave it uh as the default so again now we'll just come down here and we'll print exp moving average and again we'll do data set and three save that run that and now the first thing it printed out was our the same simple moving average we have already uh done and then the next thing it printed out was uh our exponential moving average so the idea of the exponential moving average versus the simple moving average is the exponential moving average is taking more recent data and applying a a larger weight to that more recent data than the older data so right that's like our weights here and we applied the weight here so um just to explain a couple things now uh numpy's convolved function what it's doing is it's it's returning a linear convolution right it's convolving two sequences right so in this case it's returning the two sequences with the length um or at least to the length of those values now the reason why I kind of just want to leave uh convolve out of it and same thing even with numpy's exponential uh functionality convolve is basically what's running our exponential or it's it's what's doing the moving average part for us and the reason why I just kind of want to leave those out for now is it just makes it way more efficient for us to do it this way since moving average is used and if you like say you're using like a 200 moving average which is used pretty pretty commonly like 20 and 200 um this saves a lot of processing power so numpy a lot of times comes in to save the day when it comes to processing um but to explain what's going on behind the scenes with convolve is is somewhat confusing and so there's not too much uh reason to cover it because I think moving average the definition of a moving average is simple enough to the point where I think you guys understand what a moving average is uh and the use using of numpy to make it a lot more efficient isn't really as necessary right so when you do a moving average you just know like okay I'm going to use convolve if you are curious you can go into the source files and just look at the convolve function and see what it's doing same thing with the exponential function here uh but anyway I'm not going to cover that because really what I'm trying to do in this series is just simply uh teach you guys how to do the simple equations and so a lot of the L latter equations that we're going to be covering here um to do moving averages we're going to use this simply because moving averages become very processing intensive so using numpy solves that problem uh for the moving average calculation but for the rest of the calculations we aren't doing that so if you feel like you're really lost at these uh first two I almost didn't want to cover them right away but you pretty much have to because they're like the most basic formulas out there um if you feel like you're a little confused at this point don't worry it's probably uh pretty common just understand that we're using these uh purely to save on processing if anybody wants me to I could cover a uh just a pure python without numpy uh example of doing a moving average but it just doesn't make too much sense cuz it's really processing intensive so anyway uh hopefully I didn't lose too many of my viewers right away since this numai stuff can get kind of confusing um the exponential that fairly uh basic it just does the exponential what of all of the elements in the input array whatever you're stuffing in there and then the numpy repeat option option is a way to multiply um something right so we're just saying what we want to multiply and then we're dividing it by that window so again like if you ran a repeat function on let's say we ran a repeat function on data set and we said repeat uh three right so if we did np. repeat data set 3 it would do 111 555 777 222 666 777 888 and so on and so that's what's happening in the repeat function again it we're just using numpy uh because it's very efficient so that's going to include uh both of the moving average and exponential moving average now in the next videos we're going to be uh moving right along and starting to conquer a bunch of the typical stock indicator formulas so anyway hopefully uh you guys are interested hopefully like I said I didn't lose too many people with the numpy stuff uh even though moving average and exponential moving average is easy um and the you know coming up formulas are in theory more complex it actually programming wise gets easier so anyway um as always thanks for watching thanks for the support the subscriptions and until next time