Seeking Alpha

Fred Piard's  Instablog

Fred Piard
Send Message
Author of "Quantitative Investing" (4.5 stars on Amazon) and "The Lazy Fundamental Analyst" (4 stars). Independent quantitative analyst focused on market-neutral and low volatility portfolios, looking for statistically profitable combinations of fundamental factors, and... More
My company:
Ypafi.com
My book:
Quantitative Investing
View Fred Piard's Instablogs on:
  • Backtests Of GHI Year By Year

    Someone asked me if I had the year-by-year performance of GHI. I have made a summary in the next table. The appendix gives more details. This is without hedge.

    Year

    GHI return

    SPY return

    Excess return

    1999

    -7.34%

    14.81%

    -22.15%

    2000

    65.32%

    -6.20%

    71.52%

    2001

    17.08%

    -9.16%

    26.24%

    2002

    -5.58%

    -19.92%

    14.34%

    2003

    47.40%

    24.13%

    23.27%

    2004

    24.89%

    8.79%

    16.10%

    2005

    14.17%

    7.68%

    6.49%

    2006

    18.90%

    13.88%

    5.02%

    2007

    0.31%

    4.41%

    -4.10%

    2008

    -3.84%

    -34.31%

    30.47%

    2009

    48.66%

    24.74%

    23.92%

    2010

    9.27%

    14.24%

    -4.97%

    2011

    20.68%

    2.62%

    18.06%

    2012

    23.85%

    17.10%

    6.75%

    2013

    41.40%

    27.77%

    13.63%

    2014

    38.04%

    14.50%

    23.54%

    2015 to 7/28

    14.74%

    1.60%

    13.14%

    Average

    21.64%

    6.28%

    15.37%

    Median

    18.90%

    8.79%

    14.34%

    Some data not shown in the table are important: the deepest drawdown is 34.65%, the longest drawdown is 61 weeks (it is 320 weeks for SPY on the same period). Performances are better with a hedge based on MTS10 and depend on the chosen tactics.

    In summary, annual returns and excess returns look great, but frustrating periods can be on the road. We can overcome them with a long term vision based on the 2 pillars of the strategy:

    • The best companies in recurring consumption will grow because the global middle class will grow.
    • Appropriate ranking factors give a statistical edge to surf the waves of institutional buying.

    Appendix: detailed data and charts

    The equity curve of GHI is in red, SPY is in blue.

    1999

    (click to enlarge)

    2000

    (click to enlarge)

    2001

    (click to enlarge)

    2002

    (click to enlarge)

    2003

    (click to enlarge)

    2004

    (click to enlarge)

    2005

    (click to enlarge)

    2006

    (click to enlarge)

    2007

    (click to enlarge)

    2008

    (click to enlarge)

    2009

    (click to enlarge)

    2010

    (click to enlarge)

    2011

    (click to enlarge)

    2012

    (click to enlarge)

    2013 (out-of-sample after model design)

    (click to enlarge)

    2014

    (click to enlarge)

    2015 to 7/27/2015

    (click to enlarge)

    Jul 28 5:33 PM | Link | Comment!
  • Beating SPY And HFRX Hedge Fund Indices With Zero Market Exposure.

    This is a personal blog post, not a regular SeekingAlpha article submitted to editorial review. This is a follow-up of my portfolio for informational purposes, not investment advice. I am not a Registered Investment Advisor. Past performance is not a guarantee for the future.

    I have designed Ypafi Market Neutral as an all-weather portfolio to stay 100% invested with a zero-sum market exposure. Here are the results for the 6-month period from 12/28/2014 to 6/28/2015. Comments are after the table.

     

    positions

    12/28/2014 to 6/28/2015

    MARKET NEUTRAL PORTFOLIO

      

    Portfolio without hedge

    24 stocks

    7.8%

    100% hedged on margin (short SPY)

    24st+1ETF

    6.1%

    100% hedged without margin (25%long SPXU)

    24st+1ETF

    4.2%

    Components

      

    Consumer Staples Component (Defensive1)

    5 stocks

    5.8%

    Sector Benchmark XLP

     

    -0.5%

    Healthcare Component (Defensive2)

    5 stocks

    31.8%

    Sector Benchmark XLV

     

    10.5%

    Tech Component (Cyclical1)

    5 stocks

    -0.1%

    Sector Benchmark XLK

     

    1.1%

    Industrial Component (Cyclical 2)

    5 stocks

    -4.1%

    Sector Benchmark XLI

     

    -2.8%

    SP500-Growth&Value (all sectors)

    4 stocks

    -3.8%

    Benchmarks

      

    S&P 500 index with dividends (NYSEARCA:SPY)

     

    1.6%

    HFRX Market Neutral Hedge Fund Index (on 6/25)

     

    0.8%

    The first half of the year was an encouraging start for my Market Neutral Portfolio in its hedged and non-hedged versions. Healthcare stocks are responsible for most of the gains, but other sectors may help in subsequent market cycle phases.

    About stops

    Stops are very important for day-traders, my risk management is more focused on position sizing and diversification. I don't provide stop-loss instructions, and apart from very exceptional cases, I don't use stops myself. Backtests on my website (link below) have been run without stops. Backtesting a stop-loss is tricky: the number of events may not be large enough to draw a conclusion with a good confidence interval. My simulations show that stops closer than -20% usually cut the long-term return without necessarily lowering the risk (measured in drawdown and standard deviation). Stops at -20% don't seem to have a negative effect, but I have no proof of a positive effect. In a correction like August/September 2011, stops at -20% may eliminate the best stocks before a rally and let the portfolio recover with lower rank stocks. The worst case with stops is a flash-crash kicking out all positions, followed by a quick reversal with the portfolio in cash. A similar scenario may happen on individual stocks because of HFT or corporate events. Of course, subscribers can use stop tactics they are comfortable with: 20%, technical supports, or any other one.

    Ongoing research and planned improvements

    I have recently introduced the MTS10 timing indicator as a new feature for risk management and dynamic hedge sizing.

    Market Neutral Portfolio is the strategy I am focused on, because it is my own core portfolio. I think its structure can be improved. I have started to test double momentum filters to go in cash for a sub-model when the outlook is bad in a sector. It looks interesting. Ideally, I would like to make the sector selection more adaptive.

    An all-weather portfolio to stay 100% invested with a zero-sum market exposure: http://ypafi.com/YFI_en/market.html

    (in my main account I use it with a 75% hedge since December, so with a 25% exposure).

    Jul 01 4:04 PM | Link | 5 Comments
  • Fuzzy-Logic Hedging For Ypa Market Neutral Portfolio

    This post contains information directed to ypafi.com subscribers. All readers are welcome and may find here some ideas to improve their own investing strategies.

    Financial markets are driven by human group dynamics, not statistics and physics. To make investing decisions, a flexible systemic model is safer than both discretionary analyst opinions and spotless hard science. Multi-valued logics (popularized as fuzzy logics) was a very fashionable research topic a few decades ago for I.A., automatism, information theory. As a result you have fuzzy logics in cars, washing machines and other household appliances. They aim at modeling uncertain information: an interesting characteristic for portfolio management.

    Introduction to multi-valued Market Timing Indicators

    The MTS concept (Multi-Timing Scores) aims at creating systemic aggregate indicators, more robust by design than any single optimized indicator. It is focused on a long-term investing horizon, based on research and consensus, without curve-fitting, and includes the 4 main categories of market analysis: sentiment, economy, fundamentals, technicals. MTS assumes that any timing indicator may be wrong at any time, but not most of them at the same time. MTS10 is equal to the number of bearish indicators among a set of 10 elementary ones based on various data series: average short interest of S&P 500 companies, U.S. unemployment rate, S&P 500 current year EPS estimate, S&P 500 EPS trailing 12 months, S&P 500 price, new house starts. As a consequence, MTS10 value is an integer between 0 and 10.

    Fuzzy-logic hedging applied to Ypa Market Neutral Portfolio

    The next table shows the result of applying timed hedging tactics based on MTS10 to Ypa Market Neutral Portfolio. All tactics are IRA-compatible (no leverage, no short selling). There are two categories: tactics with a constant hedging ratio, and those with a variable ratio. For the variable hedging ratios, the allocation uses the following rules:

    • 75% of the portfolio is invested in stocks, 25% in the hedge.
    • The hedge is allocated in SPXU for a fraction H, and the rest in cash.
    • H has a "fuzzy logic" value between 0 and 1 (0% and 100%). H is also the hedging ratio: when it is equal to 100%, the portfolio is 100% hedged.

    H=100% means 25% of the total equity in SPXU, so 1$ in the 3x inverse index ETF for $3 in stocks: in this case the portfolio is in a market-neutral state.

    Here, the fuzzy values are not continuous between 0 and 1, but discrete. The second column gives the rules to calculate H for each version of hedging. Trading costs are included with a rate of 0.3% by transaction standing for commission and slippage. There are two differences with my previous backtests of Ypa Market Neutral Portfolio:

    - Because of the look-back period of moving averages embedded in MTS10, simulations start in 2001 instead of 1999.

    - Orders are executed the first day of the week at closing price instead of opening (the difference is less than 1% in CAGR and about 2% in max drawdown).

    Caution: past performance, real or simulated, is not necessarily representative of the future.

    Version

    Hedging ratio

    CAGR

    MDD

    MDL

    K

    no hedge

    0%

    27

    36

    103

    23

    Constant_100

    100% ($1 in SPXU for $3 in stocks)

    14

    9

    54

    34

    Constant_75

    75% ($3 in SPXU for $12 in stocks)

    17

    11

    49

    35

    Constant_50

    50% ($1 in SPXU for $6 in stocks)

    19

    18

    51

    31

    Constant_25

    25% ($1 in SPXU for $12 in stocks)

    23

    25

    64

    26

    Binary_base_0

    H=0% when MTS10<7

    H=100% when MTS10>=7

    25

    15

    54

    33

    Binary_base_25

    H=25% when MTS10<7

    H=100% when MTS10>=7

    22

    12

    51

    34

    Binary_base_50

    H=50% when MTS10<7

    H=100% when MTS10>=7

    19

    10

    50

    36

    Binary_base_75

    H=75% when MTS10<7

    H=100% when MTS10>=7

    17

    9

    49

    37

    Proportional

    H=MTS10 x 10% (0 to 100% by 10% increment)

    21

    13

    50

    33

    Proportional

    w/Threshold

    H=MTS10 x 10% when MTS10<7

    H=100% when MTS10>=7

    21

    12

    50

    34

    CAGR: Annualized return in %; MDD: Max drawdown in % on weekly rebalancing (it is deeper intra-week); MDL: Max drawdown length (duration in weeks); K: Kelly criterion of weekly returns in % (0 to 100, good from 20, higher is better).

    H=0% means 1$ in cash for $3 in stocks.

    H=25% means 1$ in SPXU and $3 in cash for $12 in stocks.

    H=100% means 1$ in SPXU for $3 in stocks.

    For the more complicated proportional hedge, MTS=n means $n in SPXU and $(10-n) in cash for $30 in stocks.

    I would like to highlight the most interesting versions:

    • Constant 75: best risk adjusted performance (measured as CAGR/MDD ratio), best MDL and best K among constant versions.
    • Binary base 0: best CAGR among all hedged versions.
    • Binary base 50: best risk adjusted performance (measured as CAGR/MDD ratio).
    • Binary base 75: best MDD (lowest expected risk), best K (highest expected robustness).
    • Proportional with Threshold: progressive, intuitive, and close to the best ones.

    The cash part may be invested in an ETF matching the investor's taste and risk profile: SPY, DIA, MDY, SPLV, PHDG, ALFA, etc... Using ETFs for this part of the portfolio allows more flexibility and lower transaction costs in case of a false signal. This is an example of using MTS. Investors with access to margin, more sophisticated hedging instruments and appropriate skills may improve the performance.

    SPXU prices are synthetic before inception (6/23/2009). Like all leveraged ETFs, SPXU has a path-dependent drift accounted in the calculation (beta-slippage). It might become more harmful in the future if the market becomes more volatile on a daily basis than it was last 15 years. Investors willing to avoid this risk may hedge with other instruments requiring margin. I am currently using SPXU with the 'Binary base 75' tactic without margin, and consider the risk acceptable.

    This work is licensed by Dr Fred Piard under a CC BY-NC 4.0 International License. Concepts and excerpts can be copied for non-commercial purposes with the author's name, a link to this material and a link to the license. Contact the author for all rights related to any commercial use of original concepts.

    Jun 05 11:45 AM | Link | Comment!
Full index of posts »
Latest Followers

StockTalks

  • Hedge reloaded before SPY became red. Main account market neutral on close. $SPXU $SDS
    6 days ago
  • New Zealand $ENZL is a more diversified dividend play than Australia $EWA ; both yield 6%
    7 days ago
  • $ITB entered my tactical ETF portfolio today (weekly newsletter). Bought at -17.5%, now at -4%. 13% gain in 15 min. Should have bought more
    Aug 24, 2015
More »

Latest Comments


Posts by Themes
Instablogs are Seeking Alpha's free blogging platform customized for finance, with instant set up and exposure to millions of readers interested in the financial markets. Publish your own instablog in minutes.