Correlations: At the heart of systematic strategies

The use of technology (IA, algorithm) highlights correlations between type of assets and various elements of the economic context. Phiadvisor Valquant is investing massively in this direction to improve the performance of its indicators.

Correlations between assets and factors are commonly used in management/trading, with the aim of taking advantage of statistical data. The main difficulty in their use, in addition to the modeling requirements, comes from the sometimes limited validity period of these correlations and the fact that an asset most often are subject to multiple correlations and in varying proportions.

These correlations most often occur in relation to the evolution of the security's reference index, its sector, its reference currency, various macroeconomic indicators or even geopolitics. For example, an American semiconductor company will be influenced by the evolution of the Nasdaq, tensions between the USA and China, the level of long-term bond yields, etc.

The systematic mathematical use of multiple correlations implies automated management which will be the more often oriented towards the short term with frequent switches between the respective weights of the factors. This approach is one of the strategies favored by Hedge Funds.

The discretionary manager will also be interested in correlations based on a few long-term factors whose evolution he will carefully monitor. For example, the surge in long-term rates historically favors sectors such as Insurance and Banks and disadvantages Utilities and Real Estate. Anticipation of the evolution of these factors will be key for discretionary management.

Volatility and reversal of correlations

Correlations are by definition volatile and change over time. For example, inverse correlations between government bonds and stocks have been observed most of the time over the last decades with the notable exception of periods of inflation (in particular between 1974-80 and 2022-23) during which stocks and bonds are positively correlated. In these atypical periods, other factors can be introduced - such as oil prices, gold prices, or even the evolution of sectors sensitive to long rates whether favorably.

Observation often shows that equity markets will be positively correlated in the first part of the curve to a factor like oil for example - a period where certain sectors benefit from it without harming the macroeconomy - then in the second phase the correlation reverses and becomes negative because the increase in the factor ends up weighing on demand and costs.

The reaction of an asset in relation to a factor is often not linear and there are threshold effects which can be psychological (e.g. a barrel of oil at $100, US long rates at 5%, etc.).

Today the main factor slowing down the equity markets are the sharply rising bond yields, which increase the risk of an incident (bank or real estate bankruptcy), have a negative impact on the financing of companies and households and on growth stocks for which most of the valuation is based on cash flows far into the future.

Will the fall in long-term rates be followed by a “rally” in stocks? If we judge by historical data, this is quite probable initially, then on the contrary in a second phase the fall in bond yields will cause the fall in stocks, because this movement will then correspond to the anticipation of a slowdown in the economy with a classic shift out of stocks and into bonds.

In addition, certain factors are linked together, for example the rise in oil prices can fuel the rise in long-term rates just as they cause the rise in the dollar.

?

What strategies to use correlations?

Some funds exploit disparities in volatility between different stocks or sectors, the idea being to take positions that benefit from changes in relative volatility.

Hedge fund managers can identify price discrepancies between two highly correlated assets. When the correlation between these assets deviates from normal, the fund may buy the undervalued asset and short sell the overvalued asset, betting that the historical correlation will be restored.

As part of momentum/trend following strategies, favored by Phiadvisor, the contribution of correlations can also be very useful for the discretionary manager.

For example, when it comes to observing that a security “can escape gravity” and its natural correlations, a sign of powerful fundamental traction. This was the case for example of Nvidia or Microsoft whose stock market performance in the first part of 2023 reflects impressive prospects in artificial intelligence (AI) despite negative macroeconomic conditions such as the rise in long-term bond yields and tensions with China.

Conversely, a security that is unfavorably uncorrelated with its sector carries a high risk.

?

Lionel Pellicer, Founding Partner Phiadvisor Valquant