MarketTrak Question/Comment Message


Posted By: david   Date: Sun Apr 18, 2010

Title: re the cash only rule after short

Message:
  You have a rule that the signal only can go to cash, 1-day after a previous signal in the opposite direction. Now, for example, we would be long, other than the fact that there was just a short from Thursday's close into Friday.
 
  In reviewing the ANO and position values from jan 08 to dec 09, for 25.1, I see at least 3 days on which being one was in "cash" in a similar situation. the ano was greater than .2; one would have been in cash, but the day before, one was short. The first 3 days I looked at would all have been a gain if one had been long on those days, and Oct 13, 08 is worth nearly 1000 points.
 
  Does this rule actually save "us" or the system money?
 
  thanks!
 
  dz

Response:
 The optimization program that determines the trading setpoints and rules uses the last 10 years of market data to get good statistical results. There must have been some trades during this time when going through a cash position made sense. The program accounts for trading profit and loss of capital. It uses a variation of the Sharpe Ratio.

 Since we modified the network weighting algorithm, I believe that the model is now more accurate and perhaps the requirement of going through a cash position on a trade will go away. I would expect that the setpoints would also change. I will look into rerunning the program. Of course, you can decide to ignore this rule in your trading. I did take a small long position in the after-hours session on Friday.

Wednesday, April 21 Update:  I ran a test case where I allowed trades to proceed directly from long to short and from short to long. In the last 165 days, there was only a single incidence when this happened and that was a few days ago. The setpoints were not modified. The performance results are below:

Forecast Model Performance (standard trading model)
 
 Y-T-D
 165 Days
45 Days
 Buy and hold return
  6.37 percent
  18.63 percent
  9.34 percent
 Forecast model return
  4.44 percent
  18.63 percent
  8.66 percent
 Max model drawdown
  4.27 percent
  4.27 percent
  0.97 percent
 Total stops
 1
 1
 0
 Days long
 39
 122
 20
 Days cash
 33
 42
 24
 Days short
 1
 1
 1
 Trading days
 73
 165
 45

 

Forecast Model Performance (direct trades between long and short allowed)
 
 Y-T-D
 165 Days
45 Days
 Buy and hold return
  6.37 percent
  18.63 percent
  9.34 percent
 Forecast model return
  5.14 percent
  19.41 percent
  9.38 percent
 Max model drawdown
  4.27 percent
  4.27 percent
  0.97 percent
 Total stops
 1
 1
 0
 Days long
 40
 123
 21
 Days cash
 32
 41
 23
 Days short
 1
 1
 1
 Trading days
 73
 165
 45
 

 As you can see, there is a small difference in favor of allowing direct trades between long and short positions. I will probably go ahead and implement this change after some more testing.

Friday, April 23 Update:  I will install an update this weekend to remove the requirement of going to cash when leaving either a long or short position. The long and short setpoints will not change.

Return to message board




Bag Toss: high quality, made in America, and makes a great gift.
 

Copyright © 1996-2017 MarketTrak LLC.  All Rights Reserved.