Today I’d like to wrap up this series on using economic indicators to time the market. In this final post I’ll look at using the unemployment-200day SMA indicator I’ve used in the first 3 parts of the series (link to part 3) but this time apply it to individual stock quant portfolios. All of the strategies mentioned are listed in the Portfolios page. Lets jump right in.
The analysis here is pretty straightforward. I’ll take 3 example quant portfolios I’ve discussed many times here, then compare the portfolio stats of the stand-alone quant strategy with that of the quant strategy using the unemployment-200day SMA indicator (UI) on a monthly basis. I’m using the trending value (TV2) quant portfolio, the consumer staples value (CS Value) portfolio, and the large stock index value portfolio (Large stock SHY). The other quant portfolios react similarly to these three. All the portfolios consist of 25 stocks, equal weighted and the backtests are run from Jan 1999 through the June 10, 2016. Results are shown below.
The table show 3 variations of each strategy; the stand-alone strategy, the strategy using the SPY 200 day SMA to manage risk (note: in quant strategies simply using the 200 day SMA of the individual stocks leads to horrible performance), and the strategy using the UI indicator to manage risk in the strategy. The characteristic of quant strategies that makes them hard to invest in is the increased drawdowns and volatility despite their market crushing returns. Most investors abandon quant strategies during volatile times. Using a risk management strategy makes sticking with these strategies much easier.
Overall, the using the UI indicator improves the risk-adjusted performance (higher sharpe and sortino ratios) of all three quant strategies, while still maintaining their market beating performance. The consumer staples value strategy is the one the benefits the least from risk management. This is probably due to the fact that it is fundamentally a risk off strategy (the utilities value strategy reacts the same way).
That’s about it. These results are consistent with the findings in the earlier posts. Using the UI-indicator enhances risk-adjsuted portfolio performance for individual stock quant portfolios.
12 Comments
Victor · June 13, 2016 at 7:16 pm
How would you implement this? Sell all individual stocks when UI/200SMA combination triggers, or would you hedge it with short S&P500? There are a couple of things at work here… transaction costs (25 stocks out and back in) and whether you get a better bang for your buck by shorting equities in general while holding onto the individual stock selections. One way to do this… go long 2x short S&P500 ETF with margin available from individual stocks. There is the volatility drag to be concerned about on leveraged ETFs, but this could potentially be a quick way to neutralize the downside, particularly if individual selections dwindle as everything tumbles. Would be interested in your views on this.
-V
paul.novell@gmail.com · June 14, 2016 at 11:09 am
Hey Victor, great question. And definitely an advanced implementation topic. The great thing abut the UI/200SMA combo is that it doesn’t trigger all that often so the transaction costs and slippage don’t have a huge impact. For example, modeling transactions costs and slippage for the Large stock SHY strategy using the UI/200SMA combo reduced returns by about 5%, vs a 2% reduction for the normal model.
Hedging the portfolio is another way to go. Performance depends on the correlation between the hedge and the actual individual stocks in the portfolio. In general, it increases returns but magnifies drawdowns. For example, using the Large Stocks SHY strategy again, hedging increases returns from 15% to 19.6% but drawdowns go from -38% to -44%. Worth it? Maybe. Turnover in the individual stock portfolio goes down but not by that much since the signals don’t trigger that often.
Then there is how to implement the hedge. It’s a bad idea to use the leveraged ETFs to implement that hedge. One, they’re just awful products in general and only good for short term trading. The risk off signals can stay on for long periods of time, over a year, so a leveraged ETF would be quite destructive. The most efficient and cheapest way to hedge is through futures contracts. The simplest is simply to short the SPY ETF. Then of course there are options but they are a costly way to hedge. My preference is to use futures.
Then there is always the option of running a more concentrated portfolio, say 10-15 stocks, and using advanced order management like MOC orders, to control costs and slippage.
Long answer. Lots of options. Hope that helps.
Paul
Mike · June 14, 2016 at 4:31 am
Hi, thanks for the posts and improving on models. Very impressive. Is there a way to implement the UI indicator in Dual Momentum strategy? What are the returns? It is very simple to follow but lately it’s results are off. Thanks for your response.
paul.novell@gmail.com · June 14, 2016 at 10:33 am
Hey Mike, yes. just like I did in my models. When the UI indicator is on, you default the the dual momentum risk off indicator (12 month return). When the UI indicator is off, you ignore the the dual momentum risk off indicator. Don’t know what the returns are. I don’t use that model.
Paul
Richard Wilkes · June 14, 2016 at 8:50 am
Very good series and it’s always good to see if any model can be improved. A couple of years ago the Fed created another composite index the Labor Market Conditions Index or LMCI. This presumably because the unemployment rate alone does not tell everything and indications that is tainted by some of the other indicators like the participation rate. It looks as though if modeled that there may be more triggers in and out using the above or below zero of this metric. Have you looked at that one? I may do some modeling of this myself. I enjoy your work very much!
https://research.stlouisfed.org/fred2/graph/?g=4FMI
You are nearby in our area now I see in the northwest. Great time of year here… Usually supreme after July 4th! thru October.
paul.novell@gmail.com · June 14, 2016 at 10:35 am
Richard, I haven’t looked at that one myself. Phil econ looked at composite indicators and they didn’t do any better than a the simple UI indicator.
Would love to see your results if you do model it.
Paul
We do love the PNW in the summer.
Frank · June 14, 2016 at 9:21 am
Just a quick question about the Employment or UI indicator, is that one considered to be a lagging indicator?
paul.novell@gmail.com · June 14, 2016 at 10:38 am
Yes. But usually that is in reference to the economic recession, not in reference to the stock market. The data shows that the unemployment rate has been a very good risk off indicator for stocks, although by no means perfect. Also, it’s important to note that I’m using a composite indicator made up of the unemployment rate and the 200 day SMA.
Paul
Ronnie Rhyne · June 15, 2016 at 8:42 am
Paul, it there like a beginners blog or website you would recommend for someone just getting to involved in stocks and understanding ?
paul.novell@gmail.com · June 15, 2016 at 10:09 am
Here’s a few sites that may help:
investor.gov
investopedia
vanguard blog
Joe the Computer Guy · June 17, 2016 at 10:04 am
Hey Paul, As always more great information.
In a prior post you mentioned finviz as a free screener where the data can be downloaded. Checked this out and noticed the data cannot be downloaded without paying for a subscription. Did something change or am I missing something? Also noticed the limitations on the criteria that can be used.
I don’t see myself running screens on a regular basis so was reluctant to buy Stock Investor Pro plus membership. Looking like this might be the only advisable alternative though (and least expensive). Portfolio123 is even higher priced.
Thoughts?
And thanks again for all the information and help provided to a relative newb to quant investing.
paul.novell@gmail.com · June 30, 2016 at 2:55 pm
Hey Joe, yes unfortunately finviz is now charging for the downloads.
Paul
Comments are closed.