My portfolio is focused on uncorrelated asset allocation with risk parity target and use of high leverage with the objective to maximize expected return at reasonable risk.
Portfolio / TAN / ^SP500TR old performance stats, 2015-12-31
- Inception 2009-01-01
- Return since inception 131.58% / -56.52% / 163.02%
- CAGR 12.75% / -11.22% / 14.81%
- Max drawdown -97.85% / -87.94% / -27.19%
- Risk-Return-Ratio 0.13 / n/a / 0.54
- IRR 14.56%
Portfolio / TAN / ^SP500TR new performance stats, 2016-12-31
- Inception 2016-01-01
- Return since inception -1.06% / -43.24% / 11.96%
- CAGR -1.06% / -43.24% / 11.96%
- Alpha -6.24% / -60.26% / 0.00%
- Volatility 23.93% / 29.35% / 12.89%
- Max drawdown -18.16% / -44.43% / -10.27%
- Return = Stocks + Funds = -7.05% + 5.99% = -1.06%
- Return = Gross Return - Expenses = 8.87% - 9.93% = -1.06%
- Gross Return = Securities + Currencies = 3.45% + 5.42% = 8.87%
- Expenses = Trade Commissions + Interest + Tax = 1.63% + 6.84% + 1.46% = 9.93%
- Trade Commissions = Securities + Currencies = 0.95% + 0.68% = 1.63%
Since more was invested in the old portfolio during its higher return periods the IRR has been better than the CAGR.
The old portfolio volatility has been extreme. The -97.85% drawdown during 2.5 years (from late 2010 to early 2013) means that the total return was 10667% during the 4.5 years outside that period to reach a total 131.58% over 7 years. The CAGR was thus -78.47% during the 2.5 years drawdown period and 182.86% during the other 4.5 years. Extreme indeed.
During the same 7 years period the solar ETF TAN, which is the closest portfolio benchmark since my portfolio has been US listed solar stocks focused, returned -56.52% and the S&P 500 returned 163.02%.
At this point in time I'm satisfied with the return achieved during the past 7 years and I am now reducing my risk significantly in a new portfolio with the below described new allocation and leverage strategy. The goal is to almost double the average annual return of the portfolio to 25% while reducing its max drawdown risk more than a factor 3 to 30%, thus improving the Risk-Return-Ratio more than a factor 6 to around 0.8. To track these new targets the portfolio performance stats are reset 2016-01-01 (old closes and new starts).
The portfolio Alpha (return over risk-free and market risk return) needs to average around 21% (corresponding to 7% on an unlevered basis) to attain the 25% CAGR target. My fund picking should be able to generate this amount of Alpha. Thus my stock picking only needs to generate 0% Alpha, which is not as easy as it sounds since many professional stock fund managers fail to generate positive Alpha.
- Cash 10%
Low Risk Securities 15%
- Money Market Mutual Funds 0%
- Fund of Hedge Funds 15%
High Risk Securities 75%
Hedge Funds 50%
- Relative Value Funds 30 %
- CTA Funds 20%
Mutual Funds 20%
- Bond Funds 10%
- Stock Funds 10%
- FSLR 2%
- JASO 1%
- CSIQ 1%
- JKS 1%
- Hedge Funds 50%
The Cash and High Risk Securities classes are actually allowed to slide in the 0-10% and 75-85% ranges respectively and re-allocation (back to closest range limit) between the two classes are not forced until they slide out of range.
The Low Risk Securities class represents any virtually risk-free securities that can offer slightly higher return than interest rate with minimal additional portfolio risk. This includes risk averse funds of hedge funds. The purpose of this asset class is not liquidity but rather to sacrifice liquidity for some small extra return without adding significant risk, while the purpose of the Cash class is both liquidity and minimal risk.
The internal allocation of the Low Risk Securities class can change with changes in interest rate while the internal allocation of the High Risk Securities class should not change with changes in interest rates as it is already "all weather" tuned.
The Securities classes will be leveraged in order to boost return. All parts of each class will be equally leveraged. This lever will start at 5 for the High Risk Securities class and 1 for the Low Risk Securities class, i.e. a High Risk Securities allocation of 5 x 75% = 375% containing a Stocks allocation of 5 x 5% = 25% etc. The High Risk Securities lever will then be reduced by 1 and the Low Risk Securities lever increased by 1 every five years until it they have reached 1 and 5 respectively and thus equal allocation as 1 x 75% = 5 x 15% = 75%. In that simple way five different 5-year plans of risk allocation for the coming 25 years have been created.
If interest rates are too high then the Low Risk Securities are not expected to be able to beat the return on Cash and the Low Risk Securities lever must be set to 1 regardless of plan. If interest rates are extremely high to the extent that high risk assets are not expected to be able to beat the return on Cash then the High Risk Securities lever must be set to 1 regardless of plan.
The performance targets above are based on a High Risk Securities lever at 5. With a 10% return target on High Risk Securities assets and a cost of leverage at 4% the return target from the Risk class is 75% x (5 x 10% - 4 x 4%) = 25.5% with a lever at 5. For each reduction of the lever by 1 the return target is thus reduced by 75% x (10% - 4%) = 4.5% and thus when the lever is 1 the return target is only 7.5%. The max drawdown risk will be reduced with as much as the lever is reduced, which means it will be reduced from 30% with lever at 5 by 6% each step to end at 6% with lever at 1.
The high 5x leverage of the High Risk Securities class is enabled by risk parity in the asset allocation, where the sum of the bond and hedge funds (60%), despite its large size, has significantly lower max drawdown risk than the sum of the stocks and stock funds (15%). The reason the stocks and stock funds then don’t have even less allocation is that they contribute with more stable returns. Although they suffer big drawdowns on occasions, they consequently return above their CAGR most of the time. The bond and hedge funds have the opposite profile of never suffering big drawdowns and most of the time they return less than their CAGR due to occasional surges. The chosen mix thus lowers the return deviation from CAGR and consequently lowers the risk of IRR deviating largely from portfolio CAGR based on timing of investing in equity additions to the portfolio. So the mix is a compromise between the drawdown risk and the return deviation risk.
The leverage risk in terms of margin-to-equity ratio for the High Risk Securities class is maxed out 75% when the lever is 5 but it is quite evenly distributed on 4 uncorrelated asset sets (Stock, Bond, CTA, Relative Value). The stocks and the stock funds are the only correlated asset classes, which means that the largest sum of margin-to-equity ratios for correlated assets is no larger than 25%. The bond and hedge funds asset classes not being correlated to each other, together with their individually lower max drawdown risks; enables their much higher sum at actually significantly lower combined max drawdown risk.
Market correlation stats
The stocks part of the portfolio has a high 78% correlation to the TAN and a medium 47% correlation to the ^SP500TR. The funds part (including stock funds) has very low correlation to both the TAN and the ^SP500TR at 4% and 5% respectively. The whole portfolio's correlation to the TAN and ^SP500TR is low at 35% and 24% respectively. The portfolio beta (with the High Risk Securities lever at 5) is 0.30 against the TAN and 0.47 against the ^SP500TR. These stats have been measured since the portfolio inception 2016-01-01.
Return distribution analysis
Past 10 years annual (AR) and quarterly (QR) return distributions of allocations corresponding to 100% S&P 500, the portfolio stocks (simulated by S&P 500 x beta 2), the portfolio funds and the whole portfolio (= the stocks + the funds) have been calculated in the histogram charts below:
The Stocks class follows an aggressive re-balancing scheme while the Funds Class follow a fixed allocation re-balancing scheme just like between the asset classes. Basically this means that the stock allocations are adjusted by their share price moves, i.e. if the share price goes up its allocation goes down and vice versa. The idea with this aggressive scheme is to effectively reduce cost of shares held. While the total stocks asset class allocation is kept constant the internal allocation for the stocks with currently bargain share prices are over weighted and the high-flyers are under weighted.
To evaluate the scheme I will look at the effect on two now exited stocks that triggered a lot of re-balancing trades, TERP and GLBL. The gains include trading and forex fee costs as well as forex gains/losses in order to reflect net effect of re-balancing trading. For the same reason dividends are net of tax. Numbers are per initial position share.
Note that the transaction amounts in local currency for entry price, dividends earned and trade gains have been converted to USD at the exit date exchange rate.
- Entry price: $16.19 (September 28-30, 2015)
- Dividends earned: $0.86
- Trade gains: $5.33
- Net cost: $10.00
- Exit price: $9.58 (April 8, 2016)
- Entry price: $6.72 (October 5-6, 2015)
- Dividends earned: $0.44
- Trade gains: $4.01
- Net cost: $2.27
- Exit price: $2.47 (April 6, 2016)