Random Trading

Loki,

If you are simulating (not really trading) just count your entry/exit price as the bid/ask midpoint (or last sale price) and re-run your simulation.

If it evens out, then at least you know that bid/ask spread was the cause.

But in real life you can't just out-wait the bid-ask spread, hopefully you have some actual insight guiding your actions :)
 
Quote from ProfLogic:

If it consistently losses money, program it to trade in the opposite direction of the parameter.
More...
This is a very good point. Perhaps you can run 2 systems taking exactly opposite trades.. and see if both of them loose at the same time.
 
Quote from LeeD:

This is a very good point. Perhaps you can run 2 systems taking exactly opposite trades.. and see if both of them loose at the same time.
More...

with enough of a sample, they should both lose.
 
Quote from olias:

with enough of a sample, they should both lose.
More...

I think you are right. Because I already randomized the entry direction (long or short), there is no point in letting two strategies run with different entry directions.

What I will try is to increase the average duration of a trade, and see if the loss is the same as with short-term trades, based on the same trade count.
 
Quote from Loki45:

I think you are right. Because I already randomized the entry direction (long or short), there is no point in letting two strategies run with different entry directions.

What I will try is to increase the average duration of a trade, and see if the loss is the same as with short-term trades, based on the same trade count.
More...

Ok, but as a mental exercise, think about what would happen if you simultaneously buy and sell the same contract. ...you won't be buying and selling at the same exact price. Your buy price will be at least one tic above your sell price, and that's in a liquid market. You are always giving up the spread. So over enough of a trade sample you get a negative expectancy for both strategies (because the entries are random). If your results show profit in one of the strategies, it just means you haven't done a large enough sample.
 
Loki,

A better / more legitimate test of the random trade generator program would be to enter randomly but exit only after either a 10% profit or a 10% loss. (or some other percent, as long as it is the same for the profit and loss).

Then, over a long period of time, the results should be ~50-50, less the ask-bid spreads.

Ed
 
(1) What is the timeframe? Tick? 1 minute?

(2) How many bars are you testing?

(3) Have you tried moving your initial entry point to see if the results change a lot?
 
You must log in or register to reply here.
Back
Top