After my last post updating quant system performance I realized I’ve never posted on one of the quant systems, the microcap trending value system. In this post I’ll describe the microcap trending value system, its historical performance, and a tweak from the O’Shaughnessy version of the system which improves performance.

The top performing quant system, by annual return, from O’Shaughnessy’s What Works On Wall Street is the microcap trending value system. From the Table 28.1, page 597, from 1965 through 2009 the strategy returned 22.33% per year with a standard deviation of 20.38%, Sharpe ratio of 0.85, and a max drawdown of -53.89%. Below is a snapshot of the top 10 performing strategies by annual return. 5 of the top 10 strategies are microcap strategies.

IMG_2108

The top strategy, which I call the microcap trending value strategy, takes the top 30% of microcap value stocks ranked by P/B, then sorts that list by 52 week momentum. Microcap is defined as $50M to $200M in market cap. It further narrows that list by requiring that the stocks had positive returns over the last 3 and 6 months. The strategy then buys the top 25 by equal weight. Hold for one year. Rinse and repeat. Pretty simple once you’ve implemented a few of the other quant strategies. The strategy is easily implemented in P123. Lets see how the strategy performs in P123 from 1999 through yesterday. Below are the results.

Screen Shot 2015-10-24 at 9.17.26 AM

I’ve compared the strategy to the small cap index. P123 does not have a valid microcap index that goes that far back. As you can see the microcap strategy underperforms the small cap index by about 2% a year. More interestingly, the strategy outperforms until the financial crisis then starts to underperform. So, something changed. And it has to do with P/B as a value metric.

What I found confusing is that O’Shaughnessy spends a great deal of the book talking about how a composite of value factors is far superior to a single value factor but never applies the composite value factor to microcaps. Lets do that. Below are the results of a microcap trending value screen, as above, with the only difference is that we use the VC2 composite value factor to filter for value stocks.

Screen Shot 2015-10-24 at 9.19.02 AM

Much better. The composite value factor is far more effective, not only since the financial crisis but pre crisis as well. Since 1999 the strategy has returned 23.2% per year with a drawdown of 41%.

I summary, the microcap trending value strategy is a great performing strategy with great absolute and risk adjusted performance. When I talk about the microcap quant strategy on the blog, I’m referring to this version of the microcap trending value strategy.


12 Comments

JOHN STEIN · October 25, 2015 at 12:53 pm

If I were to sign up for P123 , want level do you need to do this type research ?
Investor , creator or researcher ?

Thanks
John

    paul.novell@gmail.com · October 26, 2015 at 1:13 pm

    Haven’t checked the membership levels in a while but I’m pretty sure you can do this at all levels. Its just a basic screen.

    Paul

adam · October 29, 2015 at 8:28 am

Hi Paul,
Would you mind posting the year-by-year returns? I’d like to run an Alpha Architect style hedge against this strategy. I’ll post my results.
Thanks,
Adam

    paul.novell@gmail.com · October 30, 2015 at 10:03 am

    Don’t have calendar year returns, just 1 year returns. They are not quite the same thing. To get calendar year returns I need to run each calendar year separately. If I get some time I’ll do that.

    Paul

      adam · October 31, 2015 at 8:58 pm

      Paul,

      I’ll just share the results from my end. If you 100% hedged and shorted an S&P 500 contract at the end of each month when the S&P 500 was below its 10 month moving average, you would get these results by year (probably the same by using options on ETFs).

      2015 – TBD, but somewhere below -6% to date
      2014 0%
      2013 0%
      2012 -4%
      2011 -5%
      2010 -15%
      2009 -6%
      2008 55%
      2007 0%
      2006 0%
      2005 0%
      2004 -1%
      2003 -5%
      2002 18%
      2001 15%
      2000 10%
      1999 -6%
      1998 -14%
      1997 0%
      1996 0%
      1995 0%
      1994 -2%
      1993 0%
      1992 0%
      1991 -11%
      1990 -12%
      1989 0%
      1988 -10%
      1987 4%
      1986 0%
      1985 0%
      1984 -5%
      1983 0%

      You can also place these against the annual returns in “What Works on Wall Street” to see how this dynamic hedging can impact a trading strategy.

      Note, this worked really well when risk free rates were 3% to 6% or more b/c you picked much of this up when you went short. Now it is the opposite, you lose some small % when you short so this is a more expensive hedge these days (contracts are in backwardation).

      Adam

        adam · November 1, 2015 at 6:14 pm

        I’m revising my numbers slightly based on using index vs total return data. 1982 is a partial year.

        2014 0%
        2013 0%
        2012 -4%
        2011 -5%
        2010 -15%
        2009 -6%
        2008 55%
        2007 0%
        2006 0%
        2005 0%
        2004 -1%
        2003 -5%
        2002 18%
        2001 15%
        2000 10%
        1999 -6%
        1998 -14%
        1997 0%
        1996 0%
        1995 0%
        1994 -3%
        1993 0%
        1992 0%
        1991 -11%
        1990 -12%
        1989 0%
        1988 -10%
        1987 -10%
        1986 -6%
        1985 0%
        1984 1%
        1983 0%
        1982 -8%

          paul.novell@gmail.com · November 2, 2015 at 10:35 am

          Thanks Adam. I can run some hedged portfolios in P123 and see what it says.

          Paul

          paul.novell@gmail.com · November 3, 2015 at 9:15 am

          Adam, I ran a hedged version of the Microcap strategy in P123. Hedging against the price only SP500 provided the best hedged returns, but not much better than the strategy on its own. For example, a monthly updated B&H version of the strategy returned 41%/yr with a -36% drawdown since Jan 1 2009. The hedged version returned about 40%/yr with a -30% drawdown. The sharpe and sortino ratios were higher with the monthly B&H as well.

          Paul

      adam · November 3, 2015 at 10:21 am

      Looks like I can’t reply to your reply, so I’m posting here.

      This was my thinking – you can run any of the What Works On Wall Street portfolios and generate year-by-year returns. For example, many probably had significant losses in 2008.

      Then you can marry those annual returns to the hedge returns, in this cay a positive 55% in 2008. This would give you the total hedged return in that year.

      It simulates what a B&H strategy would be if you overlaid a S&P 500 futures hedge to the portfolio when the markets are not trending. It does not require you to sell or alter in any way your B&H strategy positions. It would require you to hold a margin account and there would be some slight margin fees for holding a futures position.

      So if P123 can generate annual returns, not annualized, you can see how this would result on an annual basis. I know that WW on Wall Street lists the TV annual returns and this method works well to flatten the equity curve.

      Food for thought.

      Adam

adam · October 29, 2015 at 11:48 am

One more comment/question. I believe that the $50M to $200M capitalization range should be inflation adjusted. Did you do this in your analysis?

Thanks Paul!

Adam

    paul.novell@gmail.com · October 30, 2015 at 10:01 am

    I did both. Doesn’t make much difference. Only been around 9% total inflation since the end of 2009 to the end of 2014.
    In my personal screen I use 50-250.

      adam · October 30, 2015 at 7:10 pm

      Cool. Thanks for clarifying.

Comments are closed.