Low-Risk investment strategy ?

aliciathyme

grey_gym_sock wrote: »

1 thing to consider is that there are a lot of ppl (mostly using computer programs) trying to make money from trading based on technical analysis. some of them trading with large quantities of money - large enough to affect the market's behaviour. and most of them with evolving strategies. this suggests that a strategy which has worked for a time may be identified by other traders, and that the weight of their money can then affect the market, which may lead to the strategy no longer working.

Thanks for that warning - there have been requests for me to publish future buying recommendations (post here or twitter etc), but that would be a mistake if it were to cause too much market movement and adversely impact the strategy.

But thanks also for your other comments about my system.
I'm glad you have found something useful in it.

Alicia

Thrugelmir

aliciathyme wrote: »

but that would be a mistake if it were to cause too much market movement and adversely impact the strategy.

Considering you trade in FTSE 100 stocks I doubt it. For example GmS deal in Vodaphone around the £10 million level per trade.

Investors Chronicle probably have more followers than you as well.

jamesmorgan

aliciathyme wrote: »

If the trend over the last 2 weeks exceeds 10% and the price today is lower than yesterday then I consider that to be a buying opportunity. This works well in a rising market and allows me to lock into companies with a well established rising status (this accounted for 50 holdings averaging 23%, plus 3 stoplosses).
Alicia

Hi Alicia,
I am a reasonably experienced technical trader (this has been my sole income for the past 10 years), and I am always open to new ideas. There a few points from your various posts that I don't fully understand. Firstly you initially stated that you look for a 10% trend over 2 weeks. Later on you stated 20% over 1 month. Which is correct, or have you changed criteria over the past 5 years?
Secondly with your table on banking shares it appears that you purchased RBS on 18/1/12 even though this does not appear to be following a down day. Is there a certain amount of discretion in your buying criteria? Also with RBS you appear to have purchased again on 18/9/12 even though you had not sold the purchase on 18/1/12. Does your system allow multiple purchases of the same share? If so is there a minimum time period between purchases - otherwise what stops you from buying on day 1 and then on day 2 if it falls again (and still above trend)?
Thirdly with banking shares I am a little surprised that there are no purchases of LLOY this year. There appear to be be buy signals at least on 18/6, 23/8 and 17/9. The last one was on the same day as the RBS purchase so there should have been funds available for the purchase.
Overall as a general observation your success rate of winning shares of 81% is very high. Most trading systems rely on a success rate of closer to 50%, but make much larger gains on their winners then loses on their losers. With such a high success rate I suspect that it doesn't matter too much what your sell strategy is. The key is understanding why your success rate is so high and whether it would continue to be successful in all markets - especially a relatively flat market.

aliciathyme

jamesmorgan wrote: »

Firstly you initially stated that you look for a 10% trend over 2 weeks. Later on you stated 20% over 1 month. Which is correct, or have you changed criteria over the past 5 years?

Thank you for your interest jamesmorgan.

By way of clarification, the trend is measured over a period of the last 20 trading days.
As it's a linear trend, a rate of 10% per 2 weeks is the same as 20% per month.
I apologise for being a bit lax on this.

Secondly with your table on banking shares it appears that you purchased RBS on 18/1/12 even though this does not appear to be following a down day. Is there a certain amount of discretion in your buying criteria? Also with RBS you appear to have purchased again on 18/9/12 even though you had not sold the purchase on 18/1/12. Does your system allow multiple purchases of the same share? If so is there a minimum time period between purchases - otherwise what stops you from buying on day 1 and then on day 2 if it falls again (and still above trend)?

I have RBS at 250.2 on 17/1/12 and 245.6 on 18/1/12. My data is timed mid afternoon, so if you are looking at closing prices there would probably be a variation.
The system takes no notice of existing holdings when searching for a new opportunity, so there is occasionally a new purchase of an existing holding - note that much of the discussion has revolved around trend processing, but I also use an anomaly function which does not use trends in the same way.
There is no holding time or interval time function: the trend function does not give multiple buys because it only accepts the first point at which the conditions are met.
(If the trend continues at >20% and the share price continues to rise and fall, only the first fall meets the conditions)
However, if the trend falls below 20% and subsequently regains 20% the first price fall within the >20% trend period will meet the conditions

Thirdly with banking shares I am a little surprised that there are no purchases of LLOY this year. There appear to be be buy signals at least on 18/6, 23/8 and 17/9. The last one was on the same day as the RBS purchase so there should have been funds available for the purchase.

In the banking sector I currently have just the 2 RBS previously mentioned (post No 22) holdings bought on 5/08/11 and at last coming into profit at 9% and again on 18/9/12 now at 12.6% profit.

On the days you identify for LLOY, I have trends (%/month) as follows:
18/6 -7.0%,
23/8 +14.4%
17/9 +15.6%
So none met the criteria for 20%.

Interesting that you make a living out of investing, If that means a regular income, I can't see my system being helpful. I view it as a means of long-term growth.

Alicia

jamesmorgan

aliciathyme wrote: »

On the days you identify for LLOY, I have trends (%/month) as follows:
18/6 -7.0%,
23/8 +14.4%
17/9 +15.6%
So none met the criteria for 20%.

Interesting that you make a living out of investing, If that means a regular income, I can't see my system being helpful. I view it as a means of long-term growth.

Alicia

OK - I must have misunderstood what you meant by trend. I was simply looking for a share price close 20% higher than 20 days ago. How exactly are you measuring trend?

Regarding taking a regular income, I trade more frequently than you (typically around 100 trades/annum). Every few months I simply take some of the excess cash out as income. I have a different approach to asset allocation in that I split my total funds in 6 and allow a maximum of 6 purchases at any one time. Once the 6 is fully utilised I don't buy again until a share is sold (I have tightly defined sell criteria). I have considered whether to sell one of my existing stocks when new buy triggers are met, however, I have yet to find a method of identifying the most appropriate share to sell. One area that you may wish to research is loss aversion. This is a basic psychology of humans (and other primates) where losses have a greater impact than wins of the same magnitude. Given the choice between always winning 1 token, or 50% of the time winning 2 tokens with the rest of the time 0 tokens, people will always chose the former (even though statistically they are the same). The reason this can be important is that investors are often reluctant to materialise loses - much preferring to sell shares that have risen. This view will persist even if statistically it is the worst approach. I'm not saying that in your trading this is the case, but just pointing out the risk that you may be acting on basic human psychology rather than objective analysis.

Jegersmart

jamesmorgan wrote: »

OK - I must have misunderstood what you meant by trend. I was simply looking for a share price close 20% higher than 20 days ago. How exactly are you measuring trend?
.

Yes, this crossed my mind too.

J

aliciathyme

jamesmorgan wrote: »

OK - I must have misunderstood what you meant by trend. I was simply looking for a share price close 20% higher than 20 days ago. How exactly are you measuring trend?

I use a simple trend formula:
Suppose you have 21 share prices in an Excel spreadsheet column from cell A1 down to A21
Insert this formula in cell B21 to give the current trend of of the last 20 values:

B21=((TREND(A1:A21,,{21})/TREND(A1:A21,,{20}))-1)*20

(that's very funny! good job there's a preview posting option! replace the "you are wonderful" image with a colon!)

Then you have a target buy if B21>20% AND today's price < yesterday's (ie A21<A20)

I hope this clarifies the trend issue.

One area that you may wish to research is loss aversion. This is a basic psychology of humans (and other primates) where losses have a greater impact than wins of the same magnitude. Given the choice between always winning 1 token, or 50% of the time winning 2 tokens with the rest of the time 0 tokens, people will always chose the former (even though statistically they are the same). The reason this can be important is that investors are often reluctant to materialise loses - much preferring to sell shares that have risen. This view will persist even if statistically it is the worst approach. I'm not saying that in your trading this is the case, but just pointing out the risk that you may be acting on basic human psychology rather than objective analysis.

I know what you mean, but have to repeat that I am not using my personal judgement, simply relying on the number crunching to do today what it has (nearly) always done in the past. And I have tried other selling options (worst performer, alternate best then worst, etc), but selling the best works best.

Although my pc is not yet up to running errands for me, it is beginning to show some human characteristics. I've got it trained to send me an SMS message when it finds a target buying opportunity. Maybe it's developing a human psychological hangup against losers that I am unaware of .............

Alicia

Jegersmart

aliciathyme wrote: »

I use a simple trend formula:
Suppose you have 21 share prices in an Excel spreadsheet column from cell A1 down to A21
Insert this formula in cell B21 to give the current trend of of the last 20 values:

B21=((TREND(A1:A21,,{21})/TREND(A1:A21,,{20}))-1)*20

(that's very funny! good job there's a preview posting option! replace the "you are wonderful" image with a colon!)

Then you have a target buy if B21>20% AND today's price < yesterday's (ie A21<A20)

I hope this clarifies the trend issue.



I know what you mean, but have to repeat that I am not using my personal judgement, simply relying on the number crunching to do today what it has (nearly) always done in the past. And I have tried other selling options (worst performer, alternate best then worst, etc), but selling the best works best.

Although my pc is not yet up to running errands for me, it is beginning to show some human characteristics. I've got it trained to send me an SMS message when it finds a target buying opportunity. Maybe it's developing a human psychological hangup against losers that I am unaware of .............

Alicia

Yes, I get the trend calculation.

You don't use closing prices then? So you run something "mid-afternoon" using the real-time price at that time?

Another question: so what happens in this scenario:

RBS shows up as having gained 22.4% in the last month with a 1% loss on 1st December - this should be a buy signal, yes?
On 3rd December it shows up as having gained 22.7% over the last month with a 0.4% loss on the day so that is another buy signal?
On 6th December it shows up as having gained 25.8% in the last month with a 0.1% loss on that day - so that creates another buy signal?
I guess this could conceivably go on for some time? In that case, do you ignore further buy signals using certain parameters or do you repeatedly sell something else and buy multiple RBS positions?

J

aliciathyme

Jegersmart wrote: »

So you run something "mid-afternoon" using the real-time price at that time?

Yes, my software collects the latest prices for the FTSE100 constituents mid afternoon, adds them to the database, applies the selection criteria and posts me the result (usually "No Buy"). The timing allows me to react before the market closes.
Having said that, my modelling shows that there is little difference if I wait another day before trading (sometimes better, sometimes worse, but not significantly different over a 5-year period)

.............I guess this could conceivably go on for some time? In that case, do you ignore further buy signals using certain parameters or do you repeatedly sell something else and buy multiple RBS positions?

It's possible in theory, but has never happened in practice, the shortest interval was 20 days (WOS in November 2008).
As I have explained before, the trend must fall below 20% and then regain 20% for a recurrence to occur.


Alicia

jamesmorgan

Alicia,
thanks for you additional clarification on your approach. When traders use the term trend, they can mean lots of different things. Examples include shares that have higher highs/lows, shares that are consistently above their moving average, shares that have RSI peaks above a certain level etc. I have not seen the use of the TREND function in Excel to determine trends in shares. That doesn't mean it is a bad approach - just unusual. For those who are not familiar with the function, it essentially tries to find a linear line of best fit to a recent set of results. Your selection process finds shares where this line has a gradient above a certain level. The optimum share is likely to be one that has a consistent rise each day (ie it already has a linear profile). Strictly speaking this would be a linear profile on a logarithmic chart, but over 20 days the differences are unlikely to be significant. So a share that rises about 23% over 20 days in equal steps each day would meet your criteria. However, a share that rose 23% over 3 days but then had a shallower rise over the next 17 days would not meet your criteria (even though the total rise could be much higher). Finally a share that rose steadily for 19 days, but then fell sharply on the 20th day would still meet your criteria (as the line of best fit tends to ignore one off days). I'm sure you already understand all of this, but I'm mainly summarising for others.

Initially you described your approach as 'sell-to-buy'. You could equally, however, turn this around and say your sell criteria is to sell as soon as the share rises by 5%, but only if there is another share that meets your buy criteria on that day. Depending on your share selection pool (and you have limited it to FTSE100) it is theoretically possible to have buy criteria every day. In that scenario you always sell as soon as a share rises by 5%.

So very simplistically your share strategy could be described as find shares rising steadily, sell once rise by more than 5% or fall by 40%. This type of strategy needs a win ratio of more than 88% to consistently make money. In practice you can make money with lower win ratios as you have restricted your share pool to 100 shares, so you often hold onto shares longer than planned as there are no new shares to buy.

In your initial post, you described your strategy as low risk. It has undoubtedly been very successful for you, but there are a number of areas that could concern me about risk.

a) You started with 3 shares. If you had been unlucky and these shares never rose by 5% (or fell by 40%) you would still be stuck with these 3 shares today. Personally I would have started with a higher number. However, this is a historic risk for you and only relevant for someone trying to mirror your strategy.

b) Stop loss of 40%. This appears high in comparison to your sell target. In a down market, all shares can drop 40% so you would lose 40% of your entire portfolio. In is not unlikely that this could be followed by a bear market rally (when you purchase new shares) followed by another 40% drop. You could soon find most of the value of your portfolio wiped out.

c) Current portfolio size is around 15 shares. This is not high risk, but you should note that as the number of shares you hold increases the more your returns will trend towards the underlying index. You have made some exceptional returns in the early years (around 90% in 2009) but with 15 FTSE100 shares currently held that is unlikely to continue. In particular you run the risk that there are insufficient new buy signals to ever trade out of the shares you currently hold.

However, don't treat these comments as overly critical on your approach. You have some very positive aspects to your trading style - it is scientific, based on back testing and attempts to remove all emotion from when to buy/sell.

As a side note what was the motivation behind your initial post - are you looking for feedback to try to improve your approach?