Portfolio Construction

@Sweep the Leg

But you probably spend time thinking how much to buy and whether to sell other investments. You may not think about diversification, but you are thinking about portfolio construction.


I have spent WAY more time on all the steps that come before doing the Black-Litterman calculations (estimating covairances, etc) than on the calculations for BL. BL is rather straightforward, but (when I use it) I throw out all the mumbo jumbo about getting market cap weights for each security. I simply use a 100% allocation to some benchmark I’m comfortable with (100% stocks or 60/40 stocks/bonds).

jmh, I had included the estimating covariances part as part of BL calculations, so I think we’re on the same page here. I’ve never actually done a BL optimization - I’ve just read about how it works and I like the methodology (term used correctly here) that justifies it.

If you use a benchmark, it sounds like you do something more like a Treynor-Black type of model, which I’ve read turns out to be a special case of a BL model. When I talk about trying to develop a simplified heuristic model that is similar to BL, I think I’m ultimately headed toward a TB model with some assumptions about correlations that don’t require so many covariance estimations.

@jmh - The only thing I take into account are capital constraints, that is to say, opportunity cost. If we’re being really picky then I suppose that counts as portfolio construction. Things like managing correlations and risk though? No, I could hardly care less. I barely even rebalance.

I presume you mean estimate correlation, volatility and weighting which give you covariance matrix in BL.

Spending so much time on BL recently, am in the middle of building a model to encompass BL and probablistic model together. The drawback of BL is that it penalise high vol asset outright where in reality, that can potentially be your return engine.

At the moment, we use risk weighted portfolio. Not risk partiy, just ensure marginal contribution to risk is appropraite for their corresponding physical weight. It is ok but in both back-test and simulation, the performance doesn’t change much after taking fees and tax into consideration. By the way, i’m working with a multi-asset portfolio.

Frank’s approach works great for PA but is a nightmare for insto, the risk is uncontrollable, unmanageable. In addition to that, I run the risk of losing job to recommend the board to put money with a person that charges fees and holds cash…

Pshhh… Muggles,

He is absolutely right…this is why I have no real interest in managing institutional money…imagine spending the day working on “building a model to encompass BL and probablistic model together”…might as well just watch the tickers roll by…

Improving performance by 10 bps because you’ve spent a year building a better model isn’t a good payoff when you are managing 500k or 1 million.

It can be an awesome payoff when you are managing $500 MM and up.

It’s a context dependent thing.

Bchad…who are some of the investing greats you imulate?

This is why managing institutional money interesting. Additionally, the investment universe is much bigger when you manage large sum of money.

Depends. I suppose if you’re referring to different types of investments - certain derivatives, structured products, PE - being available to instiutional investors then I agree.

But, too much money is a horrible thing for mutual funds. Your investment universe shrinks the larger you get.

+1. I don’t pay any attention to diversification. I know my companies and I put relatively big %s of my portfolio into them. At last glance, I believe 3 stocks made up about 60% of my portfolio.

It’s ironic, when I put on a smaller positon to “mitigate risk” it winds up being a shit*y performer that I shouldn’t have bought in the first place. The 2 smallest positions, by both cost and market value, have also been the worst performers.

Market correlations have been unusualy high over the last 5-10 years. If you’ve run concentrated portfolios with high beta and made decent calls, it’s understandable why you might think diversification is a junk idea (if you have something like a 15% hurdle rate, you are much more likely going to choose high beta stocks).

Meanwhile people who ran concentrated portfolios and made bad calls (and would have benefited from diversification) have been washed out of the market, so they’re not here to say “well, I was concentrated and it didn’t work out for me.” Those who did diversify haven’t had as large a return as those who concentrated, and so, while they are still here, maybe aren’t so excited about their rankings in the pecking order (diversified investors seldom are, because there are always some concentrated outliers performing better, even if they aren’t always the same investors getting those results).

Now, if you have put on a risk-mitigation smaller position in such an environment, it will naturally be a crappy performer at least in relative terms… that’s what diversification is about… when one set of assets does well, the other’s not supposed to do well at the same time. That’s pretty crucial - the “at the same time” part. Because diversification is *NOT* about finding something that is expected to go down if something else goes up… that’s hedging. Diversification *never* justifies investing in something that has a negative expected excess return. All that it means is that 1) long term expectations are positive, and 2) the *timing* of those excess returns aren’t synchronized.

The idea for diversified portfolios is that, over a wide range of market conditions, the diversified portfolio doesn’t go through the wild gyrations, and therefore can compound at a faster rate. The catch with compounding is that it can take a loong time to see the benefits. So diversification is more important if you have longer holding periods. For short holding periods, position sizing is more important for risk control.

If you are a value trader, diversification is probably not a big tool in your arsenal, although position size limits are. If you are a value investor, however, you’re going to be having long holding periods, and therefore diversification is more important.

if you’re a long term holder, diversification is less important…diversification is a concept poorly understood…its not meant to smooth portfolio returns but to hedge against ignorance…there is no need to talk about beta, alpha or whatever…well, unless your PM requires you do…

Oh. Ok. I stand corrected.

bchad…you going to watch the new Mad Men when it comes out…red is back, now that is one hot classy babe…

I found this comment to be satisfactory. Would read again.

I think there’s something to be said for parts of what Frank says. I’m not a fan of benchmark-relative portfolio construction. I care much more about what the returns are than what the alpha is. Pension funds/endowments care about beta/alpha, which is why mutual fund companies offer benchmark-relative products.

I also agree that diversification is a poorly understood concept. There are quantitative measures of diversification (like the effective number of bets). However, if you construct a portfolio to maximize diversification, you would likely undeperform a mean-variance optimization. That’s because people don’t prefer diversification for the sake of diversification. They prefer diversification because it reduces portfolio variance or downside risk. So I would rather focus on the effects from diversification.

I don’t get it either. Who cares about volatility and risk and beta all that boring stuff. Isn’t the point of investing long term capital appreciation? At least that’s how I invest. If you cannot invest in concentrated positions due to lack of insight, that’s fine, but uhm…if you can…why would you waste time with this stuff.

i understand why these concepts were created, but I think for practitioners to even consider it when making an investment decision is sorta like gaming…

A. I think this stock is great, good management etc…

B. No, we can’t. its too high a beta and too volatile…

A. but we know given the capital structure and industry dynamics plus our margin of safety the likelihood of permanent loss is extremely low

B. I’m aware of this. but it won’t look good. ppl don’t like to invest in stocks that move up and down, they must move along with the market, but nothing more.

A.This fckin’ job is stupid.

B. It is, but we get paid to be mediocre, i keep my job, and you might keep yours. we can tag team on a HCB afterwards…but for now, build me a model that shows me how to best track the index…

^ Good grief! That python in your pants must be huge. Do you need to tote it around in a wheelbarrow? If only the HCB’s knew. Have you ever thought of a s3x tape leaked on the internet?