Relative Strength - Internal

[Do Or Die 10]

How is is that you connected the highs on the divergences, but you connected all the other lows on the previous up trend? I know that you connect lows for up trend and highs for down trend, but if you connect the peaks of the up trend just like the divergence, then the price constantly rises while the RSI falls.

 

Thats a good detail question.

 

Uptrend is always marked by connecting lows; downtrend is always marked by connecting highs... you can take it as 'first law of chart reading"

 

If you are connecting peaks on the chart, you have to connect them on the oscillator as well. You cannot connect peaks on the chart and compare against lows on the oscillator.

 

Hope this in reference to my previous post clarifies.

[umfan92 10]

Ok so in the case of the divergence: the price increased (the second peak is obviously higher) yet the indicator (in this case the RSI) decreased. So, since you drew a line connecting the highs, are you saying that when looking for divergences, the indicator is priority?

 

What I mean is, normally ignoring indicators, if I were to draw a trend line of the price at the point of those highs, I would draw it at the lows rather than then highs. But in this case you drew the lines on the highs (up trend) so the RSI was the focus of the trend lines.

 

I hope I explained myself clearly. Sorry if it's a little hard to follow.

[Do Or Die 10]

What I mean is, normally ignoring indicators, if I were to draw a trend line of the price at the point of those highs, I would draw it at the lows rather than then highs. But in this case you drew the lines on the highs (up trend) so the RSI was the focus of the trend lines.

 

I hope I explained myself clearly. Sorry if it's a little hard to follow.

 

Yes right, for marking bearish divergence we need to connect highs, no exception.

 

The fact that bearish divergence mostly occur in a uptrend is another matter.

[umfan92 10]

Ok that helps. That makes much more sense. I'm glad I asked that and thank you for the quick and thorough response.

[Do Or Die 10]

Another attempt to measure internal RS:

 

The Choppiness Index was mentioned by Bill Dreiss in early 90's. The CI calculates the relative position between the Sum of daily trading ranges during a given period of time against the total range for that period. Bill Dreiss says "high CI readings can be used to indicate that a consolidation is about to end and a position should be entered or a breakout anticipated." Similarly, low readings in the CI correspond closely with the end of strong impulsive movements either up OR down, while High readings occur after significant consolidations in the price. Extended periods of trendless price movement are reflected in extended periods of above-average readings of the CI.

 

It's a little more complicated was to measure internal RS and seemingly less effective (see attached chart).

 

[Do Or Die 10]

The Amibroker code for choppiness index is:

range = Param("Periods", 14, 5, 252, 1 );
MaxH = Max(H, Ref(C, -1));
MinL = Min(L, Ref(C, -1));
HMax = HHV(MaxH, range);
LMax = LLV(MinL, range);
TR = ATR(range);
SumTR = (Sum(TR,range)) / (HMax - LMax);
logsum = log10(SumTR);
LOGn = log10(range);
Choppiness = 100 * logsum / LOGn;
Plot (Choppiness, "Choppiness Index", colorBlack, styleLine);

 

 

For the sake of reference, there is another indicator called b-indicator.

 

Amibroker code:

function PiecewiseEMA( array, range, sincebar )
{
 factor = IIf( sincebar, 1, 2/(range+1) );
 return AMA2( array, factor, 1-factor );
//ama2( ARRAY, SMOOTHINGFACTOR, FEEDBACKFACTOR )
}
// parameters
smoothing=Param("smooth",4,0,100,1);
//m=4;
n=250;
// generate reversal signals based on EMA crossover rule
Lpf1 = EMA( C, 7  );
Lpf2 = EMA( C, 15 );
CrossPoint = Cross( Lpf1, Lpf2 ) OR Cross( Lpf2, Lpf1 );
Periods = BarsSince( CrossPoint );
// variable bar sum
DC = Close - Ref( Close, -1 );
CPC = Sum( DC, Periods );
// smooth CPC by piecewise 4 bar EMA
Trend = PiecewiseEMA( CPC, smoothing , CrossPoint );
// noise
DT = CPC - Trend;
Noise = 2 * sqrt( MA( DT^2, n ) );
// alternative 'linear' noise calculation
// Noise = 2 * MA( abs( DT ), n ) );
BIndicator = (abs(Trend)/(abs (Noise)+abs(Trend))) * 100;
Plot( Bindicator, "Bindicator", colorBrightGreen, 1);
PlotGrid( 0 );
PlotGrid( 20 );
PlotGrid( 40 );
PlotGrid( 60 );
PlotGrid( 80);
PlotGrid(100);
//--Tomasz Janeczko, AmiBroker.com
GraphXSpace = 3;

[umfan92 10]

Hey Do or Die,

 

I've been practicing using the RSI now for a while and I believe I'm getting good at it. I would just like to ask you: How reliable is it? Would you base an entire trade on it?

 

When I look for a company to trade, I look at their chart first, then the moving averages, then RSI then I look at the earning and fundamentals (although I know for short term technical analysis it doesn't matter what the fundamentals are but I still like to know that they're doing good).

 

Specifically, I noticed that Ford Motor Company (F) has had strong support at around 9.90 and has been forming a bullish RSI divergence since August 8th. It seems to be a perfect textbook example of a bullish RSI divergence. What is your opinion?

[Do Or Die 10]

Hey Do or Die,

 

I've been practicing using the RSI now for a while and I believe I'm getting good at it. I would just like to ask you: How reliable is it? Would you base an entire trade on it?

It can help you time the market with great precision, see here a live example.

When I look for a company to trade, I look at their chart first, then the moving averages, then RSI then I look at the earning and fundamentals (although I know for short term technical analysis it doesn't matter what the fundamentals are but I still like to know that they're doing good).

 

Specifically, I noticed that Ford Motor Company (F) has had strong support at around 9.90 and has been forming a bullish RSI divergence since August 8th. It seems to be a perfect textbook example of a bullish RSI divergence. What is your opinion?

 

Yes it is a good approach. F is definitely a great pick, actually it looks too cheesy to work :p. For investing with horizon of >6 months, half can be bought here with stoploss 9.80.

 

Why half size? for trading on longer time frames, scaling-in works better than stop loss and scaling out works better than profit targets.

 

[umfan92 10]

Thanks DD! I told you I've been practicing! I'm glad it's paying off and I seem to be learning properly.

 

Do you think you could explain "scaling in" and "scaling out" because I don't really know what that means? Thanks!

[Do Or Die 10]

Scaling-in or continuity trades means buying in parts at different prices rather than buying all at one price level. For example, you buy half shares at current price and half shares (anticipatory) near 6.00 which is historically important support. Or, 40% shares at current price, 30% at 8.0 and 30% at 6.0. You can still maintain a hard-stop of 5.0 (say) for all of your position. With scaling-in, If the stock is dragged down by overall weak market, you are not stopped out. Even if it flies off from here you still have nearly half shares so it means good profit.

 

Similarly scaling out means selling your shares in parts at different price levels rather than selling all at stoploss or profit target. For example, suppose your stock hits the profit target and you book profit on all shares. Sometimes the stock will continue to run upwards and you miss part of profits. An example of scaling out will be to use profit target for half of position and use trailing stop for other half of position.

 

If you plan to hold a trade for more than 100 bars (say) this soothes the equity curve to a good extent. You can backtest on what difference it makes. This is one of the simplest approach to trade management, and you should be able to learn quickly. A more 'advanced approach' is to buy/sell each time the market gives you information about upcoming moves, discussed as the 'combination approach' in this article: http://www.traderslaboratory.com/forums/psychology/10512-sin-predicting-anticipatory-trading.html

[umfan92 10]

Interesting. I have never heard of that before but I definitely think I'm going to do that with Ford. That sounds perfect. I don't want to miss out on the potential rally if the divergence is correct, so I could buy half the shares now, and if it rallies, then I won't miss out entirely. If it falls to meet my original buy price, I would just average down and let it continue as planned. That's perfect! I will continue reading about scaling in or out, but thank you for letting me know.

[Do Or Die 10]

As a counter-view F has been in a downtrend. You can see that it has been coming down since last 6 months or so. It's RS peaked out in April and shows no improvement.

 

You might have remembered AOL, we discussed here and here. It was similar to F and broke down. I had the view that AOL should only be bought after breakout from range which it never did. (BTW the other two 'survivor' stocks after crash AAPL and KO have gone up nicely)

 

The point I want to make is divergence signal a reversal and as such more often initiate a trade against current trend. Hence it is good to look for some confirmation. Buy above 11.00 (upper part of range, near breakout), not immediately.

Edited September 19, 2011 by Do Or Die
attachment