Wednesday, February 27, 2013

How much is the rebound of S&P 500 in a drop ?

After analysing 32 years daily data of S&P 500, I use Excel to find the rebound ratio of the maximum drop in 5, 10, 15 and 20 days.

First, I look for the the highest point in different period. After finding the high, look for the minimum lowest point in that period. Finally, look for the amount of rebound before another drop. The x axis of the following 4 charts is the drop ratio compared to the price at High while the y axis of the charts is the rebound ratio compared to the price at Low.

The results are very consistent, the longer the period, the less the rebound ratio is.

They are :
  • 5 days : 0.6315
  • 10 days : 0.5512
  • 15 days : 0.5072
  • 20 days : 0.4903
According to this study, if there is a drop in the past 5 days and look for a rebound, the rebound should be around 0.6315 times of the drop.





 
 
On the other hand, the average amount of drop in these 4 periods are respectively,
  • 5 days : 2.64%
  • 10 days : 3.39%
  • 15 days : 3.94%
  • 20 days : 4.40%


Thursday, February 21, 2013

Can Harmonic Patterns give you profit in Forex ?

Common Harmonic Patterns include Gartley, Butterfly, Crab and Bat. Some people think that they are reliable and may yield very good risk to reward ratio. However, in my study, I found that these patterns are not as reliable and profitable as many people think.

I downloaded two indicators from the Internet and run the tests on MT4. They are ZUP and korHarmonics. Because MT4 does not run so well in auto back testing, I manually record the signals and calculate the targets and then check whether the price can hit the target or cut by the stop loss. What I found was very disappointing result. Although some people say that they should be used together with other signals for confirmation, I have another thought that a reliable method should be accurate itself. Therefore, I only test the Harmonic Patterns alone.

I set six targets, i.e. T1, T2, T3, T4, T5 and T6, respectively representing 0.382, 0.618, 0.764, 1, 1.272, 1.618 of the PRZ. Each trade is supposed to have 2 lots, I test combinations of the targets. The test run from Oct 2010 to July 2012. Below I only show some equity charts of the targets.






 
 
Gartley pattern is always thought to be one of the best among all these harmonic patterns. Conversely, its performance is the worst among all 4 pattens.
 
Butterfly seems to have quite good result. In the last chart, I reverse the buy and sell direction of Gartley and combine it with Butterfly to see the equity performance. Once the price reaches the PRZ, the target is set to another side of the PRZ and the stop loss is set to 0.382 of the last movement. For the butterfly patterns, I use different targets. 

The charts show the results of different time frames based on T2 and T6 targets, the time frames include 15 min, 30 min, 1 hr, 4 hr and 1 day, the shorter the time frame, more number of trades can be set up. I stop running 15 min earlier because the profit in fact cannot cover the transaction costs involved and it generates too much signals to buy and sell. Overall speaking, the performance of harmonic pattern is not so stunning. The best pattern Butterfly has also a long period of small gain, that is about 1 year in testing.

Are Harmonic Patterns profitable ? At least, I won't use Gartley in Forex after doing the test. Crab is also not reliable enough. The fluctation of Bat is quite big. Only the Butterly is a possible pattern that can bring in profit but it may have long term flat performance. Since all these tests are run on Forex data, my conclusion may be wrong for stock trading.

Tuesday, February 19, 2013

Backtesting of S&P 500 Trading Strategy 1

The following S&P 500 chart is obtained from the back testing result of one of my trading strategies recently developed, it has very good performance in terms of profit and maximum drawdown when compared to S&P 500 index. The data are obtained from FXCM. The test was run from 1/1/1981 to 20/2/2013. The total profit is nearly 5900 points if not considering transaction cost. Even with the high transaction cost of FXCM and interest fee considered, the profit is still more than 3000 points for the whole period. If using S&P 500 data downloaded from Yahoo, similar performance is obtained.
 
 

I'm still looking for more strategies for stock trading and I'll public some of the results later.