Julian’s Journal

[JulianCohen]

Rob Giles wrote:

Julian, thanks for sharing.
Are the US WTT and the US Momo Nick’s systems?

Not exact copies but the basic structure is the same yes. I coded the WTT from the book and then fiddled and the Momo is a basic rotation system that has been fiddled.

I like fiddling

[RobGiles]

Doesn’t make you a bad person

[SaidBitar]

Julian Cohen wrote:

US WTT 7.0%

Good one

[JulianCohen]

Thanks Said. I fear selection bias is at play, but I’ll take it anyway

[JulianCohen]

My US WTT system is showing a red ribbon at the bottom for the first time since April. My up 7% from last month got a right good kick in the goolies this week.

[JulianCohen]

I’ve been working on my US systems.

I managed to get the MOC to go from this

to this

It might look like it is worse but the selection bias has gone from this:

2010-2017
Selection Bias 94%
2012-2017
Selection Bias 97.6%

65 Days with 40 or more signals from 2007-2017
22 Days with 40 or more signals 2011-2017

to this:

2010-2017
Selection Bias 93.5%

2012-2017
Selection Bias 98.5%

49 Days with 40 or more signals executed from 2007-2017 (16 with 70 plus)
17 Days with 40 or more signals executed 2011-2017 (only 4 with 70 plus)

I’m happier to trade a system with less CAGR and a tighter selection bias…as long as the fills match up to the backtest which is something I haven’t been checking thoroughly but now I will. Each day I will recored the slippage for each system to monitor it more carefully.

[JulianCohen]

I have also managed to get my US MR system to this

It might not look like much but it has this selection bias

2012-2017
Selection Bias 98.5%
1 day with 20 or more signals executed

2010-2017
Selection Bias 90%
5 days with 20 or more signals executed

2007-2017
Selection Bias 82%
14 days with 20 or more signals executed (9 days with 30 or more)

So at least I know I should get 18% pretty consistently as long as the fills hold up.

What do you think? Is it feasible to get a better return with such a good selection bias?

[ScottMcNab]

What dates are the cagr/mdd/MC stats for Julian ? Are they for the same dates as the selection bias ? I can then throw up some stats for this period.

Its interesting comparing apples with apples….once selection bias is accounted for then the apparent advantage or attraction of the MRV/MOC over the rotational systems starts to fade

[JulianCohen]

Scott McNab wrote:

What dates are the cagr/mdd/MC stats for Julian ? Are they for the same dates as the selection bias ? I can then throw up some stats for this period.

Its interesting comparing apples with apples….once selection bias is accounted for then the apparent advantage or attraction of the MRV/MOC over the rotational systems starts to fade

Dates are 1/1/2012 to today

I’d be interested to see how you go

[ScottMcNab]

Give me a day or two to sort this selection bias code out..I suspect that my backtests results are originating from a parallel universe where I am a trading gun…not in this one

[JulianCohen]

Scott McNab wrote:

Give me a day or two to sort this selection bias code out..I suspect that my backtests results are originating from a parallel universe where I am a trading gun…not in this one

[JulianCohen]

August ’17

Pretty quiet month.

ASX MR 0.6%
ASX WTT 1.9%

US MOC 0.4%
US MR 0.8%
US WTT -4.0%
US Momo -0.62%
NASDAQ Momo -1.32%

However my account is in positive territory three months in succession, +2.3%, +0.8% and +0.26% so I’ll take that.

[ScottMcNab]

Julian Cohen wrote:

Scott McNab wrote:

What dates are the cagr/mdd/MC stats for Julian ? Are they for the same dates as the selection bias ? I can then throw up some stats for this period.

Its interesting comparing apples with apples….once selection bias is accounted for then the apparent advantage or attraction of the MRV/MOC over the rotational systems starts to fade

Dates are 1/1/2012 to today

I’d be interested to see how you go

OK…on thos dates:

ASX: 16.73% / -13% (ASX didnt need any changes as no real SB)
US: 24.46% / -6.63%

So ideally (if ASX doesn’t turn out to be dud with missed trades) would combine in day/night sessions to give CAGR high 30%

I am thinking of what you did and taking the pain and going back to scratch for a re-design

[ScottMcNab]

ps…those were the results with the SB worse case positionscore

[JulianCohen]

Scott McNab wrote:

ps…those were the results with the SB worse case positionscore

If those are worst case then that’s fine. No reason not to trade it while you have a fiddle from scratch because face facts, you know you’re going to do that in the end

[Author]

Posts