Tuesday, 3 June 2014

Looking in the Rear View Mirror

Learning from mistakes

Investors naturally tend to focus on their current holdings and how they are performing. Decisions about whether to buy, add, hold, or sell are weighed up and decisions reached. As time goes by these decisions are vindicated, or otherwise. Wise investors reflect and learn from their mistakes and refine their strategies to avoid repeating them.

Keeping an eye on previous holdings

While this kind of advice is often heard, not many commentators cast a backward look in their rear view mirror and systematically write about what they have sold and then reflect on whether they made the right call. One welcome exception is this very honest post from Mark Carter.

I think that this general lack of interest in previously held stocks is just part of human nature. Once we no longer have a direct stake in something, it is natural that we should lose interest and redirect our energy into our current holdings and the next crop of potential buys.

However, it seems to me that a retrospective analysis of what we sell is just as important as what we buy. Indeed, for any type of trading strategy, if what we buy does not go on to outperform what we sold to fund our purchase, what was the point in selling?

A retrospective of sold stocks

With this is mind, I have done just such as review of the first year of the Mechanical Bull portfolio. Sixteen stocks were sold and replaced by sixteen others during the year. Stock rotation is a fundamental design principle of the Mechanical Bull approach, but this was slightly higher than I was expecting.

This may have been because in setting up the portfolio initially, the methodology was picking up a couple of stocks with lower scores then what generally occurs once the portfolio enters a more mature phase. There were five stocks in the original portfolio that had scores of 102 (all now sold). Typically, new entrants score 103 or higher so all other things being equal there should be less rotation over the next twelve months.

Some stocks were both bought and sold during the period, so tracking relative performance gets quite messy. So, for simplicity's sake I have compared the performance of the original 15 stocks over the last year with the performance achieved by the MB portfolio. This will provide an indication of whether all this rotation has been worthwhile. The results are shown below:

Table 1: Annual Performance of the ‘Original 15’ Mechanical Bull stocks to 24 May 2014:


This shows that the portfolio would have returned 37.1 per cent compared with 42.6 per cent achieved with rotation. Although rotation did deliver slightly better results, this is hardly a ringing endorsement. The MB strategy does take account of spreads and trading costs, but one would hope for a bit more return for all that activity.

A new rotation strategy?

The MB portfolio requires some form of rotation. Stocks are constantly moving in and out of the zone of interest. Therefore there is a balance needed between staying in the zone and a sensible number of trades. I am not convinced that I quite have got it right.

One idea I am toying with is to refocus the rotation strategy towards a ‘buy’ trigger. Currently, I focus on when a stock drops below my MB score of 90. This becomes a sell trigger with the proceeds being rotated into the highest scoring stock not already in the portfolio.

An alternative might be to focus more on when a new stock appears within say the top five of all stocks. This would act as the buy trigger, which would be funded by selling the lowest scoring stock. This could help enforce better quality buys and also be a way of minimizing unnecessary rotation. I may go back through my spreadsheets and do some backtesting of this idea.

In any case, whatever happens I will keep on keeping one eye on my rear view mirror.


  1. This comment has been removed by the author.

  2. Error in previous post, sorry;

    Just a suggestion regarding your buy trigger, this may or may not be more logical:

    Maintain the sell trigger but reduce the criteria to stocks scoring below 80
    Introduce the buy trigger, but sell the worst performing stocks rather than the lowest scoring.

    The reason I say this is that I have found that stocks with a lower stock rank (80+) can still turn out to be great performers.

  3. That's an interesting idea. Obviously there is some correlation as the worst performing will tend to have the lowest momentum rank, although not always. That's certainly something worth investigating