Progress So Far

[Stephen James]

Luke, you dollar stop is correct as the dollar amount is just subtracted from price.

The % stop uses 1-Param(“Trailing Stop %”,20,1,100,0.1)/100 simply to see the adjustable parameter as the correct % level.
So the actual calculation in that line is 1-(20/100)

Its the same as 0.8 so you could use H * 0.8 to get the 20% stop level instead of the extra coding required to work out 20% then subtract from price.

[JulianCohen]

I am trying to improve a mean reversion system that has 10% CAGR, 16% DDs, and a Ave Profit/Loss of .57%

Any suggestions on ways to make the ave Profits higher than the ave Losses? Just generally I mean.

It is based on a Connors/Alvarez system that is:

//Enter
//C > MA(200)
//VIX 5% or more above its 10day MA.
//2 day RSI < 5
//Maybe add 3 lower lows
//Enter on limit order .25ATR(10) below low (try close) on next day

//Exits
//Exit if VIX 5% or more below 10day MA
//Close higher than previous 2 days
//RSI(2) > 65, 70 or 75
//Exit open next day

[ScottMcNab]

Found (gross generalisation) that increasing size of pullback needed for buy limit order up to 0.8 to 1.2*atr reduces maxxDD (as well as trade frequency unfortunately)

I also found exits based on next open or close>ref(c,-1) (or similar variations) seemed to work better for me than those based on indicators suchs as RSI/PDI/MDI etc

[JulianCohen]

Thanks Scott. I’ll experiment with that and see how I go. Maybe I’ll try removing all the exits except c > ref(c,-1) and see how that goes. I’m trying to work out how long I should stay in a mean reversion trade (in general)

[Nick Radge]

Quote:

I’m trying to work out how long I should stay in a mean reversion trade (in general)

I like hopping in and out quickly then getting onto the next trade. Keeps exposure down.

That said many others here are using profit target type exits.

Don’t forget that you can send your system to me to help enhance. Just let me know.

[TrentRothall]

Quote:

Maybe I’ll try removing all the exits except c > ref(c,-1) and see how that goes.

Maybe do the same with the entry criteria, strip it back and just have 1 or 2 rules and start from there instead of 4 or 5.

[ScottMcNab]

Said’s post has some great info on the impact of different exits with MRVs

[SaidBitar]

Julian Cohen wrote:

I am trying to improve a mean reversion system that has 10% CAGR, 16% DDs, and a Ave Profit/Loss of .57%

Any suggestions on ways to make the ave Profits higher than the ave Losses? Just generally I mean.

It is based on a Connors/Alvarez system that is:

//Enter
//C > MA(200)
//VIX 5% or more above its 10day MA.
//2 day RSI < 5
//Maybe add 3 lower lows
//Enter on limit order .25ATR(10) below low (try close) on next day

//Exits
//Exit if VIX 5% or more below 10day MA
//Close higher than previous 2 days
//RSI(2) > 65, 70 or 75
//Exit open next day

Maybe it will be good idea to test each system (condition) as a separate system, then you start combining them one by one and see the effect of each on the other.

[JulianCohen]

OK that’s a good idea..thanks

[LeeDanello]

When you say you test each condition in a separate system, what do use as the base system?

[Nick Radge]

Why use a different base system? If you’re testing apples vs apples then stick with them apples.

TradeStation has a great way of adding and removing elements of a system without adjusting any other parts. I could separately code 10 different exit methods then add/remove to any entry criteria.

I would use a base code strategy then test each exit individually.

[JulianCohen]

Yes I’m going to code up the whole system, and then strip it back to one entry and one exit, and go through and see the effect of adding and subtracting different entries and exits. I might even get organised enough to write it all down and report the results!

[Author]

Posts