Julian’s Journal

[JulianCohen]

December ’21

Short-Term Systems

US Combined MOC: 6.30%

Long Term Systems

Long Term NASDAQ: -8.33%
US Momentum: 4.89%
US WTT: 0.67%
ASX Growth 1.23%
ASX Momentum: 11.73%

[JulianCohen]

January ’22

Short-Term Systems

US Combined MOC: 5.48%

Long Term Systems

Long Term NASDAQ: -20.54%
US Momentum: -1.72%
US WTT: -3.65%
ASX Growth -5.20%
ASX Momentum: -9.58%

[TimothyStrickland]

Julian, the drawdown on the Long Term NASDAQ seems equivalent to mine, you mind me asking what your current drawdown is? How high was it at the end of last week?

[JulianCohen]

Hi Tim,

it went as low (high) as -25.7% last week…27th Jan was the worst day

As of today it is 19.9% DD…made a bit back from yesterday

Sorry I just realised I am measuring DD on a YTD basis, so my figures reflect that.

If I look on DD from all time highs it wouldn’t be particularly pleasant….much closer to mid 30’s I would suggest.

However I’m quite sanguine about it. I know the principle is correct, I have no concerns that Trend Following is broken, and for the last two years it has made a shit ton of cash, and NAV since the start, about four years ago is 2.5 times the start…so in a way, drawdowns from Max High can be a bit scary…

That’s why I do them on a YTD basis, so I don’t worry about them too much.

If it was anything but trend following, I wouldn’t do it that way…but the basic principle to me is so sound, you can pretty much do anything and make money at it, as long as you are buying with momentum. How much money you make is the only difference between the strategies…they should all make money in good times and give about 30-40% of it back in bad times

[JulianCohen]

February ’22

Short-Term Systems

US Combined MOC: 4.26%

Long Term Systems

Long Term NASDAQ: 0.31%
US Momentum: 4.70%
US WTT: -1.30%
ASX Growth 0.46%
ASX Momentum: 1.34%

Total Account: 3.26 %

[JulianCohen]

March ’22

Short-Term Systems

US Combined MOC: 5.68%

Long Term Systems

Long Term NASDAQ: 0%
US Momentum: 0%
US WTT: 0.10%
ASX Growth 0.22%
ASX Momentum: 0%

Total Account: 2.41%

[TerryDunne]

Wow, that’s great work Julian, well done!

[JulianCohen]

April ’22

Short-Term Systems

US Combined MOC: -3.57%

Long Term Systems

Long Term NASDAQ: -4.69%
US Momentum: -3.71%
US WTT: -9.66%
ASX Growth 0.66%
ASX Momentum: 0%

Total Account: -4.06%

[JulianCohen]

May ’22

Short-Term Systems

US Combined MOC: -2.32%

Long Term Systems

Long Term NASDAQ: -0.09%
US Momentum: -0.65%
US WTT: -0.41%
ASX Growth -1.66%
ASX Momentum: 1.83%

Just woke up in Abu Dhabi to see my exit from Origin didn’t fill in the opening auction….it’s now 2pm in Sydney….That’s unpleasant…

[Nick Radge]

Yeah…I went to the gym and had 2 from 5 fills only…

[JulianCohen]

June ’22 Short-Term Systems 
US Combined MOC: -17.01% 
   
Long Term Systems Long Term
NASDAQ: 0%US Momentum: 0%
US WTT: 0%
ASX Growth -2.4%
ASX Momentum: -5.11%

There’s a month that will stand out when looked back on, hopefully!

 

[KateMoloney]

Its been an …. interesting ? month

Currently doing some research on my systems. The daily P & L chart says it all …. last month healthy losses and the profit side is errr…. flat. In a 20 year backtest it really stands out.

[JulianCohen]

In order for this to stand out in the future, there has to be a recovery in the drawdown chart so this appears as a downwards spike.

Unless it stands out as the start of a long prolonged drawdown [GULP]

[KateMoloney]

This is our fault Julian. We didn’t appease the angry trading gods last month.

*Runs for the hills, quits trading and meditates in a cave for the rest of eternity*

[JulianCohen]

   I’ve been doing a bit of testing over the weekend and I thought I might document my thinking and process as it may help someone in the future, including me. (It’s amazing how I look back at things I’ve written in the past and not remembered any of the thoughts behind it) I have an MOC/MR strategy that is a portfolio of nine different individual strategies. I was looking at the metrics for some of them, and I was trying to work out how to improve them. Mainly looking to boost MAR and Expectancy. A couple of the strategies had expectancy of 0.15% and I’d rather see that over 0.2% with MAR as close to 1.0 or over. The portfolio as a whole has MAR of over 5 and Expectancy of 0.26% (2014 to present) so as a whole metrics are OK, I was just trying to see if I could improve the individual strategies. 

Often I have found that when a strategy has an edge but just isn’t performing as well as I’d like, combining it with another strategy that has the same issues and running as one can improve things, so that was my process. I hate wasting what I thought was a good idea and it really has to prove to me that it is not a good idea before I abandon it. 

Basically code wise I am using something along the lines of: 

Strategy1 = Three lower lows

Strategy2 = ADX(5) > 30 and C < MA(C,5)  Cond1 = Strategy1 OR Strategy2   These are just examples, not the actual strategies, although my strategies are all pretty simple.  After a fair bit of work I have reduced my 9 strategies down to 6, without wasting any good ideas. I think they all have an edge, it’s just a case of making the edge better. The combined MAR is now over 6 and the combined Expectancy is 0.28%  These 6 strategies also have the rule that they are not allowed to hold the same stock on the same date as another strategy. I have been developing this way for a while. This forces me to create strategies that intrinsically had a good edge, and didn’t rely on the same stocks to boost returns as the other strategies in the portfolio. Much more difficult to do, but over the past two years or so I have slowly worked at the portfolio to work this way.  Creating a portfolio of strategies this way means that I have had to diversify the universes, so that there are a decent amount of stocks to choose from, but also that there is as little overlap as possible. If they all ran on RUA then the strategy that was last in the list would run out of stocks to choose from fairly quickly.  Ranking is also vital here to allow the stocks selected to go into the backtest to be as varied as possible. For instance I can have two strategies running on RUA but by using a different ranking method, that allows the variation so each strategy has a decent set of stocks to choose from. Amazingly I have found a usage for Historic Volatility as a ranking method at last. Cesar Alvarez’s go to ranking method, according to his blogs anyway, has always been massively outperformed, for me at least, by ROC, but with a few different strategies all working the same stock universe, it has it’s place.  I have tried to keep parameters as broad as possible. For example if I have ATR as a parameter I use the same ATR(20) anywhere I am going to use that parameter, irrespective of the strategy. I did this to keep things as robust as possible. I’m sure I can make the backtest much much better by tweaking the parameters, but by deciding that this setting is the one only one I am going to use across the board, this stops the temptation to individualise each strategy. Once a portfolio is built I might then see if ATR(10) is better, but it will be ATR(10) everywhere ATR is used, not just on one or two strategies.  I keep the same number of entries for each strategy, however I vary the position size. I might have some strategies running on 20 positions at 5% and some at 3%…Their overall total is more than 400%, however in backtesting from 1995 to today it has never gone beyond 500%…I cover this by running the portfolio at 80% of the tested account size so on the one day that it does happen, and it will one day, I won’t be caught out. I don’t use any added selection bias, 20 entries, 20 candidates for the backtester.  It is worth remembering that not every strategy has 20 actual entries every day. Sometimes there are only 5 or so; it is quite rare to have all of them with 20 entries, but it could happen. My average overall usage for the last twenty years is 140% (One strategy running 20 positions at 20% per position is 400%)  One caveat, I use RealTest exclusively now for my backtesting so if you want to know how to do some of this in Amibroker, I’m pretty rusty at that. I’d go so far as to say some of it would be so clunky to do in Amibroker that you might not want to do it.    

[Author]

Posts