I recently seen the following on a forum:
Given the following two systems, which would you prefer:
- System A which has a 60% win rate with a ratio of average win to average loss of 1:1
- System B which has a 40% win rate with a ratio of average win to average loss of 2:1
The first step most traders would undertake is the calculation of the expectancy for each system.
E = (Win Rate x Average Win) + ((1 – Win Rate) x Average Loss)
E = (0.60 x 1.0) +((1-0.6) x -1.0)
E = 0.20
E = (0.40 x 2.0) +((1 – 0.40) x -1)
E = 0.20
Over a large enough set of trades, we can expect both systems to generate identical returns. So all things being equal, the systems are identical but all things are not equal. When you consider a trading system, what other performance measure should be at the top of the list besides return? It’s maximum draw down of course. For example, if Systems A and B both generate annual returns of 15% but one system will incur a maximum draw down of 15% and the other will incur a maximum draw down of 25% you will choose the system with the lower draw down.
If all we have is the information given above for both systems, how can we determine which system is likely to have the higher draw down? One possibility is to perform a Monte Carlo (MC) analysis.
I performed a MC analysis on both systems using a sample size of 5,000 trades. As expected, the returns generated by both systems were almost identical but System B had a much higher draw down in the MC analysis. At the 95% confidence level, System A had a Return-to-Draw Down ratio of 2.48 versus 1.47 for System B.
Other traders selected System A as well albeit by using different methods of comparison.
Lately I have had conversations with investors who can’t hold back from telling me how well their stocks have performed in recent years. I’m happy that they have done well but can’t overlook the time period that they are considering.
It is human nature to place more weight on recent trading performance than historical trading performance. I suspect most of us are guilty of that. Let’s bear in mind what happened five years ago this month by considering the following chart of the S&P 500.
The S&P 500 hit a low in early March, 2009 but then began a five year bull run. From the low of March, 2009 to now, the S&P 500 has achieved a compound annual growth rate (CAGR) of 21.4% which is incredible. The past five years have been a period in time where buying and holding SPY has worked very well but you must not confuse a bull market with brains.
Let’s look at two other time periods with the same end point, March, 2014, to see how the S&P 500 has performed. For the seven-year period from March 01, 2007 to now, the S&P 500 has achieved a CAGR of 4.0%. For the ten-year period from March 01, 2004 to now, the S&P 500 has achieved a CAGR of 4.8%. Clearly, the past five years have been exceptional for the S&P 500 but, in my opinion, there is no possibility of a repeat performance for the next five years.
After three months of losses, my trading systems performed well in February and posted a 3.3% gain.
As per the above chart, my stock and ETF trading continues to strongly outperform my benchmark, the IASG Systematic Trader Index.
I mentioned in my January update that I would perform the normal process of monitoring and adjusting my trading strategies. Some trading strategies were dropped at the beginning of February and the position size was increased for other strategies. Those adjustments, in part, accounted for the strong performance of my trading last month.
The trading results I present here are from my Collective2 account for which you can always see the performance statistics here.
For what it is worth (very little actually), I am up 2.1% already in March. Let’s see if I can put in another strong month.
If you browse Youtube for trading videos, you will likely be disappointed with what you find. When I used trend following strategies, I occasionally corresponded with Nick Radge of Australia. Nick is a good guy who will give you a straight answer that demonstrates his many years of trading experience.
I decided to search for Nick on Youtube and the first video I found was indeed a good one. This video was taken when Nick made a presentation to the 2009 Australian Technical Analysis Association’s annual conference in Melbourne. Unfortunately the audio is less than perfect but the messages presented are excellent. You can learn more about Nick by visiting his website.
To go to the video, click here. Dr. Howard Bandy makes a cameo appearance in this video. If you don’t know who Dr. Bandy is, I strongly suggest you visit his site and read his books.
You can also listen to Nick on one of Michael Covel’s podcasts (click here).
PS I noticed a mistake in one of the slides that Nick used. I checked with him and he is aware of the error but obviously he can’t make a change to the video. The slide in question appears around the 23 minute mark and is shown below.
In the calculation for the expected losing streak (i.e. LS = expected maximum number of consecutive losing trades) for a system that has a 50% win rate, the “/100″ should not appear in the formula. The equation should be:
LS = (LN(50000)/-(LN((1-0.5)))
In this example, over the course of executing 50,000 trades with a system that has a 50% win rate, the trader can expect to have 16 consecutive losing trades and this has an obvious impact on the position size that should be used.