Reflation: A Risk-Parity Strategy
NDX vs. SPX
NDX has outperformed SPX since the ’80s on an absolute return basis. However, it has suffered uglier losses (-78% in 2001) and higher volatility (21v vs. 18v, 1yr average).
NDX Outperformance as a Factor Return
Can we get NDX’s superior returns without the additional risk?
If we invest only in its outperformance to SPX (long NDX and short SPX, equal-weighted), we can maintain a bet on greater productive returns of high-growth, capital-efficient, balance-sheet rich tech companies while hedging against broader macroeconomic declines.
2002-Present
As a factor return, NDX outperformance is much less correlated with either index’s performance (15% corr to SPX, 46% corr to NDX, compared to 95% NDX/SPX corr, since 2002).
This feature has made it more robust than a pure long-only strategy in the previous two recessions:
1) In the 2020 COVID-19 Recession, it lost only -5% vs. -30%/-33% in SPX/NDX.
2) In the 2008 Financial Crisis, it lost only -14% vs. -55% in either index.
| SPX | NDX | NDX_SPX | |
|---|---|---|---|
| Sharpe | 27.4% | 42.2% | 54.3% |
| Return | 5.4% | 9.3% | 3.9% |
| Vol | 19.8% | 22.0% | 7.2% |
| MaxD | 56.8% | 55.6% | 14.1% |
| SPX | NDX | NDX_SPX | |
|---|---|---|---|
| SPX | 100.0% | 94.6% | 15.2% |
| NDX | 94.6% | 100.0% | 46.4% |
| NDX_SPX | 15.2% | 46.4% | 100.0% |
Taking a risk-parity approach, we can lever NDX - SPX to the total return of NDX (roughly 4:1 for the period) for an apples-to-apples comparison: 
Gold as a Reflation Hedge
2002-Present
Gold has outperformed either equity index in the period on a lower vol with no correlation, although with a sizeable drawdown in 2012-2016. The majority of the returns in gold can be attributed to unconventional Fed policy in the lead-up to and during both 2008 and 2020 recessions.
Gold is also uncorrelated with NDX outperformance (kind of a no-shit obvious result, but important), which leads us to our final strategy:
A Risk-Parity Approach
The Power of Risk-Parity
Given two, high-sharpe, uncorrelated streams of returns (XAU and NDX outperformance), we can build a portfolio levering each to an equal level of risk, reducing overall portfolio risk while maximizing returns, significantly outperforming either:
| rebal | XAU | NDX_SPX | |
|---|---|---|---|
| Sharpe | 94.6% | 58.0% | 54.3% |
| Return | 6.4% | 10.2% | 3.9% |
| Vol | 6.8% | 17.6% | 7.2% |
| MaxD | 11.3% | 44.6% | 14.1% |
Performance against Benchmark
SPX gets destroyed:
| rebal | SPX | |
|---|---|---|
| Sharpe | 94.6% | 27.4% |
| Return | 6.4% | 5.4% |
| Vol | 6.8% | 19.8% |
| MaxD | 11.3% | 56.8% |
| rebal | SPX | |
|---|---|---|
| rebal | 100.0% | 8.1% |
| SPX | 8.1% | 100.0% |
And if we lever the strategies to match the volatility of SPX: 
Appendix
The primary benefit of a risk-parity approach is in its lack of use of model projections: there are no requirements for an estimated forward projection of either returns or volatility, using only backwards looking historical volatility metrics. This avoids issues with overfitting or trained model dependency. Furthermore, weights are much more stable compared to a classic Markowitz Efficient Frontier asset allocation strategy, avoiding high portfolio turnover:
In 2020, the portfolio roughly holds 30% of assets in GLD, 70% in QQQ, and borrows 70% to short SPY, for a total absolute levered notional of 170%.
Other Sub-optimal Portfolio Strategies
XAU + SPX
| SPXXAU_rebal | SPXXAU_static | SPX | |
|---|---|---|---|
| Sharpe | 72.4% | 46.4% | 27.4% |
| Return | 9.2% | 6.6% | 5.4% |
| Vol | 12.7% | 14.2% | 19.8% |
| MaxD | 33.7% | 37.7% | 56.8% |
| SPXXAU_rebal | SPXXAU_static | SPX | |
|---|---|---|---|
| SPXXAU_rebal | 100.0% | 93.5% | 71.4% |
| SPXXAU_static | 93.5% | 100.0% | 86.7% |
| SPX | 71.4% | 86.7% | 100.0% |
