bostonbubble.com

GenXer · Joined: 20 Feb 2009 Posts: 703

balor123: A typical long term (10 years maturity) muni bond fund pays 3.5% or so tax free. If your tax bracket comes out to 50% (think business owners, federal AND state taxes), you are making 7% interest tax free AND backed by assets and taxing authority of the state (in this case, MA). I do not think you can beat THIS deal. Short term muni bond funds pay around 1.6%, which is 3.2% tax free. Not bad, but a much lower risk. Intermediate (7 years maturity) are probably around 3.3% and will make you 6.6% tax free - you get the idea.

balor123 · Joined: 08 Mar 2008 Posts: 1204

balor123 · Joined: 08 Mar 2008 Posts: 1204

GenXer · Joined: 20 Feb 2009 Posts: 703

Time diversification has been shown to be non-existent. I'm sure you can find the results on the web, but in a nut shell, a study by Philippe Jorion it was shown that if you take a big enough dataset and assume as little as possible about the data, you will get no evidence of time diversification (i.e. that risk decreases over time), aka mean reversion. Rebalancing actually increases risk when the stocks go down because you get back into the risky assets, and the studies I've seen are so naive that I'd hold out saying that it works at all. Markets are LESS predictable over the long term (read Jorion's study for more info). Historical information is useless as a predictor, and is only useful to perform mathematical exercises ex post facto.

Correlation (as in Modern Portfoio Theory) is a single number. It is completely useless when talking about portfolios, because in reality it is a time-varying random process, and you can not use it in any shape or form to make any type of predictions, because you'll have huge errors which will make your predictions meaningless. You have to know how your correlation is distributed in time to get something useful out of it.

Munis are for people whose tax bracket is 30%-50%, and all you care about is relative safety of your principal (as in medium-term cash reserves). You are right that munis typically are not used for 'classical' portfolio construction, but many high net worth individuals (with good advisers) always have munis in their portfolios.

balor123 · Joined: 08 Mar 2008 Posts: 1204

GenXer · Joined: 20 Feb 2009 Posts: 703

I think threads have a way of running off on their own...but to bring us back to the original question...if Guest who asked it is still reading this...
Did you make up your mind? What did you decide? I'm curious as to the results of this lengthy discussion.

balor123: This can be arranged. The question is, what type of analysis would you like to apply to the data?

CR Money Adviser? The tabloid of finance. Come on, this is not a serious publication. I subscribe to Financial Planning magazine, which I think has a little higher caliber articles by some well-known authors, and I'm yet to see a financial analysis article which actually does the math right. Sorry to say that most 'analysts' use in-sample testing of their hypothesis, which coupled with a tiny dataset limited to a single market (US, mostly) can prove anything your heart desires. They cherrypick their datasets, avoiding anything nasty data points. Survivorship bias, assumed statistics, and other bad assumptions abound. They set out to prove what they believe in - pure and simple confirmation bias.

john p · Joined: 10 Mar 2006 Posts: 1820

I think within our brains, we have a democracy of voices. Sometimes certain voices dominate. Right now, the voices of caution dominate. The analytical side seems to be able to slip into any conversation because it is like Switzerland, it is neutral. Sometimes the dominant voice says "Shut up Switzerland". We do have an abundance of data to cull from and sure we can cherry pick data to try to fool the voice of caution. It is hard to have an overriding voice of reason when our economy is kind of dropping like a stone. It is kind of human to have emotions present during these events. One of the downsides of learning is that it sensitizes you to certain behaviors (behaviors that are going haywire turbulent). These sensitivities create emotions so managing all these voices in your head during a heightened state is important.

One thing I've found out about myself is that I sometimes get the emotion before I get the understanding. After thinking about stuff and going through the data, often times I realize that it was some voice in the back of my head that had the alarm bells going off and once I understood the situation, it sort of validated my emotions. I think you need to read all you can but at the same time trust your instincts and values.

balor123 · Joined: 08 Mar 2008 Posts: 1204

GenXer · Joined: 20 Feb 2009 Posts: 703

Actually, fiance yahoo has dividends and splits if you load it right. Data set is quite limited, though. I prefer to understand what's out there first - when you learn what mistakes other studies made, you can sleep better knowning you are not simply going to re-create the same mistakes.

I get a several online publications. I've gotten CR Money Adviser - unfortunately they are only good when it comes to very basic financial planning. The instant you get into portfolio design, risk, etc, they are just as clueless as most fiancial publications. Though they have some great suggestions (like, annuities are mostly unsuitable - try to find THAT one in Financial Planning magazine, paid for in part by annuity salesmen).

In fact there is a discipline called financial engineering. Financial engineering publications cover their (more often than not) incorrect assumptions with a thick coating of math which most people (myself included) would break their teeth on. So its no help, unfortunately. One breath of fresh air is Nassim Taleb, but unfortunately he is not in the business of providing solutions that can be applied by an average investor (his solutions involve some serious options trading, which is probably not going to work for most people).

I think there are some good studies out there - I really wish I had more time to do the actual math myself (maybe in the future someday), but I think at this point its enough to be able to recognize bad math and not act on it.

Some people think NY times calculator is a good 'financial engine' for assessing buy vs. rent. Some think mean-variance optimizers are the best thing since sliced bread. Others think that Monte Carlo simulations (with Gaussian statistics!) are great, and then there are those who take their random walk seriously and use this math to compute risk and to predict 'optimal' portfolio allocations. Don't waste your time. There is no predictive engine that works as of today.

balor123 · Joined: 08 Mar 2008 Posts: 1204

Yeah but good luck trying to study how commodites have done over the past 100 years. You'll have to stitch together multiple indices or extrapolate from other sources. I know these things exist becuase companies like Financial Engines use style analysis to do their portfolio optimization and they have to have a set of styles to match investments with.

I think Yahoo has dividends for stocks but I don't think they have capital gains distributions for mutual funds. I could be wrong but maybe I just need to look harder. I've had trouble in the past with some investments going back far enough. For example, an index may have been defined in 1985 but no ticker symbol existed for it until 1995. Maybe there's an open market for datasets here but I suspect that someone already provides this for a fee.

I use quantitative tools not to determine investment strategy but to reveal parts of the picture. Most of the picture is still blank and you are left to extrapolate the rest. Its too bad that a lot of the quantitative work is buried in expensive publications because you end up having to rely on these tools (like Financial Engines) just to get the benefits because you can't get the data from the papers. I wanted to find a good glide curve to use, for example, which is based on statistics but its mostly buried in expensive publications or proprietary. I eventually found it through the power of search engines though Smile

GenXer · Joined: 20 Feb 2009 Posts: 703

Well, we do know something about one commodity - gold. What we know is not very encouraging though - it basically gets you inflation, but this assumes that you do not hold gold during a downspike.

By the way, when you 'exptrapolate' statistical data, you are creating data, which is a big no-no. I do not believe honest researchers extrapolate data, since we do not know the DISTRIBUTION of these returns, hence any extrapolation will create NEW data which is simply wrong. Just a word of caution Wink

Sounds like a great topic for the presentation...the fallacy of extrapolation of data generated by a fat-tailed distribution Wink

Another by the way. Proprietary for the most part means it does not work. Why? If they are not willing to publish it, they do not want anybody to burst their bubble. All good stuff is published, because if anybody can actually SOLVE this problem (i.e. getting data to spill the beans), they can easily get a Nobel for it (and not just for a bogus theory like MPT).