Goldman Turns "Tactically" Neutral On Stocks, Believes S&P Not Pricing In "Downshift In Macro Picture", Proposes "Zero Cost Cross-Asset" Hedge For SPX Drop

Tyler Durden's picture

In another key note out of Goldman, we now learn that it is not only Jim O'Neill who is the natural hedge to the firm's other diametrically opposing views (as discussed earlier). While it is no secret that the firm's chief strategist David Kostin continues to ignore warnings from Jan Hatzius et al warning that the economy is set for a period of slower than expected growth, we now find that Goldman's Roman Maranets is out with a note in which he says that "the downshift in the macro picture (excluding Friday’s payrolls print) has not been fully reflected in the level of SPX, prompting us to shift the tactical trading stance in equities to neutral." Hmm, it is somewhat odd that nobody else has noticed that Goldman is now, well, neutral (and we all know what that means in sellside lingo) on the market. Now, the question is whether this is an honest opinion, or merely a way for the firm's prop desk to accumulate ES at the expense of its clients. Judging by the market's sudden surge, asset managers seem to be convinced it is the latter. That said, the firm is now proposing a USD long as a "Zero Cost Cross-Asset Hedge for SPX"  due to the firm's motivation "to find application of our framework in construction of “cheap” cross-asset hedges for investors being long SPX." Of course, why not just buy the USD and the market. We are long past the point where anything makes logical sense any longer courtesy of central planning. We wonder how Thomas Stolper feels now that his thunder has been stolen by both Maranets and Fiotakis, who predicted the EURUSD toptick to within minutes of the near-800 pip plunge in the pair in less than a week.

From Goldman on using FX as an equity hedge:

In May 4th Global Markets Daily, Thomas Stolper highlighted that over the last two years the FX markets became attractive as a cross asset hedge for investors in other risky assets, and in particular that the number of currencies being used for this purpose widened out. As an example, the stability of historically low correlation in daily returns over a 3 month window between SPX and GS trade-weighted US Dollar index (average -60% over last two years) lent support to this argument. In today’s Daily, we revisit a methodology launched in 2007, which allowed us to contrast volatility against carry along a concept of risk/reward for FX portfolios and examine correlation of our optimized baskets versus global equity indices (proxies of systemic risk).

In a nutshell, our analytical framework maximizes carry per unit of volatility across a portfolio of 56 liquid crosses (28 short currency vs. USD and 28 short USD vs. currency). In our earlier empirical work we have demonstrated that higher expected returns are proportional to the level of carry we are receiving, and therefore we use current carry as a proxy of expected total return. Also for the purposes of this study, we use standard deviation of weekly FX returns over the last 52 weeks as a proxy for conditional volatility.

By introducing correlation with global equities as a third criterion in this problem, we showed that by reducing the amount of correlation with equities we would also typically reduce the net carry of our optimized basket. We linked this observation to the fact that higher carry currencies typically have higher correlations to equities and emphasized that carry traders have potential exposure not only to macro fundamentals of the cross, but also global shifts in risk appetite. A few other interesting results we extracted were: 1) Higher carry is generally associated with higher volatility (confirming the risk/reward concept), 2) In a portfolio, one could achieve a reduction in volatility for the same level of carry, by diversifying away the idiosyncratic risk. But most importantly, we have demonstrated that it is feasible to construct portfolios with non-negative carry gains that either have zero or negative correlation to the SPX (which is not straightforward to do in single cross space).

And here is how to hedge an overvalued (not our words, Goldman's):

In May 6th Global Markets Daily, Noah Weisberger had noted that the downshift in the macro picture (excluding Friday’s payrolls print) has not been fully reflected in the level of SPX, prompting us to shift the tactical trading stance in equities to neutral. With that said, on our part we would be motivated to find application of our framework in construction of “cheap” cross-asset hedges for investors being long SPX.

In our research we have often discussed that one of the best ways to reduce correlation to SPX through currencies is to increase USD exposure. However, by capping the volatility of our portfolio at 6% and diversifying away the idiosyncratic risk, our framework (in theory) allows us to create a carry-neutral FX portfolio with lower correlation to SPX than trading a basket of USD TWI would allow: -92% vs. -77%. This portfolio consists of 21 FX positions, which we break up into 10 net short and 12 net long currencies. We observe that this is a fairly diversified FX portfolio, where AUD, CAD and SEK are top three positions in net short portfolio (comprising 61%), while the net long portfolio is fairly distributed among six currencies: USD (21%), IDR (14%), CHF (13%), BRL (11%), JPY(10%) and GBP(9%).

These results are not entirely surprising. First of all, carry funding currencies such as JPY, CHF and USD have historically had negative correlations with equities, while currencies with commodity exposure, such as AUD, CAD and SEK, have had positive. Since the correlations between equity and FX returns have typically been very unstable and tended to change over time, there is no systematic way to pin down why these correlations kick in and for how long they will persist. Nevertheless, by making some reasonable assumptions about what these correlations reflect, carry trades tend to be driven by the same cross-asset sentiment for risk taking that also helps strong equity performance. Therefore, it is not unreasonable to expect that funding currencies like JPY, CHF and USD are negatively correlated with global equities, while high-beta currencies such as AUD, CAD and SEK are positively correlated.

So how would the proposed carry-correlation trade off look like?

As mentioned above, in the past we demonstrated that a reduction in correlation with SPX comes with a sacrifice of carry gains. We analyze the carry/correlation trade-off by looking in detail at optimized FX portfolios with positive and neutral correlations to SPX.

In our attempt to construct FX portfolios with positive and neutral correlations to SPX, we target investors with two motivations: 1) receiving the maximum carry for a specific level of risk (in our case 6% annualized vol), and 2) the same as 1) but also diversifying away systemic risk at the same time.

First, the positively correlated portfolio (+52% correlation) consists of five FX positions, which we break up into 3 net short and 3 net long currencies. The weights analysis suggests that this is a fairly concentrated portfolio in two currencies: TWD (61%) on the short side and INR (79%) on the long side. These results intuitively make sense given that long INR/TWD cross has demonstrated solid total return (including carry) among EM’s over time (70% since 1998). Our uncorrelated portfolio with SPX is slightly more diversified with USD (46%) and TWD (42%) as the two largest positions on the short side, and INR (43%), CHF (33%) and JPY (13%) as three largest on the long side.

From carry perspective, we highlight that we gave up 2.1% of carry to diversify away the systemic risk and arrive from positively correlated portfolio (4.8% carry) to uncorrelated (2.7% carry). To scratch the surface of this finding, we statistically analyze a sample of correlation and carry data from a full set of our optimized baskets. These results are quite striking as they imply that the relationship between carry and correlation is fairly linear and has a solid fit given the 98% R-squared. A coefficient of 0.03 suggests that a 1% decrease in correlation to SPX would typically correspond to a 3bp reduction in net carry of our portfolio.

As usual the stated message is secondary. The key theme is that Goldman is likely about to turn full blown bearish on stocks with an official stamp of (dis)approval coming imminently from David Kostin. As to whether Goldman prop is positioned appropriately already, we think that, as always, this is a rhetorical question.