Technical Analysis Predicting trend changes: The Forecast Indicator

Moving averages and their crossovers or crossovers with indicators are widely used in trading and
technical analysis. Shorter averages above longer averages or closing prices above them suggest an uptrend. An indicator below its moving average is a sign of a downtrend. It would be beneficial for a
trader to develop a technique to predict the closing price which will generate a crossover of averages as described above. This article introduces such an anticipation technique, based on moving averages, to determine the future closing price for which a
trend change is possible.

valcu-1.gif


Let's look at figure 1 representing the German
DAX index. Subchart 1 contains two simple moving averages: 5- and 10-day, SMA5 and SMA10. It is well-known among traders that SMA5 > SMA10 corresponds to a short-term uptrend, while SMA5 < SMA10 suggests a short-term downtrend. At point (A) on 18/8/2004, DAX was in a
downtrend as defined by the above-mentioned criterion. That day, an inquisitive trader could have asked: "What will the
close at the end of the next trading day be, in order to have SMA5 above SMA10, and a change of the short-term trend? Can we predict next day's close to meet this requirement"? The same question can be asked in (B) with DAX in a short
uptrend: "How should DAX close tomorrow, 13/10/2004, to trigger SMA5 below SMA10"?

The whole idea with this technique is to be prepared for a trend change in a
risk-controlled manner.

The technique we are going to describe will take into account simple moving averages and two common scenarios found on any chart:
  1. Crossovers of SMA(C,5) and SMA(C,10)
  2. Crossovers of Close and SMA(C,10)

It can be applied to other averages or to indicators and their moving averages.

5- and 10-day golden and dead crosses
Figure 2 shows Vodafone during November 2003 and May 2004. Between (A) and (B) the stock was in a downtrend defined by SMA(C,5) < SMA(C,10). The shorts had a good time, but the longs were impatient to take control and enter the market. They knew that a
golden cross defined by SMA(C,5) breaking above SMA(C,10) would be an entry signal, but they did not know when it would occur.

valcu-2.gif


In retrospect, that point in time was (B) with the following condition met:

SMA(C,5) > SMA(C,10) (condition 1)

Let's expand both moving averages:

SMA(C,5) = (C+C(-1)+C(-2)+C(-3)+C(-4))/5
(the sum of the past five closing prices, including the current one, divided by five)

SMA(C,10) = (C+C(-1)+C(-2)+C(-3)+C(-4)+C(-5)+C(-6)+C(-7)+C(-
8)+C(-9))/10
(the sum of the past ten closing prices, including the current one, divided by 10)

After simple algebraic operations and extracting C, condition 1 is reduced to
C > (C(-5)+C(-6)+C(-7)+C(-8)+C(-9)) - (C(-1)+C(-2)+C(-3)+C(-4))

With this anticipation technique in mind, C is in fact the close tomorrow C(+1) which makes condition 1 true. In this case the above formula translates into

C(+1) > (C(-4)+C(-5)+C(-6)+C(-7)+C(-8)) - (C+C(-1)+C(-2)+C(-3))

How can we translate it into plain language?
  • In a downtrend and to change it, the close of the next bar, C(+1), must be greater than the sum (C(-4)+C(-5)+C(-6)+C(-7)+C(-8)) - (C+C(-1)+C(-2)+C(-3))
  • In an uptrend and to change it, the close of the next bar, C(+1), must be less than the sum (C(-4)+C(-5)+C(-6)+C(-7)+C(-8)) - (C+C(-1)+C(-2)+C(-3))

The sidebar contains the calculation of a
Metastock indicator, X-SMA5-SMA10, which computes and displays C(+1) for the next bar (in our case, for the next trading day).

Case study - Vodafone
We consider the Vodafone chart again and discuss the technique during period (A)-(B). The downtrend starts at (A) on January 21, 2004. That day, SMA(C,5) goes below SMA(C,10). The shorts are in charge as long as the two averages remain in this sequence. Subchart 1 displays X-SMA5-SMA10 indicator which calculates the next bar's close in order to change the current trend. During the downtrend, we reach January 29 and the longs are asking: "At or above which value should the close be on the next trading day, January 30, to have SMA5 above SMA10"? To find the answer traders consult indicator X-SMA5-SMA10 and get a reading of 30.60. One day later, Vodafone closes at 25.60 which is far from 30.60, so the longs must postpone their bullish action.
Markets go in cycles so we get closer to a bottom. On February 11 Vodafone is still in a downtrend and the longs are asking the same question. They refer to the indicator in sub-chart 1 and the answer for the next trading day, February 12, is 23.92. The bulls feel that this price can be reached. On the 12th the close was 24.91 which made SMA5 go above SMA10 and signalled a trend change. The bulls were happy... and on that day they should have had two options: (1) to enter half an hour before the market's close if the price was >23.92 or (2) to defer their entry for February 13 at the open if the price closed >23.92. The first choice would have been better.

The segment (B)-(C) was an uptrend with SMA5 > SMA10. Shorts were contemplating an entry but looking at X-SMA5-SMA10 in subchart1, they saw that the right moment was not there. During the uptrend, they noticed a bottom of the indicator in B1 and they realised that the trend was going to change shortly. On February 26 the close was 25.60 and the required close value for February 27 was less than 25.90 in order to change the trend. The close on the 27th was 24.97; everything was going as the shorts wanted and SMA5 went below SMA10.

The closing price and its 10-day average
The anticipation technique described above can be also applied for the close and its 10-day simple moving average. Figure 3 shows Deutsche Bank during June and December 2002. Between (A) and (B) the stock was in an uptrend defined by C > SMA(C,10). The longs were in charge, but the shorts were ready to take again advantage of the overall bearish market. That moment would have been with Deutsche Bank closing below its SMA(C,10).

valcu-3.gif




Looking in retrospect, that point in time was C with C C SMA(C,10) (condition 2) met.

Now let's expand the right term of condition 2. SMA(C,10) = (C+C(-1)+C(-2)+C(-3)+C(-4)+C(-5)+C(-6)+C(-7)+C(-8)+C(-9))/10 (sum of the past ten bars, including the current one, divided by 10)

After simple algebraic operations and extracting C, condition 2 is reduced to
C < (C(-1)+C(-2)+C(-3)+C(-4)+C(-5)+C(-6)+C(-7)+C(-8)+C(-9))/9

Shifting everything one day, C is in fact the close tomorrow C(+1) (or for the next bar) which makes condition 2 true. With this adjustment in mind, the above formula translates into:
C(+1) < (C+C(-1)+C(-2)+C(-3)+C(-4)+C(-5)+C(-6)+C(-7)+C(-8))/9

How can we translate this formula?
  • In a downtrend and to change it, the close next bar C(+1) must be greater than the sum (C+C(-1)+C(-2)+C(-3)+C(-4)+C(-5)+C(-6)+C(-7)+C(-8)) divided by 9
  • In an uptrend and to change it, the close next bar C(+1) must be less than the sum (C+C(-1)+C(-2)+C(-3)+C(-4)+C(-5)+C(-6)+C(-7)+C(-8)) divided by 9

The sidebar contains the calculation of a Metastock indicator, X-C-SMA10, which computes and displays C(+1) expected to change the shortterm trend defined by crossovers of close and its 10-day moving average. Table 2 outlines all steps taken during period (A)-(B) until a bearish crossover occurred in (C). It also shows where and how the
trader would have taken advantage of this technique.

Step-by-step
The best way to understand this technique is to outline a roadmap and see how it works for a sequence of trading days.

valcu-4.gif




Step 1: Use your technical analysis software to define the two indicators described here, X-SMA5-SMA10 and X-C-SMA10

Step2: Build a screen containing [at least] the following views:
  • View 1: X-SMA5-SMA10 or X-C-SMA10
  • View 2: Price displayed as a line graph, SMA5, and SMA10
  • View 3: (optional) another indicator, preferably Bollinger bands, Z-score,
    ADX, or any other of your choice. This will be used as a second opinion to avoid possible whipsaws due to the short-term periods for the averages

Step 3: Define your next action, either enter long (exit short) or exit long (enter short), based on the positions of the pairs (SMA5, SMA10) or (C, SMA10).

Step 4: In a downtrend defined as SMA5 < SMA10, check daily the forecast indicator displayed in view 1. If it approaches a top, the price may soon be close to a bottom and you should consider preparing a long entry at a value which is greater than the one shown by the indicator. Consider the initial stop-loss as the recent price low. Once in a long position, implement a trailing stop equal to C(+1) or based on volatility (ATR).

Step 5: In an uptrend defined SMA5 > SMA10, check daily the forecast indicator displayed in view 1. If it approaches a bottom, the price may soon be close to a top and you should consider preparing for a long exit at a value which is less than the one shown by the indicator.

To illustrate all these steps, let's create an action table during the downtrend (A)-(B) for Vodafone. We will monitor on a daily basis the anticipated close for the next trading day, C(+1) which may change the trend from descending to ascending.

How could we have taken advantage of the anticipation signal on 11/02/04? During the last half hour of trading, we were facing a decision about whether to enter long positions or to delay the entry for the next day. With an update of the indicator X-SMA5-SMA10 during the last 30 minutes of trade, C(+1) would have read a value greater than 23.69 for the next day. Comparing this value with the pre-close of around 24.89, the chance for a trend change would have been excellent and we could have been in anticipation of the crossover. Another entry would have been at the open of 12/2/04. The final confirmation came at the end of 12/2/04. Sometimes when signals are strong it is a good idea to participate with small positions (end of 11/2/04) until a full confirmation (end of 12/2/04) occurs.

Now, let's apply the same procedure for Deutsche Bank during the period 7/08/02 - 30/08/02.

valcu-5.gif


In the case of these crossovers, the situation is a little different. The lows and peaks of X-C-SMA10 do not anticipate tops and bottoms for the price. The days prior to the trend confirmation, 07/08 and 28/08, do not allow accumulation based on the probability that the price the next day would meet the targets indicated by C(+1). This scenario works best using confirmations at the end of days of trend reversals like (C).

Conclusion
This technique is simple and can be implemented by any trader who works with moving averages. It can be used in any timeframe, with various periods, and with all trading instruments. In the case of SMA5 and SMA10 crossovers, the forecast indicator displays very useful information by forecasting minor tops and bottoms for the price. A bottom, respectively top of the indicator announces a top, respectively a bottom of the price. The values computed by the forecast indicators can also be used as stop-loss or cover stop-loss.

Since moving averages are taken into account, there are delays for the crossovers. They can be reduced by using zero-lag averages, but in this case, the calculation of C(+1) may become more complex, yet more rewarding. Other areas to develop this technique for are
crossovers of indicators and their moving averages. An immediate example is the forecast of closing prices to generate
MACD crossovers.

If your trading platform allows for the building of custom indicators, then you can apply this technique in a real-time environment. Everything depends on some basic algebra and willingness to "beat the system".

valcu-6.gif
 
Last edited by a moderator:
This article introduces an anticipation technique based on moving averages to determine when a trend may change.
 
MA cant predict future trends or targets. check out latest traders mag. article there about pointlessness of inds.
 
Ma's do get slated as pointless and tests on mechanical systems that use them are in general not good so on the surface your statment would appear correct ,but unfortunately this equates to saying this pear is not a very good doorstop...in other words it is not that ma's are not useful it is that they are misused when applied mechanically (thoughtlessly)..you could generalise that and say most indicators applied mechanically (thoughtlessly) are likewise misused....ma's are certainly useful contextually in helping to anticipate in which timeframe the market will be shown to be wrong and at what point that will be transparent to other timeframes.
In my view it's all about anticipation ..confirmation and transparency summed up by Bernard Baruch who said something similar to this..."that what is known to other men is not worth knowing"
 
As chump says, MAs have to be understood in context, ie; only place value on them in a trending market, and treat them with caution in a ranging market. ( How you do that is another matter )
( the lack of context results in poor performance of MAs in tests )

As for "prediction", I wonder whether the need to "predict" is more a psychological need for some risk-averse type of traders.

MAs DO NOT PREDICT the direction of a trend, ( or anything else ), they provide re-assurance of the continuation of a trend, once it has started. ( since, once trending, they are regarded as dynamic support in an uptrend, and a dynamic resistance in a downtrend ).

I wonder what psychologists would make of peoples desire to chase "predictive indicators" ?
( as opposed to successful traders, who understand the nature of probabilities, and just plough through their plan, knowing the potential / downside of same )

Since MAs are lagging, they can only tell you a move is continuing or stalling.

You need something else to identify ( not predict ), a trend change. ( narrow range days, for example ), or Head and Shoulders. etc.
Even a MA catching up with price may indicate trend weakening. ( funnily enough, this also identifies a good re-entry to a trend. )
 
All trading is based on making intelligent choices on high probability winning trades. Therefore any technique that aids such a decision is bound to be useful rather than throwing darts in the air. The author's technique is one fo the few soild quantitative tools I have come across and I commend his contribution. It would be useful to have some statistics to test this technique versu different markets at various times. the DOW and Oil is both keen on reverals lately but the currency markets are not so. As with all mechanical indicators some discretion on the part of the brain is still necessary as there can be no Holy Grail.
 
Tee hee... What's Terry Wogan doing writing articles on moving averages?
 
Didnt like this whole MA article, they cant show a real direction of the stock, they are just history showers to tell you wether the stock is in pure uptrend or downtrend.
 
Tomerep said:
Didnt like this whole MA article, they cant show a real direction of the stock, they are just history showers to tell you wether the stock is in pure uptrend or downtrend.

hmmm yes, but if you extrapolate that trend based on its smoothness todate into the future n all that, and do things move in cycles, or waves? waves within waves, inner outer , channels within channels. Its man tinkering with a kind of time machine , hmmm . very cool........ your own interactive quest of time price into the future, I think Brian J Millard. did or does some work on that side, Sigma P probability analysis and extrapolating trends into the future, interesting maybe, but is it required.? exercise of stretching the grey matter maybe..
 
Tomerep said:
Didnt like this whole MA article, they cant show a real direction of the stock, they are just history showers to tell you wether the stock is in pure uptrend or downtrend.

You are right. They simply attempt to reassure that the share is going in the direction that you are thinking of trading. The longer the ma, IMO, the more inaccurate they are because the share will change direction well before the ma does. In any case, I don't trust any of them and I could show you some scars to prove it......

However, if you use, say, a 13 day once the trend is established the bar does, often, pierce it, signalling trade entering possibilities. However, once the trend turns watch that you do not go long when you should be short! Sooner or later, that will happen. The ma will predict nothing, it's all hindsight.


Split
 
Using 5 and 10 days moving average to predict a trend is a sure way of missing it.
It gives signals way too late, it is often confusing, and it signals squat.
Valcu posted previously articles on this method, which I think it is not only confusing but gives a lot of false signals.

Just a thought.
 
Top