Forums  > Pricing & Modelling  > Setting limits given probability distribution  
     
Page 1 of 1
Display using:  

secsy


Total Posts: 10
Joined: Mar 2019
 
Posted: 2019-03-21 12:41
Hey all, I am new here. I have lurked for a while but decided to create an account and start contributing to the community. I have been working on the side on algorithmic trading for the past 7 months and actually have started to see some good results as of late.

I am wondering what is recommended for setting take profit and stop losses if my model generates an expected probability distribution for the price change of an asset over the next N minutes. Right now I am using a stop loss based on the standard deviation of the distribution, and the take profit based on the mean (I only would trade if the mean of the distribution is significantly away from 0). I am familiar with Kelly Criterion although I am not sure if I could apply that here.

I am also thinking in basic terms of a risk reward ratio based on the distribution, but for some reason it is seeming complicated and I am not sure what to do.

Any thoughts would be appreciated. I look forward to contributing here from now on.

gaj


Total Posts: 49
Joined: Apr 2018
 
Posted: 2019-03-21 14:08
If you take profit at the mean, your profit is capped at that level. So your realized profit will be lower than your model's mean.

Instead of using stop loss/take profit, why don't you get out after N minutes, which is what your model is predicting anyway. If slippage is an issue, you can work it more slowly.

You can also look at how much of the alpha is realized within that N minute interval. If for example 50% of the alpha is realized in 1min, 85% in 2min, 99% in 3min, 100% in 4min and levels off beyond that, you should probably get out in 3 minutes.

secsy


Total Posts: 10
Joined: Mar 2019
 
Posted: 2019-03-21 16:04
Okay gotcha that makes a lot of sense. Thanks man
Previous Thread :: Next Thread 
Page 1 of 1