Calendar-year 2004 total return3, 17.3%
Click on either chart for a larger graphic
Realized volatility measures remain imperfect.
R-squared remains way too low. Don't do what Donny Don't does.
My usual analysis is an unconstrained regression of monthly return premiums against a basket of dissimilar indexes.
Estimate Std. Err t-stat Pr >|t|
(Intercept) 0.004 0.003 1.5 0.15
non-US equity 0.25 0.14 1.8 0.08 .
small US equity 0.37 0.08 4.9 6e-06 ***
large US equity -0.21 0.14 -1.5 0.13
non-US gov bond 0.29 0.15 1.9 0.06 .
volatility -0.05 0.03 -1.6 0.12
---
Multiple R-Squared: 0.7, Adjusted R-squared: 0.6
F-statistic: 26 on 5 and 67 DF, p-value: 3e-14
The low R-squared highlights my portfolio's high idiosyncratic risk. Intercept is positive, but not significant, so it walks but does not quack like alpha. A negative volatility bet congrues with a contrarian style, or may reflect mechanically trimming overweight positions.
The basic ANOVA table included the US 10-year bond.
Sum Sq Df F-stat Pr >F
GFD.WdxUS 0.002 1 3.1 0.08 .
Russ.2k 0.013 1 23.6 8e-06 ***
S.P.500 0.001 1 2.3 0.14
JPM.WdxUS.Gov 0.001 1 2.2 0.14
US.10y.Gov 9e-09 1 2e-05 1.00
Vol 0.001 1 2.4 0.12
Residuals 0.036 66
It was unneeded, resulting in the regression above. The intercept of the full model was the same 0.4% per month.
Total return vs S&P 500, 40/20/40 blend4, allocation5
| year | TR | S&P 500 | 40/20/40 | nonUS/eq/bd/cs |
| 2004 | 17.3% | 10.9% | 13.8% | 39/20/18/21 |
| 2003 | 36.3% | 28.7% | 24.2% | 43/20/24/13 |
| 2002 | 3.0% | -22.1% | -4.6% | 40/23/30/7 |
| 2001 | 0.6% | -11.9% | -5.9% | 41/21/32/5 |
| 2000 | 9.6% | -9.1% | 0.3% | 41/23/31/4 |
| 1999 | 12.3% | 21.0% | 11.4% | 15/14/18/54 |
| 99-04 | 12.6% | 1.3% | 6.0% | 39/21/25/14 |
Diversification is the only free lunch.
| year | TES&P | IRS&P | TEblend | IRblend |
| 2004 | 9% | 0.7 | 10% | 0.3 |
| 2003 | 11% | 0.7 | 9% | 1.3 |
| 2002 | 18% | 1.4 | 8% | 1.0 |
| 2001 | 13% | 1.0 | 10% | 0.7 |
| 2000 | 21% | 0.9 | 11% | 0.9 |
| 1999 | 11% | -0.8 | 6% | -0.2 |
| 99-04 | 14% | 0.8 | 9% | 0.8 |
Lots of relative risk. Cf Donny Don't above.
| year | RP | RPS&P | RPblend | SR | SRS&P | SRblend |
| 2004 | 1.3% | 0.8% | 1.0% | 0.4 | 0.4 | 0.5 |
| 2003 | 2.6% | 2.1% | 1.8% | 0.8 | 0.6 | 0.7 |
| 2002 | 0.2% | -2.0% | -0.5% | 0.0 | -0.3 | -0.2 |
| 2001 | -0.1% | -1.2% | -0.7% | -0.0 | -0.2 | -0.3 |
| 2000 | 0.4% | -1.2% | -0.4% | 0.1 | -0.2 | -0.1 |
| 1999 | 0.6% | 1.3% | 0.5% | 0.2 | 0.3 | 0.2 |
| 99-04 | 0.8% | -0.0% | 0.3% | 0.22 | -0.01 | 0.10 |
Risk premiums are monthly arithmetic averages.
Not currently traded -- long story.
1This page is not a formal presentation or an advertisment. I am an individual, not an investment firm, and hence I have no composite. This page is for illustrative purposes and I make no guarantees about its contents. I endeavored to follow calculations standards and have tried to make "every reasonable effort to assure that [my] performance information is a fair, accurate, and complete presentation of [my] performance." That said, I very explicitly have not prepared or presented this information in compliance with the Performance Presentation Standards of the Association for Investment Management and Research (AIMR-PPS®), the U.S. and Canadian version of the Global Investment Performance Standards (GIPS®). AIMR has not been involved in the preparation or review of my performance calculation or presentation in any way.
In re the size and density of that footnote, better safe than sorry.
2 Growth of $100 represents the hypothetical return to $100 using monthly total returns of portfolios and indexes. Return vs volatility shows annualized six-year total returns vs annualized six-year standard deviations of monthly total returns. Regression shows an ordinary least-squares fit of monthly portfolio risk premium versus 40/20/40 blend index risk premium.
3 Total return calculated using the hoary old modified Dietz method.
4 40/20/40 index blend represents a monthly total return blend of 40% GFD World Ex-US, 20% Russell 2000 and 40% US 10-Year Government Bond indexes.
5 Allocation format is non-US equity/US equity/bonds/cash. Numbers may not add to 100 due to rounding. Exchange-traded funds and mutual funds are classified according to their mandates.
6 Tracking error (TE) is the standard deviation of monthly out-/underperformance versus a benchmark index. Information ratio (IR) is the out-/underperformance versus the index divided by the tracking error.
7 Risk premium (RP) is the monthly total return in excess of T-Bill return, a proxy for the true risk-free return. Sharpe ratio (Sharpe) is this average risk premium divided by its standard deviation, the original [1966, 1975] version of Sharpe's differential ratio. Cf his updated 1994 paper in the Journal of Portfolio Management.
8Regressions statistics calculated with R.
9 Indexes used are the Global Financial Data World ex-US for non-US equity, the Russell 2000 for small US equity, the S&P 500 for large US equity, the JP Morgan World ex-US Government for non-US bond, the US 10-year Total Return Bond from GFD for US bonds and the month-to-month change of the CBOE VIX for volatility.
