PDA

View Full Version : probabilities for the combined Long term / Short term settings



TraderD
12-02-2011, 08:39 AM
Pascal,

You've shown elsewhere that (on average) returns deteriorate when delaying execution of a trade on a buy/sell signal by X number of days. The robot page displays the expected 3D/10D returns due to LT/ST settings. My question is - do you indeed use the "execution delay" as a dependent variable in calculating the LT/ST stats shown on the robot page?

Trader D

Pascal
12-02-2011, 09:18 AM
Pascal,

You've shown elsewhere that (on average) returns deteriorate when delaying execution of a trade on a buy/sell signal by X number of days. The robot page displays the expected 3D/10D returns due to LT/ST settings. My question is - do you indeed use the "execution delay" as a dependent variable in calculating the LT/ST stats shown on the robot page?

Trader D

The LT/ST stats (% of gain and probablilities of gain of a LT/ST combination) are calculated a posteriori. These stats themselves are not used by the robots. The robots use their different indicators, combine them in to LT and ST edges and then take an investment decision. One of the indicators is the delay since the last signal. Hence, the "execution delay" is used to calculate the LT/ST levels, but not to recalsulate teh stats linked to such levels.


Pascal

TraderD
12-02-2011, 09:49 AM
The LT/ST stats (% of gain and probablilities of gain of a LT/ST combination) are calculated a posteriori. These stats themselves are not used by the robots. The robots use their different indicators, combine them in to LT and ST edges and then take an investment decision. One of the indicators is the delay since the last signal. Hence, the "execution delay" is used to calculate the LT/ST levels, but not to recalsulate teh stats linked to such levels.
Pascal

Thanks for clarifying. If I'm not mistaken, there is also an optimization phase for the robot operation. Would it make sense at all to segment the LT/ST levels' stats (a posteriori) by execution delay and somehow improve the LT/ST levels/edges?

Trader D

Pascal
12-02-2011, 10:02 AM
Thanks for clarifying. If I'm not mistaken, there is also an optimization phase for the robot operation. Would it make sense at all to segment the LT/ST levels' stats (a posteriori) by execution delay and somehow improve the LT/ST levels/edges?

Trader D

I don't know. Something to test, but there might be some more urgent things to do...


Pascal