Weighting of Holdings in Portfolio

[Packer16]

How do you weight the stocks is your portfolio any why?  I have value weight my positions based upon discount from intrinsic value and losely use the Kelly Formula to set maximum percentage allocation.  I will re-balance if stock rallies.

 

Packer

[sigis]

Very interesting. I am a big fan of Kelly's formula but always thought that it is too hard to apply it to stock investing because of uncertainty of the estimates. Is it a 'real' Kelly or is it just loosely based in a sense that higher conviction/probability ideas get higher weights? If it is real Kelly, do you use fractional Kelly? You are probably aware that over allocating is much more dangerous than under allocating. How do you decide when to rebalance? For example,  if your target allocation is 10%, do you rebalance at 11%? 15%?

[Packer16]

I use it as a maximum allocation.  So for instance, if a buy a stock and it appreciates but I think it still has a ways to go in terms of upside I would sell when the weighting goes above the Kelly fraction.

 

Packer

[gary17]

Interesting... So what variables do you use to come up with the probability figure in Kelly's formula.

 

And what if your portfolio consists of different securities... Do you look at each security independently Or relatively.... 

 

For me I weight them based on what I think I understand the most... I think I can see the story behind why banks are cheap and the steps needed to realize the value so I have a much greater concentration than telcos

 

Thanks

[SharperDingaan]

We use Kelly as the starting point, at initiation, then discontinue its use.

 

We see the highest risk as being at initiation; a high conviction, leveraged punt, that could fall 30% the day after we buy it. The lower end of our uncertainty sets the initial size, the upper end sets the limit if we have to average down. Around 25% appreciation we sell down to remove the leverage, thereafter its a rolling quarterly 6-9 month forecast.

 

Averaging down is a very dangerous game, when leveraged; the upper end Kelly forces selling.

We focus on risk; so no leverage unless its tactical, & removal as soon as practical.

Once we 'know' the security, we usually find our forecast picks up the seasonality & add a few points.

 

It means you cant own the whole world, & that you have to be the guru/analyst on each stock you own; therefore you need the same CFA, & other technical certs, that most sell/buy side analysts have. Your advantage is few stocks, & you do not have to toe your employers line.

 

Your disadvantage is that you are prone to sipping cool-aid; usually not a problem if you keep a lot of knuckle-draggers in your orbit!

 

SD

 

 

[SharperDingaan]

We should add that each stock is stand-alone idea, independent of everything else in the portfolio, and that the portfolio itself is a stand-alone asset.

 

Example:

 

ABC may be a new addition. It is Kelly sized, leveraged, independent of the portfolio, & a bet on European recovery. Most likely a very large bank as any kind of recovery will have to pass through the banking system.

 

DEF may be an older addition. It has been sold down to remove the leverage, but its too early to reduce the position any further. 

GHI may be a maturing addition. Leveraged to remove our capital investment & fund a FI instrument, & ideally a dividend payer entirely funded with house money.

JKL may be a series of FI instruments; convertibles, laddered T-Bills etc.

MNO may be a leveraged portfolio overlay; options, special situations, etc. where we have high conviction.

PQR may be the cash/margin you have. + 25% through -40% plus, depending on circumstance.

 

Total of 6 investments, with only 1 being Kelly sized .... & then only temporarily.

ABC investments become GHI investments over time, & CF increases semi-annually as long as we continue to invest.

 

SD

[gary17]

SD thanks for sharing this - I have a similar 'system' in my head but yours is clearly very organized, well thought through and probably based on years of experience...

 

I don't have JKL or MNO - just stocks -

 

I'm not sure how relevant this is, but I have a list of criteria that I came up with myself and assigned some weight to each factor.  I then give a score of 0 to 10 for each company I hold. The final score will have an influence on my allocation , I guess in terms of 'confidence' in the pick, and so far I have tend to not include the upside in the calculation as I'm more concerned about downside. 

Untitled.tiff

[wachtwoord]

I don't think Kelly applies as this requires bets to be independent. Every two investments in the world are likely correlated and most are correlated quite strongly.

 

Further, it's really hard to estimate an accurate win/lose chance for an investment. If using this system makes you feel safer I guess it's okay but it's all a bit arbitrary to me.

[Palantir]

Core-satellite. I take concentrated positions in larger firms like BRK, GOOG, MSFT, and small positions in speculative microcaps. Granted, even my small positions are like 3-5%.

[Hielko]

I'm not sure how sensible it is to use Kelly when:

 

1. Outcomes are not binary (it's not like you can see it's either +50% or -10%, range of outcomes is bigger)

2. Not all bets have the same timeframe, and usually the timeframes are unknown as well

3. Some bets are more correlated than others. Kelly assumes independent bets.

[Packer16]

Thanks for sharing SD.  That is a great way to us the Kelly.  Even though the events are not totally independent I use the Kelly as tool to reduce risk by providing me with some indications of how much to sell to prevent me from over allocation when a stock appreciates in value.  I use conservative win/loss ratios (typically 50%).  So I will give you an example.  Right now I own and have liked AIQ.  At this point, I estimate its upside in 116%.  So putting this in the Kelly gives you an allocation of (.5*(2.16 + 1)-1)/2.16 = 27%.  So when AIQ approaches this percentage I will sell and invest more in one of my most undervalued names.  I use it as guideline which allows me to concentrate but puts a limit on how much in one security to reduce the chance of risk of ruin.

 

Packer

[SharperDingaan]

Couple of add-ons:

 

Re Slave to Kelly: Kelly is just a principle applied in a particular way. Nothing prevents you from applying that principle a little differently, according to situation; as many here do.

 

Re Concentration: For practical purposes, most concentrated positions are actually independent of each other. You invest your whole financial services sector allocation in 1 bank, not 5; & it is in either Europe, Asia, or NA. Your next allocation is in some other sector, in some other geographic location. With only 3-4 stocks & only 3-4 locations, co-variance is pretty mild.

 

Re Time Horizon: Strategies change as time & investment thesis advance; they do not stay constant. Todays dividend paying equity investment can easily become tomorrows synthetic FI, with all accumulated house money removed via margin.

 

SD

[jay21]

I think understanding the core concepts of Kelly is helpful, but I find its use impractical because of the subjective inputs.

 

The way I have been investing is I think about the position and its risks vs upside.  Then I bucket it into a general 5% 10% or 20% "optimal" bucket.  I generally buy in around 50% to 75% of the optimal bucket in order to leave room to average down.  The idea can move into different buckets if something changes (e.g. my knowledge, price, company event, etc.). 

 

I also try to limit the amount of industry overlap I have.  For example, I think banks are very cheap and a fair amount fall into the 20% bucket, but I do not own 4 or 5 banks.

[constructive]

I regularly rank my top 20 ideas. I've assigned each rank a weight from 12% down to 3%. I compare those weights to my portfolio and if they are far off I think about adjusting them.

 

In the right circumstance I'd like to put 20%, 25%, 30% into a single idea, but generally I don't have that much more certainty in my top idea compared to my other ideas.

[sigis]

I think applying Kelly to avoid over exposure makes sense, this thread definitely got me thinking about what I should be doing on my portfolio.

 

[gary17]

 

SD

Don't take this the wrong way, but I'm wondering if you mind sharing how well has this concentration approach worked for your portfolio in the past? I believe in concentration, but have a difficult time seeing just picking one bank.... although anyone who concentrated in the Canadian banks the last decade probably did well

 

Couple of add-ons:

 

Re Slave to Kelly: Kelly is just a principle applied in a particular way. Nothing prevents you from applying that principle a little differently, according to situation; as many here do.

 

Re Concentration: For practical purposes, most concentrated positions are actually independent of each other. You invest your whole financial services sector allocation in 1 bank, not 5; & it is in either Europe, Asia, or NA. Your next allocation is in some other sector, in some other geographic location. With only 3-4 stocks & only 3-4 locations, co-variance is pretty mild.

 

Re Time Horizon: Strategies change as time & investment thesis advance; they do not stay constant. Todays dividend paying equity investment can easily become tomorrows synthetic FI, with all accumulated house money removed via margin.

 

SD

[Hielko]

The kelly criterion gives us a max betting limit, but it doesn't necessarily mean that we should have positions close to this limit. His formula is all about maximizing the growth rate of a sequence of bets, but a stock portfolio is not a sequence of bets: you are betting in parallel! If you are betting in parallel you want to make your bets a small as possible if you can find enough with a good risk/reward ratio.

[Packer16]

That is the rub - there are not that may stocks out there with good risk/reward.  It is sort of counterintuitive that at market bottoms you can be diversified because you can find a lot of securities with high reward/risk ratios but as markets advance the number of securities declines and using the Kelly at least provides some guidance on how concentrated you can get before you get into trouble from overbetting.

 

Packer

[SharperDingaan]

.... I'm wondering if you mind sharing how well has this concentration approach worked for your portfolio in the past?

 

We have extreme ROI volatility, worsened by few equities & the use of a max $ investment in equities. Rolling 10yr compound returns of 25-30%, but in any given year we can range from -35% to +140% (43% 2013 YTD). We pick well, but are not good at assessing how long an idea may take  to come to maturity, hence our focus on risk.

 

We buy cheap punch cards & hold forever, which can be costly. ie: we hold a big position in PD at an ultra-low cost base, entirely funded with house money; but year over year it has been a terrible investment (& per convention, should therefore be sold). Totally misses that we are already up 250%+, & when we do eventually sell ... it will be a 12 bagger, with a div yield in the 20-30%/yr range, with proceeds more than enough to buy a large apartment in Manhattan - if we so desire. But in the meantime ... put up with negative ROI.

 

Over time our volatility (& extremes) have declined. In our early days of 100% equity we relied on small size, options, & cash inflow to cover our risk. Option use changed to hedging as we could afford the actual common, & dropped away entirely as positions got big enough to warrant hedging by other methods instead.

 

We hold FI to park $ for distribution, & hedge our positions. If we have to average down, the incremental investment is usually no more than what we have already recovered of our original investment, & invested in that FI. With growing size we add distressed debs to offset the portfolio drag; but the higher FI weighting overall lowers compound ROI & reduces volatility.

 

You could not do this as an institution, & most would not do this as their retirement strategy. But if you think that you might want to buy a business at some future point; the equity build, & the business/industry experience that comes with it, is very useful. Our Cdn partnership was originally created to fund a brewery  ;)

 

Hopefully the patrone also gets to be a Master Brewer!

 

SD

 

 

 

 

 

 

 

 

[gary17]

Thanks SD.

I have been learning a great deal from thoughts shared by you and many others (Packer, Eric, cardboard, sanjeev, etc)

 

I've just started and have a lot of interest in this.

 

Now I have also just been offered to be a partner in the engineering firm I work for which would mean I need to sell some stocks for this and borrow money. Interesting you mentioned using the investing experience and $ to fund business.

 

The partnership will sell for 3x book and earn about 12% a after tax (profits paid out as wage bonus not dividend) which seems attractive , but definitely not as attractive as some of the returns I have seen here ...

 

But it offers great cash flow security I suppose. 

 

Any thought / advice would be greatly appreciated.

 

[SharperDingaan]

Check your messages .

[gary17]

Thanks SD!! =) 

You seem to know this kind of thing well - will take your advice into great consideration.

 

 

[Packer16]

In terms of weighting I adjust to undervaluation level.  As they approach fair value I sell them.  I guess I am like Kraven anything in my store is for sale at a reasonable price.  In terms of volatility it is still there.  I was down 51% in 2008 but up 109% in 2009 and YTD up 140% but who knows how I will end up.  I don't go on margin so I start and end unlevered and since most of my finds are in IRAs, I am not sensitive to taxes.  I guess I just live with volatility because it is there.  I do notice if stock has run up and is still undervalued it has a larger impact on the returns.  If you have any ideas to reduce volatility and not sacrafice upside, I would like to know.

 

Packer

[SharperDingaan]

If you have any ideas to reduce volatility and not sacrafice upside, I would like to know.

 

This is the main reason we hold onto our low cost bases; if the stock goes down tomorrow we can afford not to sweat it.

Problem is that the lower the cost base, the poorer the price signal - so we're also prone to holding when we really shouldn't be.

 

The cure is judgement.

 

SD