Putting together quant portfolios

November 22, 2014

One of the more overlooked areas of quantitative investing is how to integrate quant strategies into a diversified portfolio. Let’s look at a few of the issues involved and some possible solutions.

I’ve discussed the big allocation decision already, i.e. how much to put into quant strategies vs bonds already. I think this decision should be primarily driven by how much risk one is willing to take. I use maximum portfolio drawdown as the primary metric to judge how much risk I’m willing to take. Here is the max drawdown table from that post.

Quant system vs max DD nov 2014

For example, to limit annual drawdowns to less than 10% and 30% quant portoflio 70% bond portfolio would be the appropriate allocation. If a 15% max DD is more suited to your risk tolerance then you can go up to a 70% quant portfolio 30% bond portfolio allocation. Of course, this is dependent on the type of quant strategies used but surprisingly it doesn’t matter as much as you may think. The XLP/XLU quant strategy is one of the most conservative strategies while TV is one of the most aggressive and best performing yet in terms of max DD there is not that big a difference. OK, now for some of the subtle details of building quant portfolios.

First, why pick more than one quant strategy? Shouldn’t you just pick the best one with respect to whatever metric you want? I would argue no mainly due to behavioral factors. Investors have a hard time with under performance no matter what the long term track record of a strategy may be. Lets take this year as an example and compare one of the best performing quant strategies over time, a value and momentum microcap stock quant portfolio, to a conservative utility value strategy. The micro cap quant strategy has returned over 22% a year since the late 60s as compared  16% a year for the utility value strategy. In 2014 the micro cap strategy is up 3% while the utility strategy is up over 29%! That kind of divergence is hard to swallow for most investors and leads to performance chasing. The best way to avoid this is to choose multiple strategies and choose strategies with high base rates. At the simplest level using more strategies increases the odds the at least one is outperforming and thus lessens the odds of making behavioral mistakes.

OK, now that we’re going to choose multiple quant strategies we need to avoid a few pitfalls. The two big ones for me are over diversification and excessive fees. The quant strategies I’ve presented on the blog usually have 25 individual stocks in a portfolio. If  you’re going to go with 3 quant strategies then we’re talking about 75 individual stocks. Besides paying way more in fees by holding too many stocks we reduce the power of the quant strategies. Two great post on this subject are, Avoid Diworsification, and You need to Dare to be Great. Here is the money pic.

return vs number of stocks for quant systems nov 2014


10-15 individual stocks gives about the best bang for the buck for any one of the individual quant systems. So lets say we go a bit crazy and choose 5 different quant systems that we’d like in our overall portfolio. We’ll choose the TV2 systems, the EDY system, the Large stock SHY system, the XLP system, and the XLU system. I’ve posted all of these systems before. See here for a list of all my quant posts. 10 stocks each. Yearly holding period and go back as far as we can, Dec 2008, for the P123 book simulation tool that I use. I’ll call this the Quant 5. Results of the overall quant portfolio are below.

the quant 5 performance nov 2014

Pretty impressive results. Over the full period all of the quant systems beat the market handily. But there have been significant periods of under performance for several of the systems. For example, over the last 3 years, the EDY systems and the XLU system have under performed the market and the other systems. The top quant system over the full period from 2008 turned out to be the large stock SHY system even though going back further to 1999 other systems have performed better. Of course, how could you know that before hand? You can’t.

Bottom line. It pays to diversify, but not too much. Not just in terms of risk adjusted returns but probably more importantly in terms of an investor being able to stick with quant investing over the long term.

Tactical asset allocation november update

November 2, 2014

It’s been a while since I’ve done updates for the various IVY portfolios. I’ve noticed that investors still have trouble implementing these portfolios so I decided to start doing the updates again. At least for a few months. Hopefully this helps investors who are trying to implement these portfolios or are just starting their research into these strategies. Also, I’ve changed the name to tactical asset allocation update because these strategies are independent of the ‘IVY’ models popularized by Meb Faber. As I mentioned in my last post you can apply these strategies to many other diversified portfolios. For now, I’ve added the Permanent Portfolio to my IVY portfolio updates. In the future, I will probably add other tactical asset allocation strategies. In particular, I will probably be adding a volatility weighting to some of these strategies. On to the November update.

Starting with the most basic portfolios, below are the November updates for the GTAA5 and the Permanent Portfolio. I keep a spreadsheet online that is update automatically.

IVY5 timing model Nov 1 2014 update

Perm Port timing update nov 1 2014

Now for the more broadly diversified GTAA13 portfolio. Online spreadsheet for this and the GTAA3 and GTAA6 portfolios.

GTAA13 timing update nov 1 2014

And finally for the more aggressive GTAA3 and GTAA6 portfolios.

GTAA6 GTAA3 Nov 1 2014 update

These portfolios signals are valid for the whole month of November. Also, I’m considering replacing the MTUM momentum ETF with a better option. My concern with MTUM is the small market cap and low liquidity. There are potentially some better options out there. I’m considering IUSG or PDP for the US momentum ETF and am even considering adding an international momentum ETF, for example PIE.

One portfolio to rule them all?

October 31, 2014

Unfortunately, portfolios are not like a certain type of ring, forged with magical powers to rule over the world, man/hobbit/elf kind, and all other types of investment portfolios. There is no one portfolio to rule them all. There are many different styles and types of portfolios with very different characteristics, asset allocations, and histories. Most importantly, the appropriate type of investment portfolio can and should change for different types of investors. Investors have different risk tolerances, goals, biases, starting points, end points, etc… In this post I want to present a summary of various investment portfolios that I track and discuss a few of their characteristics and the types of investors that they could be suited for. Also, I hope the data and discussion dissuade investors from becoming dogmatic in their portfolio choices or recommendations. All right, let’s get to it.

In May of this year I posted a summary of the portfolios I had been tracking up to that point. Since then I’ve expanded the number of portfolios that I track and added some more statistics as well. I collect all the data from varied array of free sources, e.g. Shiller, MSCI, the Federal Reserve, generous financial professionals, and my own data. I grouped the portfolios into types; equity only, diversified buy and hold, tactical asset allocation, and bond only. The data is from the 1973 to 2013 period. Within each type the portfolios are sorted by compounded annual return (CAGR), the number most investors look to first. Without further ado, here is the data.

Diversified Portfolio Summary Data Oct 31 2014

Lots of data, I know. In the equity only portfolios there are the basic SP500 index and the MSCI world stock index (ex US). I also added two of the best quant portfolios I’ve discussed on the blog many times before. The diversified buy and hold portfolios show the most common and popular portfolios. The traditional 60/40 or 70/30 US stock US bond portfolios are the most basic and maybe the most common asset allocations for US investors. They are definitely the most talked about allocations, partially because they have the longest historical record. There are several portfolios named after their namesake advocate/creator; Bernstein, El-Erian, Arnott, Swenson, and Tobias. The IVY buy and hold portfolios, and the Permanent Portfolio are also shown on the list.  I also added two diversified buy and hold portfolios created with the quant equity strategies. The tactical asset allocation section shows the various trend following, risk manage IVY portfolios discussed on the blog many times. I then show various bond portfolios and inflation mainly for comparison purposes.

Phew! Now, what to make of all this? First, notice that within a broad swath of the diversified buy and hold portfolios they are all very similar in terms of returns and risk. One thing I see all the time is investors becoming overly zealous and even dogmatic in their portfolio choices. IVY is the best or 60/40 is the best because such and such. Don’t do this. Spend your time and energy somewhere else. There is not that great a difference in returns across the various portfolios and more importantly no guarantee those differences will persist over time. What about the first two portfolios on the diversified list? They show great return and risk characteristics. These are examples of low beta, high tilt portfolios. I chose the allocation among the quant portfolios and bonds specifically to limit risk to a max drawdown of approx 10%. Great returns and low risk. Problem is 90% of investors can’t or won’t implement these. In my opinion, more investors should look into these portfolios but I’ll leave that topic for another post. Overall, 4 of the buy and hold portfolios stand out – the top 2 quant/bond portfolios, the Permanent Portfolio, and the Risk Parity portfolio. They stand out because they have great returns and very low risk (low drawdowns, high sharpe/sortino ratios). The low risk characteristics of these portfolios make them much easier to stick to than the other diversified portfolios that can have very bad years. It also makes them attractive to retirees where bad years can have a huge impact on retirement.

Now for the tactical asset allocation category. These portfolios take the diversified buy and hold portfolios, in particular the IVY portfolios, and apply trend following and momentum to increase returns and reduce risk. Trend following and momentum could also be applied to other asset allocations which would also improve risk adjusted returns. What makes these portfolios appealing is the incredibly strong risk adjusted returns, without sacrificing too much in absolute returns. Again, that makes the portfolios easier to stick with and more suitable for retirees. And there is even more that can be done with these types of portfolios as I’ll explore in future posts. This is an area of active research for me.

There you have it. Enough portfolio data to drive one nuts. Use it as a reference, a starting point for further research, etc… I’ll keep these portfolios updated as best I can in the future and publish updates periodically.


Using dividends to cushion against market gyrations

October 20, 2014

Early in 2013 I presented a quantitative strategy based on dividend paying stocks from OShaughnessy’s What Works On Wall Street. The Enhanced Dividend Yield strategy. The strategy provides market beating returns, higher sharpe ratios, and a healthy dividend stream. One of the best things about this strategy is it’s high stick-to-it-iveness. That’s a highly technical financial term that means that it is one of the easier quant strategies to adhere to and implement because of the high income stream from the strategy. The income stream provides a cushion that helps investors weather the tough times. Now, about 22 months after I first presented it I thought it would be good to go back and look at the strategy in some detail to point out it’s strengths and weaknesses. Also, we’ll look at the different results of this strategy for an investor pre-retirement who is focused on building wealth and for a retired investor spending the income from this strategy.

The Enhanced Dividend Yield strategy basically takes cheap market leading stocks and ranks them by dividend yield. An investors buys the top stocks from this ranked list and holds them for one year. Rinse and repeat. For details of the implementation of the strategy see my earlier post. Lets look at the recent performance of the strategy. The table below shows the performance of the Enhanced Dividend Yield strategy from the beginning of 1999 through Oct 19, 2014. I also list several other metrics which I’ll discuss below.

Enhanced Div Yield 1999 to Oct 2014

The strategy returned 17.4% a year over the last 15 years and 10 months. That compares to the 4.6% a year for the SP500 over the same time period. Not bad. As the dividend yield column shows the starting yield of the portfolio varies quite a bit every year. This is really a value strategy first, then a dividend strategy. The potential dividend income from the portfolio rises at 15.5% a year as well. An investor who is trying to build an income stream as retirement approaches would have been well served with such an approach. But what about a retiree? A retiree would maybe like to spend the dividends instead of re-investing them in the portfolio. The table below shows that analysis along side the previous one.

Enhanced Div Yield for retirees 1999 to Oct 2014

For a retiree, who is not re-investing the dividends into the portfolio the results of the Enhanced Dividend Yield strategy are quite impressive as well. Without re-invested dividends the portfolio grows at 9.9% a year plus the dividend income stream grows at 8.6% a year. That’s still a lot better than the SP500 and the income stream handily trounces inflation as well. Sounds like a great strategy to consider for a retirement portfolio. It is one of the core strategies I use in mine. But it’s not all roses. There are two negatives. One, is the strategy does not limit drawdowns. This is typical of all quant value strategies. You get higher returns in the long run but still suffer large drawdowns. The Consumer Staples and Utilities Value strategies do a better job of limiting drawdowns while still providing some level of income, all be it lower. The other downside, in particular for retirees, is the fluctuating level of dividend income from year to year. As the last column in the table shows, from year to year the level of dividends can decrease quite a bit, even though it is rises quite strongly over the long run. A retiree using this strategy could apply a smoothing approach to the income stream over a 3-5 year period to avoid this downside. Also, an allocation to bonds in an overall portfolio can accomplish the same thing.

Finally, what is the Enhanced Dividend Yield strategy telling us today. I ran the screen this weekend and the top 15 stocks are shown below.

Enhanced Div Yield Screen as of Oct 18 2014

The portfolio above yields 5% as of last Friday. That’s on the low end of the yield range for the last 15 years (compare to the first table in the post) which is to be expected in the current valuation environment but still a robust yield. We’ll see how it does over the next year compared to the market. I have a tracking portfolio set up in FINVIZ to monitor its progress.

In summary, the Enhanced Dividend Yield quant strategy is a great strategy for investors in the wealth building part of their lives and for retirees in the withdrawal phase. The high level of dividends not only enhances returns but also makes it easier to stick with the portfolio during rough times knowing that at least that income stream is there.

Future returns and their impact on SWRs

October 16, 2014

Today I wanted to talk about the forecasting of future returns and more importantly what implications future returns have for SWRs (Safe Withdrawal Rates). As I showed in my last post, the first 10 year period real return in retirement is the best predictor of SWRs for 30 year retirement periods. Thus by creating a model for returns for the next 10 years based on where assets are priced today we can get a better idea as to the applicability of the traditional 4% SWR for future retirees.

There’s a been lot of discussion over the last couple of years in the financial blogosphere about future equity returns. Most of the that discussion is centered around US equity markets being over valued and therefore leading to very poor returns going forward, for anywhere from the next 5 years to the next 20 years. The analysis goes something like this; you take historical valuation figures based on different parameters such as Shiller PE, MarketCap to GDP, Q ratio and see what the historical forward returns where for different levels of starting valuation. Then you compare that history to today’s valuation level and make a statistical forecast for future returns. It’s more complicated than it sounds but that is essence of what the process is to arrive at return forecasts. The best summary of these types of analyses I’ve found anywhere is this post from GestaltU. There is a lot of gory detail (or awesome detail for folks like me) in the post and associated caveats but here is the punch line that we need to consider.

Butler forecasted returns oct 2014

Not so pretty right? The most probable compounded US equity returns for the next 10 years is about -1%. This could imply a rethinking of the 4% SWR. This being a statistical analysis there is a potential range of outcomes. I wanted to get a feel for what those range of outcomes were in the past with specific years attached to it – after all, this is based on historical data anyway. Sometimes the human brain needs context and a story behind the data to derive a better understanding. I went to my handy dandy data base of historical stock returns that I use for my SWR analyses and added the Shiller PE (CAPE) data to it, as well as bond returns, 60/40 portfolio returns and inflation. At the end of the day we care about total portfolio returns, not just equity returns. I then sorted all 10 year period returns starting in 1929 by CAPE range. Below is the summary. Highlighted is the CAPE range of the US equity market today, greater than 25.

Ten year historical returns by CAPE range oct 2014

The data in the table pretty much agrees with the GestaltU data. Forward stock returns from high valuation markets are low. Stock returns pretty much go up with decreasing starting valuation as expected. A couple of interesting things to note in the table are that bonds returns are highest at both high starting equity valuations and low starting equity valuations. Also, inflation increases as starting equity valuation decreases. This is not the place to delve into the reasons why but the important point is that these effects are all important to and affect what we care about which is total real portfolio return. The portfolio returns also decrease directly with increasing equity valuation but note that the absolute numbers are not as bad. A 2.85% avg portfolio real return from the highest valuation tier is a lot better than a -1% stock only return that starts freaking investors out. Now recall that the first 10 yr period return for the worst case retiree in history, which defined the 4% SWR rule, was -1.26% real per year for a 60/40 portfolio. So, this simple look into the details tells me that the 4% SWR rule is probably still OK. But could some of the future periods be worse than any other period in history? Yes. I think for more insight here we need to look into even more detail at the historical data. Below I break out all the 10 year periods with CAPE ratios greater than 20. The table is sorted by CAPE ratio from highest (most expensive) to lowest (cheapest).

Detailed Ten year historical returns by CAPE oct 2014

Along with CAPE I also added the yield of the 10 year note at the start of the period. For example, in the begging of 2000 the CAPE ratio was over 42 and the 10 yr yield was 5.11%. Note that indeed the 10 year period starting 1966 started from a high valuation, although not in the top tier, and produces almost the lowest total portfolio return with both negative real stock returns and bond returns! Now, could it be worse than 1966 in the future thereby forcing a lowering of the 4% SWR? To look at that I highlighted all the past years in yellow that could be called similar to today, with high equity valuations and low bond yields. Looking at the data and thinking about the story behind those periods makes me think that 1937 could be a similar parallel for today. Beginning of 1937 was about 8 years after the Great Depression. Today we stand at close to 7 years after the Great Recession. Maybe? Who knows but worth a think. The data is the data though – high equity valuations and low bond yields is what we face.

You could even say that there is no precedent for where we are today with CAPE ratios of over 25 and bond yields below 2.5%. And that could imply that the 4% rule may fail going forward. Definitely a possibility. But I think the important point is that it would take more than weak first 10 year equity returns to kill the 4% SWR. It would take a similar but even worse brew than the 1966 period which combined high starting equity valuations, with rising bond yields, and rapidly rising inflation.

If that was where we ended it would indeed be a worrying picture. And then maybe the best recommendation would be to reset expectations to a lower SWR. But the problem with the all the above is that we have taken a very narrow view of the brave new investing world we live in. The entire discussion above and in most of the blogosphere is solely focused on US equity markets. There are other markets out there with better valuations. Most of the discussion around the potential failure of the 4% SWR assumes a US large cap stock and US intermediate government bond only allocation. There are better asset allocation methods out there. There are portfolio strategies that seek to enhance return and minimize risk which also help maintain SWRs. There are better retirement withdrawal and spending models that help combat poor early returns in retirement. Before we go off concluding that a 4% SWR won’t cut it anymore we need to look at all of these and what impact they have on SWR. I’ve done that in many previous posts but I plan to recap and summarize most of that work in future posts.

In summary, future US equity returns are probably going to be lower than in the past due to today’s high valuations. But that is only part of the story when it comes to determining SWRs. Despite these headwinds I think that with the combination of better asset allocations, better portfolio strategies, and better spending and withdrawal models the 4% SWR can continue to serve retirees well into the future.

When you retire matters a lot more than you think

October 13, 2014

The starting year of retirement makes a huge difference in the success or failure of a retirement plan. This is the key difference between the accumulation phase of investing and the withdrawal phase of investing. Yet the majority of investing writing does not take this key difference into account when speaking of returns, asset allocation, planning, etc. So, lets look at how important the starting year of retirement is to a successful outcome and what potential a retiree can do to mitigate this risk.

Why does the starting year of retirement matter in the first place? It is because retirement involves yearly withdrawals from the portfolio. Once you start making yearly withdrawals from a portfolio the sequence of investment returns, the year by year order of those returns influences how much you can withdraw from that portfolio and have it last for the full retirement period. This is NOT the case pre-retirement where there are no yearly portfolio withdrawals. The figure below (from Gestaltu) shows three separate return paths to the same final total portfolio value.

Alternate return parths chart oct 2014

The blue line shows a constant return path. The red line shows an early bull market followed by a weak market and the green line shows an early bear market followed by a strong bull market. For an investor in wealth accumulation mode, all three return paths have the same result – the same end wealth. But that’s not the case for retirees. The blue line retiree enjoys a consistent retirement and in fact their safe withdrawal rate (SWR) can be set equal to the portfolio return. Unfortunately, the blue line return assumption does not exist. It never has. Annual returns are lumpy. There are some good years and some bad years. And it turns out the order of good vs bad years makes all the difference for retirees. In retirement literature this is called sequence of returns risk or path dependence in retirement. The red line retiree in the figure above can enjoy much higher withdrawals from their retirement portfolio than the green line retiree even though the returns over the entire period are exactly the same. Using the blue line return as the SWR, the blue line retiree would have a successful retirement. The red line retiree would have a wildly successful retirement and the green line retiree would run out of money. For a great case study of this particular scenario see here.

If the sequence of returns matter in retirement then how can we measure it’s impact? We can look at the correlation between time period returns and SWRs. As hinted at in the previous paragraph early returns are much more important than later returns. The chart below show this quite clearly. From Michael Kitces who has a great recent post on this topic.

SWR correlation by return time horizon oct 2014

Turns out that 10 year real returns have the highest correlation with SWRs. Also, notice that nominal returns don’t matter nearly as much as do real returns which makes sense since retirement is about keeping inflation adjusted spending consistent. 10 year real returns explain almost 85% of the SWR. This is where the attention should be focused – on the first 10 years of retirement. Now lets look at some specific retirees of the past to give you a feel for the impact of these numbers. See the table below.

Rolling 10 and 30yr 60 40 portfolio return history summary oct 2014

In the table I show the 60% stock/40% bond portfolio 10 year and 30 year period return figures for various years of starting retirement. The first two rows show the worst case retirement periods in history, the 30 year periods starting in 1966 and 1929. The 1966 period alone solely determines the historical SWR. See here for more on 1966. As the table shows, the 30 year period returns are pretty darn good even for the 1966 retiree. However, what really hurt the 1966 retiree was the first 10 years of retirement where their returns were negative on a real basis. The 1929 retiree despite the worst stock market decline in history experienced higher real returns than the 1966 retiree (even though nominal returns were lower). As they say, you can only eat inflation adjusted income. Just for kicks, now look at Mr and Mrs 1982. Wow! What a retirement. Simply, really, by luck of the draw of when they retired.

What does the future hold? For the 1999 retiree, the first 10 years have definitely not been great but they have not been as bad as the past. In fact it looks like the 1999 or the 2000 retiree are proceeding to a successful retirement (see here for the latest data). But it was dicey those first 10 years. And what about those who are considering retiring shortly? That is a whole other topic for a post but what needs to be considered is clear. Retirees need to consider what weak returns during the first 10 years of retirement can do to their portfolios and put plans in place accordingly. I’ve touched on this topic in other posts on various portfolio options and spending strategies that can improve the probability of a successful retirement. I’ll return to it again in future posts. There is much to say.


Investing during market turmoil

October 10, 2014

Maybe the best thing about quant or automatic investing is the peace of mind it provides during times of market turmoil. While I would by no means characterize the recent market action as ‘turmoil’, many investors seem to think it is. It always amazes me how short term the investor memory is. Has everyone forgotten the corrections of 2010 (16%) and 2011 (19%)? Of course, in theory, the buy and hold investor should be in a state of peace, confident in the long term, and ignoring the short term fluctuations of the market. But as that wise man, Yogi Berra said, “In theory there is no difference between theory and practice. In practice there is.” As evidenced by the abysmal historical record of the average investor. This is where automatic or quant portfolios shine by having built in automatic decisions and risk management. So, lets look at some examples of what quant investors are doing today.

First, I’ll start out with the automatic diversified ETF portfolios like the IVY varieties. This also includes the Permanent Portfolio and many others but I’ll focus on the IVY versions. Below are some screenshots of what and investor in these portfolios would be doing today. Lets start with the basic IVY portfolios. The status of the GTAA5 and GTAA13 portfolios is shown below. I keep spreadsheets for these portfolios that update automatically which you can access here and here.

GTAA5 status oct 2014

GTAA13 status oct 2014

A GTAA5 investor would be 60% invested and 40% cash right now. A GTAA13 investor would be 55% invested and 35% in cash. Due to the recent market pull back some of the riskier assets have triggered their risk management provisions and commodities have been in cash mode for a while. No thinking required, automatic decision making once a month. Now for the more aggressive portfolios, GTAA AGG3 and AGG6. Their status is shown below.

GTAA AGG status oct 2014

Both GTAA AGG3 and AGG6 are still fully invested. Notice how the GTAA AGG portfolios have shifted more to risk off assets (bonds) during the recent turmoil. In general, the AGG portfolios stay more invested than the basic GTAA portfolios because they always seek the assets classes that are moving higher the fastest. It is quite quite rare for these portfolios to move completely to cash. In fact, in the last 10 years these portfolios only spent 3 months at 100% cash. Again, automatic decision making.

The IVY portfolios are just a few examples of the many types of diversified portfolios out there. The same risk management, automatic decisions can be applied to other portfolios as well. Below are some of the other portfolio choices out there (from Meb Faber). By the way, if you are a die hard buy and holder then the Permanent Portfolio is the easiest to implement and due to its low volatility and drawdowns has shown that is probably the easiest portfolio to stick with.

Diversified portfolio examples oct 2014

Now what about the individual stock quant portfolios that I’ve presented here? Those quant portfolios are buy and hold portfolios with a holding and re-balancing period of 1 year. As I detailed here you can then use a bond allocation to manage risk and drawdowns. For example, you can target a max drawdown of less than 10% by choosing a 30% quant stock portfolio with a 70% bond portfolio. Otherwise, there are methods to implement automatic risk management into individual quant stock portfolios. It is something I’m using myself and will have some posts on the topic in the future.

In summary, it’s times like these where automatic investing really shines. Having a systemtatic investing process takes much of the worry out of wildly gyrating markets and the investor uncertainty that comes with it.



Get every new post delivered to your Inbox.

Join 1,726 other followers

%d bloggers like this: