Low-Risk investment strategy ?

grey_gym_sock

i imagine that different selling strategies work well with different buying strategies. your buying strategy is 100% technical analysis. a lot of the criticism of the selling strategy may have been more relevant to ppl relying on fundamental analysis.

you would expect short holding periods when using technical analysis. so selling winners too soon may not be such a big problem.

1 thing that strikes me as more questionable is that you could end up accidentally holding a share for a long time when it's failed to move much (up or down). if the buying signal indicates an uptrend, and the uptrend has broken down, shouldn't you be selling (without waiting for a 40% fall from your entry point)? what does the data show: have longer-term holdings performed worse?

i should mention that i know nothing about technical analysis. i don't decide what to buy or sell based on technical analysis, though i may make amateurish attempts to time entry or exit based on short-term trends.

bowlhead99

aliciathyme wrote: »

Out of curiosity I have re-run my model using just the trend function on the same data as previously so you could try to emulate it if you felt inclined.

It is clear that just using the trend function in isolation still gives great results, and maybe shows that the clue is perhaps not in the buying strategy, but in the selling strategy

Maybe this will help others to try out the Sell-to-buy method.?

In Method A, you selected good stocks for a 400% return, using a strategy which found entry points using trend analysis and other factors. It also had a selling strategy which was not based on trend analysis or fundamentals like financial reports and news releases, but just on an arbitrary coin toss.

Now, you've tried a different stock picking strategy which uses trend analysis but not those other proprietary factors from Method A, so does not select the stocks and time the entry points as efficiently. It still uses the same arbitrary selling strategy. The return went from 400%+ to 200+.

I can't see, other than blind faith, why you are convinced the selling strategy is producing these good positive returns.

Think of it this way. You start with 3 stocks at 3k for 9k invested. Over 6 months, the portfolio increases in value to 12k. Stock A increases in value to 4.7k, B increases to 4.5k, C declines to 2.8k. At this exact moment, you want to go and buy a fourth stock and need 3k. Broker fees are 0.00595 per sale and pretty inconsequential to a strategy which generates huge returns, we'll ignore.

You could sell 3k of A and leave 1.7k of A, 4.5 B, 2.8 C, 3k of D
Or sell 4.7k of A and leave 1.7k cash, 4.5 B, 2.8 C, 3 D
Or sell 2.8k of C and 0.2 of B, leave 4.7 A, 4.3 B, 3 D
Or sell 3 of B, leave 4.7 A, 1.5 B, 2.8 C, 3 D
Or sell 1.5A and 1.5B, leave 3.2 A, 3 B, 2.8 C, 3D. Possibilities endless.

Lets say you do the first one and sell the largest holding leaving 1.7k of A, 4.5 B, 2.8 C, 3k of D.

In next few months until next opportunity, A stays flat, B puts on 10%, C puts on 77%, D puts on 60%.
The portfolio is 1.7A, 4.95B, 4.96C, 5 D.
To get the cash for E you can either sell the one that has stopped growing (A), the one that has drastically reduced growth (B), the one that is growing fastest (C) or the one that's growing second fastest (D).
You're going to choose D here because it has the largest absolute amount of cash in it, which will leave you with
1.7A, 4.95B, 4.96C, 2 D, 3 E.

Evaluating the portfolio which now has an average holding size of 3.32, you are now:
About 49% underweight on the zero performer but previously good A
About 49% overweight on the drastically reduced performer B
About 49% overweight on the best current performer C
About 40% underweight on the second best current performer D
About 10% underweight on an untried performer E which just hit its buy signal.

Basically the construction of this portfolio for someone who believes in trends for stockpicking is garbage.

It has just come from selling the one with the biggest pound profit showing at a point in time. When you need to consider exiting a stock you could try and take some cues from the market trends (rank them in likelihood of what the trend says they'll do next according to your massive databank - technical analysis) or evaluate the underlying financial indicators (trading performance, press releases, news stories affecting their sector - fundamentals). Instead you are just selling one that holds the most pound profit with no regard for growth rates, current market trends or fundamentals of why the trends are happening.

aliciathyme

grey_gym_sock wrote: »

1 thing that strikes me as more questionable is that you could end up accidentally holding a share for a long time when it's failed to move much (up or down). if the buying signal indicates an uptrend, and the uptrend has broken down, shouldn't you be selling (without waiting for a 40% fall from your entry point)? what does the data show: have longer-term holdings performed worse?

I have looked into the issue you raise on profitability of longer term holdings.
A graph of of returns vs duration is fairly flat, showing gains and losses at any duration, although as you suggest, both are higher for longer terms, there is no particular correllation. If anything, the trend shows slightly better return with longer holding, but nothing to get excited about.
The median duration of all sold holdings is 51 days and the median return is 16.1%. The maximum return was 97%, achieved over a duration of 179 days.

I have also run my portfolio data (to 14 Dec 2012) through the model using a series of reducing trend rates at which to sell, from 20% per month reducing in 2% steps to a loss of 20% per month.

The results are as follows (selling when trend falling below the rates shown):

T 20% wr 84% MR 13% md 39 NP 287.6%
T 18% wr 84% MR 13% md 36 NP 278.2%
T 16% wr 85% MR 12% md 37 NP 261.7%
T 14% wr 85% MR 12% md 37 NP 258.0%
T 12% wr 84% MR 12% md 38 NP 237.1%
T 10% wr 85% MR 10% md 38 NP 228.8%
T 8% wr 85% MR 9% md 36 NP 196.5%
T 6% wr 85% MR 10% md 38 NP 212.1%
T 4% wr 85% MR 10% md 38 NP 218.4%
T 2% wr 84% MR 12% md 38 NP 204.2%
T 0% wr 84% MR 12% md 38 NP 213.4%
T -2% wr 84% MR 11% md 39 NP 225.7%
T -4% wr 85% MR 12% md 42 NP 234.5%
T -6% wr 84% MR 12% md 40 NP 231.6%
T -8% wr 85% MR 11% md 40 NP 250.4%
T -10% wr 84% MR 12% md 45 NP 249.9%
T -12% wr 84% MR 12% md 43 NP 247.5%
T -14% wr 82% MR 12% md 45 NP 230.9%
T -16% wr 83% MR 14% md 51 NP 304.1%
T -18% wr 81% MR 15% md 51 NP 317.7%
T -20% wr 81% MR 16% md 51 NP 316.2%

Where T = selling when trend falls below this rate per month
wr = win ratio
MR = mean return
md = mean duration
NP = Net Profit

By comparison, the respective results of the existing portfolio (where selling trends are not used) are:

wr 81% MR 16% md 51 NP 343.3%

So, it would appear that there is no advantage in selling early as the rising trend slows down (or even reverses)

Alicia

aliciathyme

bowlhead99 wrote: »

...........I can't see, other than blind faith, why you are convinced the selling strategy is producing these good positive returns............Basically the construction of this portfolio for someone who believes in trends for stockpicking is garbage. It has just come from selling the one with the biggest pound profit showing at a point in time.

I was trained as a scientist.
We work with ideas, theories and experiment.
When we get consistent results we believe them .......... until someone disproves them or comes up with a better theory.

We observe and take note, commenting only when we can provide helpful constructive criticism.


Alicia

aliciathyme

bowlhead99 wrote: »

Think of it this way. You start with 3 stocks at 3k for 9k invested. Over 6 months, the portfolio increases in value to 12k. Stock A increases in value to 4.7k, B increases to 4.5k, C declines to 2.8k. At this exact moment, you want to go and buy a fourth stock and need 3k. Broker fees are 0.00595 per sale and pretty inconsequential to a strategy which generates huge returns, we'll ignore.

You could sell 3k of A and leave 1.7k of A, 4.5 B, 2.8 C, 3k of D
Or sell 4.7k of A and leave 1.7k cash, 4.5 B, 2.8 C, 3 D
Or sell 2.8k of C and 0.2 of B, leave 4.7 A, 4.3 B, 3 D
Or sell 3 of B, leave 4.7 A, 1.5 B, 2.8 C, 3 D
Or sell 1.5A and 1.5B, leave 3.2 A, 3 B, 2.8 C, 3D. Possibilities endless.

Lets say you do the first one and sell the largest holding leaving 1.7k of A, 4.5 B, 2.8 C, 3k of D.

In next few months until next opportunity, A stays flat, B puts on 10%, C puts on 77%, D puts on 60%.
The portfolio is 1.7A, 4.95B, 4.96C, 5 D.
To get the cash for E you can either sell the one that has stopped growing (A), the one that has drastically reduced growth (B), the one that is growing fastest (C) or the one that's growing second fastest (D).
You're going to choose D here because it has the largest absolute amount of cash in it, which will leave you with
1.7A, 4.95B, 4.96C, 2 D, 3 E.

Evaluating the portfolio which now has an average holding size of 3.32, you are now:
About 49% underweight on the zero performer but previously good A
About 49% overweight on the drastically reduced performer B
About 49% overweight on the best current performer C
About 40% underweight on the second best current performer D
About 10% underweight on an untried performer E which just hit its buy signal.

Lets take your example.
My Sell-to-Buy strategy has only one option:
First sell ALL 4.7k of A to buy 3k of D and bank 1.7k cash.
Next time sell ALL 5k of D to buy 3k of E and take another 2k cash, bank now has 3.7k, enough to buy 3K of F without selling anything.
So as not to introduce any new valuations, assume all remain level:
Portfolio is now 4.95B, 4.96C, 3E, 3F, 0.7Cash

The portfolio has grown from 9k with 3 holdings to 16.61k with 4 sound holdings (84% return) or an average holding value of 3.977 and 0.7k cash.


Alicia

grey_gym_sock

aliciathyme wrote: »

I was trained as a scientist.
We work with ideas, theories and experiment.
When we get consistent results we believe them .......... until someone disproves them or comes up with a better theory.

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.

another point is that markets go through different phases. a strategy that works in 1 phase may not work in another. and it may not be obvious when we are in a new phase. so you may have a strategy that's worked in recent phase(s) of the market. which is not quite the same thing as saying you've been lucky. or is it?

bowlhead99

aliciathyme wrote: »

Lets take your example.
My Sell-to-Buy strategy has only one option:
First sell ALL 4.7k of A to buy 3k of D and bank 1.7k cash.
Next time sell ALL 5k of D to buy 3k of E and take another 2k cash, bank now has 3.7k, enough to buy 3K of F without selling anything.
So as not to introduce any new valuations, assume all remain level:
Portfolio is now 4.95B, 4.96C, 3E, 3F, 0.7Cash

The portfolio has grown from 9k with 3 holdings to 16.61k with 4 sound holdings (84% return) or an average holding value of 3.977 and 0.7k cash.

The portfolio has gone up in value because the growth rates of all the shares have been pretty great. The decision to buy those particular shares A to D from a universe of A to Z has been very succesful and so the technique for picking A to D out of A to Z must have been great, and the highly technical trend analysis to support these 'buy now' selections seems to have paid off.

But we're now looking at the rationale for a sell decision - looking at the choice you made to sell D and buy E and F and leave 0.7k cash:

You had 3 shares B, C or D with large overall positive returns which were well overweight in the portfolio (4.95, 4.96, 5.00) and whether you'd sold B,C or D you'd have more cash than you actually need to buy E and F; the only difference is that the spare cash you'd be left with is either 0.65, 0.66 or 0.7 which is not a usefully different amount of cash in a nearly 17k portfolio.

So the goal should be to leave the portfolio in the best position for future growth. Should you have slightly less in equities and slightly more in cash? Or slightly more in equities and slightly less in cash? You don't specifically need the cash for anything right now and the 0.05 cash/equities difference is not really meaningful in terms of prospects, so the remaining choice is which equities should be kept to drive the portfolio higher as we go foward.

All of the shares have produced very similar lifetime growth at 65-67% and are all worth about the same so we need to come up with a rational basis for the decision. Their recent performance has been very different and as the use of trends in price to buy the stocks in the first place was very successful, this seems like one place to start (presuming you're not going to actually review fundamentals, financial results, news releases etc).

You have share B which had great growth previously but dropped back to only 10% in recent months. You have share C which had initially been flat to marginally down but is now the fastest growing performer at 77% in recent months. You have share D which has done almost as well, 60% in recent months- it's not the fastest growing thing you have, but at least it doesn't seem like it's gone ex-growth like B has. All of these shares are worth between 4.95 and 5.00 if sold today (i.e. materially the same), and the cash needed could be obtained from any one of them, or by selling all in equal proportions to lock in gains from all and rebalance the portfolio.

The choice therefore, for someone who believes in trends, to keep the ex-growth share B, and keep the fastest growing share C, and dispose of the marginally less fast growing but still great performing D, is a little odd. It's been selected, effectively, almost at random, picked out by your system of "whatever's the highest after time period 2". You didn't even own it in time period 1, although you do have data in your database about its two-period performance which could be brought into the assessment if wanted - it might be better or worse or equal to either B or C over that timescale. But for the one period you held it, it went up second-best. Your system says sell, and you put blind faith in that system to do so.

I think many would agree that if they are making reasoned buy decisions on trends or fundamentals they should not make sell decisions on blind faith. The system told you to sell D. D was not the fastest or slowest grower in Period 2 and the system did not look at how it performed overall over both periods because it hadn't been owned in Period 1. This strategy is only likely to be neutral and not earnings enhancing.

aliciathyme wrote: »

I was trained as a scientist.
We work with ideas, theories and experiment.
When we get consistent results we believe them .......... until someone disproves them or comes up with a better theory.

Your data has validated your theory that the algorithm used to pick stocks based on their price movements allowed you to buy them at the right time during a certain type of market.

You don't have a theory about when to sell any of the stocks other than 'whenever you need 3k'. On which stock to sell, or how much of it, your answer is not driven by comparing the shares' most recent period's performance, or looking at each share's peformance over an equal time period - and the 'how much to sell' is always 100%.

So Sell-to-Buy is not an investment theory about when to sell, what to sell and how much of it to sell based on reviewable market data at the time of sale. It is a betting strategy. At a card game requiring skill, Sell-to-Buy is telling you how much and when to bet rather than how to play the hand.

Jegersmart

One could argue that "sell to buy" is something that instills discipline in terms of taking profit which is very important in my experience, gains on paper cannot be spent and are not real until the position is closed. Whether one closes the entire position or enough to enter another position of a set size could perhaps be discussed, but most stocks will reach exhaustion, and unless the yield is unbelievable I prefer to exit and potentially re-enter unless something else presents itself as an opportunity. Entry AND exit is important in equities unless one just buys and holds endlessly - which at least is less hassle I suppose, and the holder does not have to take responsibility for selling too early or too late etc. - lots of people do this, and over longer term periods has lost out considerably as we go through the cyclical nature of markets.

The "stock selection" part of this process has been tested in a rising market out of a bloodbath and from that point of view I would have a problem with blindly following in a lot of market conditions, and as Alicia herself points out, you need to sell everything when the market tanks.....

It is an incomplete system, but has some interesting elements that I will be considering going forward.

imho
J

aliciathyme

bowlhead99 wrote: »

So Sell-to-Buy is not an investment theory about when to sell, what to sell and how much of it to sell based on reviewable market data at the time of sale. It is a betting strategy.

Any view of the future based on the past is a betting strategy.
That applies with your assessments of future performance of stocks.
But that does not invalidate the concept.
Mine is 81% successful over a 5-year period of ups and downs.

I have been trying to get people interested in modelling my Sell-to-Buy concept on their own data just so that I can get an idea of it's general applicability.

Alicia

Jegersmart

aliciathyme wrote: »

Any view of the future based on the past is a betting strategy.
That applies with your assessments of future performance of stocks.
But that does not invalidate the concept.
Mine is 81% successful over a 5-year period of ups and downs.

I have been trying to get people interested in modelling my Sell-to-Buy concept on their own data just so that I can get an idea of it's general applicability.

Alicia

It is a 5 year period though when it has been a bull market in equities, or a bear market rally depending on the time frame you are looking at - but the results are impressive nonetheless and I think it warrants exploration or thought. The "sell to buy" concept I have already commented on - it is very important to take profits but I don't have the time or inclination at the moment to try to do some modelling on my own history, especially when I already sometimes do the same - i.e. take profits on holdings in most profit but will usually do so in line with price targets that I set using technical analysis. Sometimes these are good, others they could be better but I would assume it would be the same with a Sell to Buy technique in that you only sell when there is another opportunity so will often sell to early or too late. Having said that, if the selection criteria for buying is sound then what better time to sell when the "next best thing" comes along?

I appreciate my comments are completely unscientific, but as I said I will be taking a look at this in more detail going forward.

J