Monday, May 8, 2017

Dynamically Reweighted Equity Balanced Portfolio Current Review

Current Commentary/Analysis For:

## [1] "May 08 2017"

Dynamically Reweighted Equity Balanced Portfolio Current Review

Paired correlation of the six month run (below) shows the US Treasurys (UST) has dropped to negative values for the most part but for consumer staples (XLP). Consumer staples are showing as large weight in the optimized solution, so this would be a problem but for the fact US Treasurys are still “blown” from the Federal Rserve pressures and are not to be used. (See the Origin Process rates and yield curve analysis.) But there is now an optimized solution as shown by the “normal” outward bowing of the effecient frintier for 6 months. The equal weighted portfolio is still on the frontier but now not well to the left, again showing active management is now additive.
From November 2016 to February 2017 the 6 month effecient frontier was a vertical line showing the swamping impact of finance (XLF) surge and the large loss from US Treasurys (TLT). The XLF and the TLT are still at extremes. But when the 6 month frontier was a vertical active managment was not useful.
Since US Treasurys are “blown” and still too rich to provide a hedge or potential offset to a SP500 and sectors correction, the ability to take equity risk is reduced even though active management via a optimization now had good risk reward ratios.
The protfolio is now roughly 50% equity stressing consumer staples (XLP), consumer discretionary (XLY), finance (XLF) and technology (XLK).
The following analysis uses a read of graphing the efficient frontier at various time spans, and then derives the optimized weights of allowed assets - the SP500 major sectors in ETF form - and an analysis of the market volatility and an analysis of the correlation between US Treasurys and the SP500.
Using the correlation and the volatility of the assets, a frontier of the optimal portfolios are mapped for 6 months, 1 year, and 3 years.
The 6 months frontier was “swamped” by finance (XLF), but recently an outward bowing of the frontier to the left has showed risk of the protoflio can be reduced yet expected return realized. That an optimization solution can be found and that active portfolio management is additive. The equal weighted portfolio (EWP) is now on the effecient frontier while it had been well to the left of the frontier after the election and then after the State of the Union speech. But since the absence of US Treasury negative correlation which reduces risk of the equity sectors with a balanced weight of long duration US Treasurys, from the extreme pressure exerted by the Fed to maintain low rates, a large cash weight is still required.
Equity weights are now 50% of the portfolio, with no US Treasurys and cash 50% of the portfolio.
The 6 month frontier map is outward bowing and the equal weighted protfolio (EWP) is migrating to a more typical place inside the pocket to the right of the outwardly bowing frontier.
The one year efficient frontier graph will likely follow the 6 month map and start showing a more typical outwardly bowing to the left curve. But since there is still an absence of curve and most of the frontier is a vertical, it also suggests a large cash weight is appropriate.
The 3 years frontier graph shows the normal efficient frontier shape bowing out allowing return to be maintained yet risk greatly reduced by weighting low correlated sectors.
Histograms of the above time periods show the weights for the least risk solution portfolio and the risky solution of the tangent line portfolio.
The the more risky tangent portfolio solution on 6 months of data shows finance (XLF), tech (XLK),consumer staples (XLP) and consumer discretion (XLY) weights. The least risky portfolio is similar but with a larger consumer staples (XLP) wieght.
6 Months data histogram of weights for the optimized portfolio:
One year of data histogram of weights for the optimized portfolio:
The 3 years efffecient frontier, staples (XLS) sector is a keystone weight with tech (XLK) and finance (XLF).
Three years of data histogram of weights for the optimized portfolio:
The above frontiers and optimal weights provided are put through a backtest for the last ten years. This is not forward looking but is plotted to see if there is a coherent structure to the optimized portfolios over time, such that active management is useful.
And good results for active management shows the use of US Treasurys and then reweighing to more optimal equity sector mix can keep pace with the index - the SP500- yet result in about 65% of the index risk (the backtest shows 61% for the last 10 years).
Note that in theory the main tool of reducing risk is the use of US Treasurys (TLT). Since US Treasurys are relatively the same credit quality as SP500, this explains why investors will accept US Treasurys yields of about 1/2 of SP500 yield. That differential is the price for the US Treasury insurance in for those economic downturns.
The “drawdown” of the theoretical backtested portfolio shows that in theory the model recommended weights that would have resulted in 50% of the market loss - the index SP500.
The summary of the above in the backtest plot shows the potential for good risk adjusted returns in comparison to the index, the SP500 with the Origin Process active management. Past returns and certainly not backtesting is no indication of future returns realized, they only frame the analysis of the portfolio process.
A theoretical “what if” test where the backtest is applied to only equity and with no US Treasurys. This shows the usefulness of US Treasurys during major index downturns, or recession. A sector dynamic re-weights into US Treasurys does reduce risk. However US Treasurys are currently several sigmas rich and as the Fed normalizes rates large downward jumps are occurring. Currently no US Treasurys are used in the portfolio. The backtest shows that the use of sector rotation optimization only slightly reduces risk without losing rewards and is smilar to a simply long SPY position. However sector rotation is used as within a short time frame - say under 5 years - risk is reduced an reward maintained via sector rotation.
The backtest for this two asset portfolio, SPY and TLT (long duration US Treasurys) is derived.
Return and correlation is graphed. The offset of US Treasurys is missing as summed up in the rolling 2 months correlation graph. US Treasurys are not providing offset to equity.
The correlation grids is again provided.
This supports the 0% weight of US Treasurys now, but this is more clearer in our US Treasury and curve analysis.
The realized vol for the SP500 is graphed for short dated to long term. The vol shows levels of near historical lows which when combined with the vertical graphs above affirm the max cash weights. This presents a problem and is felt to be a near artificial results of the large Fed pressure to keep rates abnormally low and thereby “suck” the risk our of the market. Again this suggests a large cash weight as the SP500 volatility will likley return to mid 20% upon normalization and thereby risk greatly increase which may not be offset by lower pair wise correlation.
US Treasurys vol is mapped in comparison to SP500 vol. The extraordinary low volatility for both assets classes is shown and that SP500 volatility is two low in relationship to US Treasury vol. This again is red flag that insists upon large cash weights.
The weights applied:
AssetWeights
Cash52%
US Treasurys0%
Equity48%
Sectors:
Tech-XLK12%
Finance-XLF12%
Consumer Discretion - XLY12%
Consumer Staples - XLS12%
Necessary increased risk control comes about with more frequent model runs and more frequent reweights during this unusual speculative phase; reweighting as frequently as twice a day on market open and market close.
Once a normal distribution of risk is shown by a normal outwardly bowing frontier in 6 months, 1 year, and 3 years, the reweighting will be reduced to once a day and then weekly and then likely only monthly or months. If the curve analysis shows the Federal Reserve has normalized the US Treasury weights will be applied using the optimized weights.

No comments:

Post a Comment