Connect with Peter on LinkedIn!

In this news letter

(Longer than most)

Market Review

Reply to an Email

Portfolio update

Peters next speaking engagement.

Welcome to this news letter, longer than most because of the detailed reply to a great question from a client, so if you can stick it out and read it to the end it will be a great benefit to you I believe.

The Market:

The XAO continues to fall, the signal given 4 weeks ago (as seen in the top chart) when it closed below a weekly 6 EMA proved to be an early warning sign.

I see support around 4900 to 5000, (2nd chart) where there is a combination of trend line and horizontal support and resistance lines. There are still many stocks in a solid uptrend in this market and I would expect those stocks to reassert themselves quickly once the nervousness expires and  the buyers  start to outweigh the sellers again.

I see no upward turnaround just yet so perhaps we will have  more falls or a sideways consolidation like we had in June and July. I do see some positive volume support on the weekly charts but it’s too early to call anything yet. I think this is a normal correction in a Bull market but of course time will tell.  As I always say I am not in the prediction game but I know a lot of people appreciate my opinion so there it is!

My weekly system is in drawdown, and probably are many other traders portfolios. This can be a tough time psychologically and its where some faith and discipline in your method becomes paramount, there is much more discussion about this in my reply to Nicks question below. Nick works as a medical professional in Rockhampton and has been trading for some time, we have been working together to create a system to meet his objectives.

Hi Peter ,

I have a query in relation to the back testing/Monte Carlo approach*(see below at the very bottom for an explanation of what Monte Carlo is) with regards to the stocks tested.

When you carry out a back test over say a 7 or 20 year period for example, are you running the back testing over a ‘ universe of stocks ’ created TODAY that fit the appropriate fundamental criteria in the present ?

Yes

That is , are the approx. 150 stocks (fundamentally strong) that you generate a list of today, the only stocks that are run through the back testing process over whatever time period that may be ?

Yes

Alternatively, is the back testing process independent of a list generated today, and based on trades during the selected time period, that would have fitted the fundamental criteria at that particular time?

No, however,

I understand your thinking. It is an uncommon and astute question. I gather what you are really asking is if the testing has bias created by either selecting a time period or a group of stocks not relevant to the past, or a combination of both. The answer is unfortunately yes.

Bear in mind that testing is a simulation, not a certainty, as there is no certainty in trading but probability. We all want to see testing that reflects that probability with as little bias as possible.

Even if the testing is done on a group of stocks formulated from a period in the past (that represented the index as it existed in the selected time period)...for example you tested the ASX 300 as it was in 2005, the testing would still be only a simulation because stocks are delisted all the time from indexes. Takeovers, mergers, and delisting from companies going broke happen regularly, but more in the lower indexes than the indexes containing larger stocks.

Also small companies become larger companies and test well NOW in say the ASX200 but they may have been in the 200 to 500 or even lower at some stage, that is one reason I look at stocks in lower sectors climbing into the bigger indexes hoping to catch a growth stock making high % returns. However that doesn’t always work as you could see by looking at my small cap portfolio, (the link is below) but it still made a respectable 20% p.a.

Testing over a short period , say 12 months, is less likely to have bias, be that a list of stocks like the 300 or perhaps a list of stocks containing selected fundamental criteria, because it is less likely to include the bias mentioned above.

So therefore what is a solid test over the long term and how does one reduce bias as much as possible?

Geoff Dodd’s, one of my trading friends, system developer and trader who I have featured before in one of my news letters, trades a system very similar to the ones I teach (trend following in the medium term) has always had similar concerns about bias. See that newsletter with Geoff’s email here….

http://www.easysharetradingsystems.com.au/Newsletter-Archive/archive/view/1-mailinglist/33-traders-emails/24.html

Developing a system has one major technical aim and question...is it robust over a large variant of possibilities that the market will throw at you? 

Geoff’s solution to that question has been to test the whole market as it stands now, consisting of over 2519 instruments/stocks, not just the top 100, 200 or a select few groups of stocks based on current fundamental data. The aim being to produce a system that includes the entire probable’s  of every stock that exists in the market ignoring any favoritism to a sector, or fundamental grouping selected by the operator. The aim being to give the most probable result without bias.

A liquidity filter is included to eliminate the stocks of low volume in the test that the trader would be unlikely to buy. That filter is a mean average minimum turnover of 200 k per day. A minimum share price of 20 cents is also used, to eliminate skewing the results from a big percentage winner from say 8 cents to 16, a 100 per cent gain.

So here are the results from a test on all the stocks in the market (2519)over the last 8 years. Arguably one of the most difficult time frames to trade since I started trading 18 years ago. It includes for example the commodity run up and fall, GFC, the global recession/stimulus  that followed, and the Euro/Currency crisis. (Anyone remember Greece?)

 The system is one Nick and I  have worked on, (obviously for your his protection I cant display the rules )but here are the results.

Starting Capital 100k
Ending Capital  550 K re investing the money

Using 1.5 % risk of capital per trade
That will give you an average of 13 positions in the portfolio (sometimes more if the market is volatile)
 Number of Trades
197 trades over the 8 years (25 a year)
Average of 24% p.a.
44% win rate
33% Drawdown (maximum pullback during GFC)
Avg profit of winning trade 62%
Avg loss of losing trade 17%

Is this a perfect test and result? No, nothing is perfect and I am not suggesting this is a system for anyone. Many years ago when I questioned one of my trading teachers about the validity of back testing the answer was “it’s not perfect, but it’s the best we have in an attempt to predict future results from past performance”. I thought that was a very reasonable answer, its up to you if you agree.

The above results are volatile, and have a large drawdown in relation to the percent p.a. result. I like to see the drawdown number at least half of the profit result p.a. in fact myself and clients have managed to reduce Draw Down to as low as a quarter of p.a return, but you will do more trades, there is always a pay off somewhere.

You know from the work you have done with me Nick these results can be improved significantly. The question really is “does the system have merit and is it robust?”  I think that the results above are quite good (you may argue otherwise), considering the broad data base and the fact the system is trading right across the board including stocks of a low fundamental nature that are highly volatile. The position sizing method obviously doing its job well.

Also the blue line on the graph  represents the result if the trader had spread his/her money equally across all the stocks tested and just held and not traded. In 8 years the buy and hold funds have risen to 131k but the system has gone to nearly 550k. The system clearly outperforms the market it is trading. That shows the power of sticking to a method along with compounding the results. The big recent run up on the graph above is because it is displayed in Arithmetic scale exaggerating the recent results, the Logarithmic scaled chart below displays a more true steady climb of the method over the years.

The logarihmic scale chart above.

The fact that the system has performed so well trading an index that wasn’t performing well (the whole market not just certain sectors like health and banks) is pretty impressive. It also shows why stock selection of a fundamental nature or at least in an index more selective like the 300 or 500 helps, as Nicks next question leads in to.....

* Alternatively, is the back testing process independent of a list generated today, and based on trades during the selected time period, that would have fitted the fundamental criteria at that particular time ?

The * point above would be my preference for a true indication.

Either way, I am pondering what effect this may have on the back testing results ?

Something else that can be done to test the probability of a method is to test a group of selected stocks in a past time frame. So what if we tested a group of stocks with strong fundamentals in the present (Dec 2013) but in a past time frame to remove present fundamental analysis bias? In other words, just because a stocks fundamentals look good now, how did the system work on them before that? when perhaps the fundamentals and market confidence were not so good?

The following is a test using the same system above between 14/12/2006 to 14/12/2011 a 5 year period including the GFC but on a selected group of stocks with strong fundamentals calculated in the NOW not then. Does the system have merit and pass the robust test?

The system was going well climbing 27% to 127K until the GFC, it cashed out almost even at 95K  and then re invested for an amazing climb to 256 before cashing out at 240K, all with just 85 trades in 5 years. A great example of why predicting can be a waste of time, who could predict that result after the carnage of 08? Best to follow a high probability method. The system also had a great 55% win rate on these stocks compared to the 44% win rate with the other test over the whole market (2519 instruments),  understandable because these are more solid and less volatile stocks. This system trading the strong stocks has gone on to continue to outperform the market, below is what it did next.

Notice here how this group of stocks is holding the line and not selling out during the current sell off. (The dark green on the graph is cash in the portfolio) They will exit quickly if the market continues to fall but for the moment its ok. This has been a similar run as to the 2006 to 2011 test with just 40 trades or 20 a year. Until just a year ago most market participants were full of doom and gloom, yet the system has almost doubled its money in 2 years.  Over the medium to long term the current pull back in the market and system will look like a blip on the radar. Below is the performance of the total test period from 2006 to the present. Note the low Drawdowns.

So it seems that fundamental scans over the market to decrease the number of stocks to select from makes a solid difference. Perhaps not so much in a greater p.a. return but producing a method that is EASIER to trade. Why waste time analyzing and trading 2519 stocks when we can get a better result from around 150 strong stocks, and that’s not necessarily the top 150 by size in the ASX, but the BEST 150.

All systems can be improved or adjusted to suit the trader’s objective. Subtle changes to things like the sizing of trades, when to enter the market, how to respond when the market weakens, what group of stocks to trade and of course what exit you choose, make far more difference to the systems robustness than over focusing on the entry like most novice traders do. I always say to clients DO NOT BE SEDUCED by the p.a. returns, but look at what you can psychologically execute and stick to. Lets develop a system YOU can trade and STICK to.

If you agree in the theory that “past performance is the best predictor of future results” then perhaps you are starting to feel a little more comfortable from this discussion.

Here is an analogy:  If you were going to pick a player for the Australian cricket team you would look at his past performance, batting and bowling averages, statistics, strike rate etc. If the player has a good record you would be inclined to pick him with the view he has proven himself in his current arena, but no one knows if he will perform at a higher level…the selector needs to take a risk.  In my opinion trading is a game involving risk, put the odds on your side and it gives you the best chance of winning in an uncertain environment where anything can happen, just like cricket! (Congratulations to the Aussies for winning the ashes)

Conclusions:

Analysis and lots of it is good to a point, it helps you believe in the system. If a trader doesn’t have confidence in the method he/she is using when the going gets tough (like now with the current drawdown), they won’t stick to it. The way to get confidence in the system is to understand it well. So thanks again Nick for your question and I hope this deeper analysis of my systems help others.

However, no matter how well a trader understands a method and continues to search for a better one, or one that makes them feel more secure, at some stage they have to commit. It’s by committing to a method with an amount of money that will not stress the trader that gives them the best chance of success. If you are afraid of a trading commitment than start with a small amount of money and FOCUS ON EXECUTION AND STICKING TO THE RULES.

Good trading is about good execution, sticking to the plan.  Not over trading and/or grasping for ideas or changing the plan when you feel like you are drowning in a sea of emotion. If you are drowning in a sea of emotion you are probably over trading, don’t know what your odds are or have too much money invested for your mindset. Perhaps all of those operating together is why you are struggling.

Portfolio Update:

Thank you all for being patient with this page of my website. I had a technical issue that drove me to avoid it for a long time (that is my lack of discipline) but its up and running. If you want to know more about what stocks I am holding, buying or selling  click on the link here

http://www.easysharetradingsystems.com.au/peters-portfolio.html

Want to know more?

I will be lecturing at the Sydney ATAA on January the 20^th.  The lecture is free for first time visitors, or if you are a current or interested client email or ring me to arrange a free ticket. Ph: 0403821523

 The presentation will be focusing on the performance of a system for the ASX 300 I presented to ATAA in Mid 2012.

The system on the ASX 300 since entering the market in August 2012.

I will discuss the method and the  rules, look at the entry and exit points of the system and also discuss some trades in detail.

 I will also discuss pchychological aspects of trading about why people find it so hard to choose and then stick to a method, which are the 2 reasons why traders fail to make consistent profits from the market.

Here is the link to the ATAA page

http://www.ataa.com.au/meetings

I hope you enjoyed this news letter, it is my Christmas present to you as it was a lot of work! I wish you all the best for you and your family now and for next year.   Cheers! Peter.

"Geniune expertise status comes from objective testing, not their own judgement, which may be just the illusion in action"

Colin Nicholson from his book "Think like the Great Investors"

* Monte Carlo: This analysis consists of computing a large number of randomized Runs based on the original results, and can provide insight into the trading system's potential in the future.  It can also give you an idea of the probability of achieving various profit objectives.

If Facebook is more you thing, friend him here!