Trading System for Market Timing, Results

by billb 12. September 2007 11:51
The results are in. The data is from Yahoo. The system is done with RightEdge and I posted the trading system on the RightEdge web site for download here.
 
Here are the inputs:
IXIC = Nasdaq Composite. Data available was from 1972 to present
GSPC = S&P 500. Data available was from 1960 to present.
DJIA = Dow Jones Industrial Average. Data available was from 1960 to present.
 
 

 

Symbol

MA Parameter

APR

Max Drawdown %

Trades

Buy & Hold APR

Buy & Hold ($)

APR Diff

Net Profit ($)

Net Profit Diff

IXIC

50

12.47%

33.15%

220

9.27%

$ 2,565,700.00

3.20%

$ 7,297,262.31

$ 4,731,562.31

IXIC

100

9.90%

60.07%

144

9.27%

$ 2,565,700.00

0.63%

$ 3,071,949.41

$ 506,249.41

IXIC

150

10.03%

63.61%

111

9.27%

$ 2,565,700.00

0.76%

$ 3,206,736.20

$ 641,036.20

IXIC

200

9.94%

65.28%

93

9.27%

$ 2,565,700.00

0.67%

$ 3,112,338.98

$ 546,638.98

IXIC

250

9.36%

66.70%

69

9.27%

$ 2,565,700.00

0.09%

$ 2,542,909.04

$ (22,790.96)

 

 

 

 

 

 

 

 

 

 

GSPC

50

6.04%

33.04%

362

6.91%

$ 2,425,985.16

-0.87%

$ 1,540,305.64

$ (885,679.52)

GSPC

100

5.40%

50.64%

248

6.91%

$ 2,425,985.16

-1.51%

$ 1,127,345.27

$ (1,298,639.89)

GSPC

150

6.63%

38.20%

183

6.91%

$ 2,425,985.16

-0.28%

$ 2,034,375.83

$ (391,609.33)

GSPC

200

6.85%

28.27%

143

6.91%

$ 2,425,985.16

-0.06%

$ 2,263,890.37

$ (162,094.79)

GSPC

250

7.27%

21.61%

112

6.91%

$ 2,425,985.16

0.36%

$ 2,742,239.76

$ 316,254.60

 

 

 

 

 

 

 

 

 

 

DJIA

50

5.21%

33.04%

390

6.40%

$ 1,927,800.94

-1.19%

$ 1,027,406.97

$ (900,393.97)

DJIA

100

5.11%

50.64%

277

6.40%

$ 1,927,800.94

-1.29%

$ 976,990.93

$ (950,810.01)

DJIA

150

5.48%

38.20%

200

6.40%

$ 1,927,800.94

-0.92%

$ 1,173,244.20

$ (754,556.74)

DJIA

200

5.86%

28.27%

168

6.40%

$ 1,927,800.94

-0.54%

$ 1,414,448.05

$ (513,352.89)

DJIA

250

5.68%

21.61%

145

6.40%

$ 1,927,800.94

-0.72%

$ 1,297,957.03

$ (629,843.91)

 
The MA Parameter column is the moving average that the trading system was run against. I also included the number of trades taken for each run because commission costs vary and will obviously impact performance.
 
Let me make a few points about the numbers. First off, more often than not timing your entries and exits using moving averages reduces your overall profits in the long run. Second, I've included the drawdown figure, which doesn't necessarily conclude that being out of the market more means less volatility.
 
Another thing that the system cannot account for accurately without significant are dividends.   If you're not in the market, you can't collect the dividend. So how much additional did you lose by being out of the market and missing dividends?
 
Looking at this data as a trading system developer would indicate that there is no edge and potentially a negative edge to using a moving average to time the major indexes. Based on my results, I'm not in agreement with the other findings that timing with moving averages is a good idea for your long term investments.

Tags:

Trading Systems | Markets

Comments

9/11/2007 12:06:51 PM #

Roger Nusbaum

This can be problematic, I agree. In trying to use the 200 DMA I have been faked out several times in the last few years and we will see if the current dip will result in another fake out.

It is for this reason that I don't make broad sweeping changes right away.

Great to see you have a blog.

Roger Nusbaum

9/11/2007 5:29:30 PM #

Michael

I complement you for actually testing the system for yourself rather than just accepting conventional wisdom.  Like you, I've also done backtest on simple MA timing and gotten similar results.

I agree with your conclusion that this is not particularly better than buy & hold.  But let me suggest an alternative way to think about it.
1. Results are comparable magnitude.
2. Risk is comparable magnitude.
3. Correlation is not too high, and fairly low when buy & hold gets killed.

I have a portion of my portfolio in B&H, and portion using SMA timing.  The overall Sharpe ratio on backtest using the combined approach looks great.  It's not right for everyone, and has some problems (trading cost, taxes, etc.) but can make sense for some.

Also, since I have a big concern about current market valuation, it's a way to mechanically be in the market if it goes up, and know that if down in big way over next couple of years I should miss most of the downside.

Michael

Michael

9/11/2007 11:34:38 PM #

Bill B

Thank you for your comments, I was sure my way wasn't the only way to think about it :)

Michael, you do get whipsawed a bit when the market is hovering around the MA line.  In fact, the overall win % on a trade basis is low (25-30%), but of course, the winner $ is much larger than the loser $.

You make valid points across the board.  I haven't necessarily ruled this out of my life, but the rosy picture that has been painted by some doesn't seem all that rosy to me.

Bill B

9/12/2007 12:51:28 PM #

ScioTrader

How about testing S&P500 back to 1900? Someone did it:
http://ssrn.com/abstract=962461

ScioTrader

9/13/2007 6:02:00 AM #

Bill B

If I can find the data, I'd be glad to.

Bill B

Add comment


(Will show your Gravatar icon)

  Country flag


  • Comment
  • Preview
Loading



Powered by BlogEngine.NET 1.5.0.7
Theme by Mads Kristensen