OK, here it is: I hate moving averages. Well, maybe “hate” is a strong word. But what gets me is that they’re everywhere – and they’re NOT REALLY THAT GOOD.
When I hear commentators talk about “bouncing off the 50-day moving average” or the “200-day moving average is rising so the trend is up” – I want to shout at my computer screen. It’s so arbitrary. Why 200-day? What if the 200-day moving average was rising but the 175-day was falling?
You think I’m kidding don’t you? No, I’m serious. I can’t stand moving averages. And here are another couple of reasons why:
- Moving averages assume the market is either going up or down. They are either rising or falling. But the market isn’t that simple – there aren’t just 2 states. In fact there are 3 states: the market is either rising, falling or in congestion. A moving average can’t help you determine this third congestion phase.
- There are literally millions of different moving averages. Simple, exponential, weighted, etc. etc. Then 9 bar, 13 bar, 21 bar, 50 bar, etc. etc. Add all those combinations together and you literally have an infinite number of combinations. And so which one is “right”?
But here’s the real kicker – there is a better way!
There is an analytical technique to determine uptrends, downtrends and market congestion. It works for all markets, all timeframes. Yes, it’s a little more complex than a moving average calculation – but that’s the point. We now have the computer power and the digital signal processing techniques to do this. And yes, I’m talking about the Hilbert Transform and the Better Sine Wave indicator.
Increased computer power means we can do better than a moving average!
- Click for video transcript …
It is Thanksgiving weekend, so if you celebrate Thanksgiving, hope you’re enjoying your holiday. And in this video, I want to talk about moving averages and why I think you should ditch them, which is a bit of a big fight to pick, because if there’s one thing I can guarantee, you’re probably using a moving average on your favorite chart setup.
If not a moving average then a MACD, which is the difference between two moving averages. So, going against the grain here, kind of against the conventional wisdom, but just give me a few minutes and I’ll try and explain my point.
And why are moving averages so 1956? What’s the significance of the year 1956? 1956 was the year Robert Brown and his co-creator Charles Holt published a book on the exponential moving average, so it’s the first time it came into the public consciousness. They’d been using the moving average for 8 or 10 years previously.
They both worked for the U.S. Navy, although independently, and originally the exponential moving average was used to track the location of submarines. Later, it was used for demand forecasting and inventory control.
The beauty of the exponential moving average is that you only need two numbers to make the calculation. So here’s the little formula for an exponential moving average: it is yesterday’s value of the exponential moving average, or the previous bar’s value, times some factor, alpha, plus (1 – alpha) times the latest piece of data, whether that’s today’s price or the last bar’s price.
So as a little example here: if alpha was 0.8, you multiply the current value of the exponential moving average, say 100, by 0.8 then add (1 – 0.8), which is 0.2, times the current piece of data that comes along, which might be 110. Add those two numbers together, you’ve got 102. Easy mental arithmetic to do.
If you’re using a 10 or a 25 period simple moving average, you’d have to add together 10 or 25 different numbers and then divide by 25 or 10 or whatever it might be. With an exponential moving average, you only needed two numbers. And then in the days before the first pocket calculators were available, this was a godsend. Beautiful.
So 1956 was an important year. Prior to that, technical analysis has been very graphically-based. It’s all about trend lines and support resistance lines and things that you could do on a chart. After 1956, it was far more computationally-based, where we’re actually calculating moving averages and we’re using computer power basically to generate technical analysis signals.
Now, the problem with this is this increased computer power we’ve had ever since then has led us down the wrong path, in my opinion. The search for the ultimate moving average that smoothes out periods of consolidation.
Here’s a 4,500 tick bar chart of the Emini continuous contract day and night session going back about 7 or 8 days, and I’ve just got a simple moving average here place over this. It doesn’t matter on the length of this particular moving average, but if it’s falling, it’s colored in white. If it’s rising, it’s colored in red.
And you can see the trending moves quite nicely marked. Downtrend going here, uptrend going here, and so on. But, in between that, we have these periods of consolidation, where the market is trying to determine which way it’s going to go. And our moving average is all over the place. You can see here, it’s colored red and white and going backwards and forwards. Another
little bit of a trending move here: red and white going backwards and forwards and so on. And then through this consolidation phase, exactly the same thing.
During these periods of uncertainty or congestion, the moving average does not do very well at modeling. You don’t know whether the market’s going up or going down. A lot of false signals going on. So, all the activity that’s gone on in developing moving averages is all been about trying to find this ultimate moving average that smoothes out those periods of consolidation.
And the result is we literally have thousands of combinations of different moving averages. I’ve just, off the top of my head, listed some of the ones programmed in my EasyLanguage and TradeStation here. Obviously, we’ve got simple moving averages, exponential, weighted moving averages. We could also do medians of price values. There’s the T3, the TEMA, the Hull Moving Average. A quite nice little moving average. There’s the JMA, the Jurik Moving Average. Again, a very nice little moving average – very clever.
The whole of John Ehlers’ filters, and he’s done dozens and dozens of different filters, one of those is the Laguerre filter. Also the MAMA and FAMA moving averages. We’ve got linear regressions, zero lag filters, elliptics, and then multiple moving averages, and the difference between those multiple moving averages, and so on.
So, as you can see, there are dozens of different types of moving averages. Then, on top of that, you’ve got to choose the length, or the smoothing factor, for each of these to use. And then you could also be displacing these moving averages. So, put all of that together, and you’ve got a myriad of different choices of moving averages to use. And all they’re trying to do is smooth out these periods of consolidation and limit the number of false signals you get.
The problem that I have is that instead of trying to find this ultimate moving average, what we should be looking for is an indicator that tells us what kind of activity the market is going through. If the market is trending, a moving average will be fantastic at showing you signals in a trending market. But when a market is not trending, but is consolidating, the moving average will be horrible.
So you need an indicator that will tell you which kind of activity the market is going through at the moment, so that you can apply the right piece of analysis. In effect, what you’re calculating is the support and resistance levels adaptively. Because once you break out of a support or resistance level, that’s when you move into a trending period. John Ehlers to the rescue.
Beautiful piece of code he developed called the Hilbert Sine Wave, which I’ve improved on, called the Better Sine Wave, which gives you an alternate view of it. So, going back to our 4,500 tick bar chart here, and let’s look at this in terms of the Better Sine Wave indicator. Exactly the same period of time and data.
So, here we go, this is the Hilbert Sine Wave view of the world. And what it says is we’re in a downtrend period here, and we went into a pullback kind of trend where this downtrend actually came to an end. After the end of trend signal, we go through a period of consolidation.
During the period of consolidation here you can see the support, these dynamic support and resistance levels being plotted throughout this entire period here. And then the market breaks out into another trend move when it breaks a resistance or a support level, and pull back to end of trend, this tells us this is a trending period here.
After we get an end of trend signal, we’re going to go through some consolidation. And again, adaptive support and resistance levels being plotted on this data until we break out into another trend here. Pull back to end of trend. The market’s so strong. This was on Friday. The market just keeps on going there. Breaks out into another up trend, so it’s still in a trending phase here.
The Better Sine Wave indicator is showing you periods of consolidation, and periods of trending. And so what you need to do is to figure out what the market is doing and to play it accordingly. If the market is going through a consolidation phase here, having been a strong trend, you need to be playing the support and resistance levels. And then waiting for the next breakout into a trend, which happens to be an up trend here, above this level, and you can ride along the trend.
Now the beauty of using this Better Sine Wave, or this Hilbert Sine Wave approach is that, first of all, there are no inputs to tweak. You just load this onto a chart and it dynamically calculates these support and resistance levels. There are no inputs. Nothing to optimize.
The second thing is, because these levels are dynamically calculated, it’s an improvement on the traditional support and resistance view of the world, where the support and resistance levels were actually fixed values, when people were charting the market. This is calculating these and you can see these move through a fairly wide kind of range, and that the support and resistance levels are not fixed.
So, that’s my pitch for the Better Sine Wave indicator and why you should be ditching moving averages. The moving averages will always be giving you false signals through all these consolidation periods. Great during the trending periods. Lousy during consolidation. Whereas your dynamic Better Sine Wave or Hilbert Sine Wave type view of the world will actually tell you when you’re in one of these congestion zones and when you break out of those congestion zones into trend moves.
Now some of you who are familiar with the older versions of the Better Sine Wave indicator might be thinking: “Hold on, wasn’t there a Jurik Moving Average, a JMA version of the Better Sine Wave? Weren’t you using a moving average then?”
It’s true, yes there was a JMA version, but here’s a little twist and here’s how I was using that particular moving average. I’ve subsequently replaced the JMA with my own version of a fast kind of moving average. But the moving averages were being used to smooth the incoming price data or “pre-process” it in order to eliminate the noise.
You can use that technique to pre-process noisy price data into your other indicators. The difference is that all you’re using is a very short window, a short period for those moving averages. No more than 5 bars and something between 2 and 4 bars is ideal.
In fact, one of the simplest ways to use something like a moving average to pre-process, or smooth, data is to use a 3-bar or a 5-bar median of the price data coming in, in order to get rid of that noise. So, there was a moving average version that I was using, but I was using it in a totally different way. Very short term in order to pre-process data and get out the noise from some of the noisy price data streams that you get.
So, there we go, just a quick video explaining my pitch for why I think you should ditch moving averages. I get lots of charts sent to me by email by traders. If you still continue to use moving averages, I’m not going to hold it against you, but just consider there is an alternate way, and this search for the ultimate moving average I just think – it’s in the wrong direction. We need to be using approaches that are a lot smarter to understand the psychology of the market and try and figure out when we’re trying to break out of support and resistance levels.