End of the road

To me, Friday looks a lot like the end of the road. There are a couple of support areas nearby which may hold back the slide. But Friday's weakness came at the end of the day, and the day was 31st of May. That suggests that the last of the smart money has come off the table and the appetite for buying has gone.

The after-the-event news reports tell of the market's expectation of good employment figures at the end of this coming week as the cause (good economic news speeds the end of QE) but  it is more likely that it was the continued rise in interest rates that caused the sell off.

I like sharp stock market falls, as long as I have ditched the bulk of my holdings and am sitting on cash. Lower share prices mean that there are bargains are to be had. The big danger is that, with no holdings and loads of cash, I will succumb to the temptation to fiddle with 'interesting' opportunities. A good way to lose money.

But in the mean time

Remember I was struggling with backtesting my Support and Resistance (S&R) timing list. 

My problem was that I had found some excellent strategies which yielded fabulous returns over the period November 2008 and May 2013 (about four and a half years). But  they had unacceptable levels of drawdown. Drawdown is the money you risk losing when the system makes decisions that lose. The maximum drawdown levels were over 25%. Why is that bad? Just imagine that you start your investment programme just when the market falls into one of these holes. Bang goes 25% of your fund.

After a lot of thought and work I finally found an objective method of eliminating the worst drawdown culprit that occurred in autumn 2011. So I’ve added  “The Ten Day Rule” to my system. When a down signal is followed by an up signal after less than ten days then that putative up signal is ignored as are all subsequent up signals until an up signal is followed by a clear run of ten days. At that point it counts as confirmed up and the system can restart buying and selling as normal.

The additional of this rule reduced the maximum drawdown from over 25% to less than 20%. Unfortunately I could

not conceive a system which would eliminate the next largest drawdown . The only way to deal with the problem is human intervention. I am the one who is trading, not the system. If the system starts churning out unacceptable returns it is up to me to put it straight. The risk is getting it wrong, but, as the last line of the film Some Like It Hot has it: “Nobody’s perfect.”

A by-product of all this work is that I feel more relaxed as my portfolio falls into losses. The results of the tests prove that there’s a lot more money to be had if I stick to my guns.

The big bucks come from share picking

The other thing that has been brought home to me is that share picking is the way to make the really big bucks. Share picking has been the core of my approach to the market for all the years that I have been playing the markets. I have made money, but I have struggled to find a timing method that tells me when to get in and out, I have had to guess. Now, with the tools provided by Vector Vest, I think I have found and tested a reliable timing system.

However, without share picking techniques that underpin my activities my returns would continue to be decent but unexciting (my average over 12 years has been 17%pa). I have tested VectorVest’s unisearches extensively and have picked two top performers. Let's call them Picker a) and Picker b). Picker a) returns consistent results and it usually finds shares that meet its criteria. Picker b) is more wayward and often it finds no shares, so I have to wait with nothing in the market or just a couple of shares instead of the ten which normally spread the risk in my portfolio. 

Using the S&R timing system picker a) produces a 44% pa return with a 19.6% maximum drawdown. Picker b), using the same timing testbed yielded an annual rate of return of 220% pa for those four and a half years with a maximum drawdown of 27.5%.

Longer periods of backtesting are still needed. The further back I go the more reliable the results and the bigger the range of market conditions that the test encounters. But essentially I am ready to go. One part of my portfolio will be run using Picker a) and a smaller part will be tried out on riskier but potentially more profitable Picker b).

Making money is all down to doing your homework.

Stop press

I spent yesterday extending S&R timing back to March 2008. I now have just over 5 years of tests for my picking and timing systems. More importantly I have included a period of serious slide in the stock market.

Between March 2008 and November the DJI fell by 32%. Including this period applied a more stress to the tests. As I would have expected Picker a) struggled more than during the ugly down . As a result, it yielded 23% pa when tested from March 2008 compared with the 40% from November the same year.

The big surprise was Picker b) which yielded a staggering 319% pa. when tested from March. Here's a chart which shows how and when the money was made.

A note of caution in reading these results. Vector Vest focuses on the average rate of return when it reports on backtests i.e. the total return divided by the number of years. In fact the total is achieved by compounding. The compounded rate of return is a more modest but nevertheless eye watering 75% pa.

I have not finished my work and will run my tests back further but now I have a plan of action.