SPY vs S&P 500 Correlation Near All-Time High

03Jun10

Random observation for the day…

Below is the rolling 63-day (grey) and 252-day (red) correlation between daily % changes for SPY and the S&P 500 Index from early 1994.

Over the last year or so correlation between the two has been at all-time highs.

This may seem like a minor point, but fine differences in the performance of an ETF versus it’s underlying index is one reason why very active strategies (like the ones in our recent test of five short-term indicators) behave differently depending on the specific asset traded.

We trade leveraged mutual funds which more closely track the index itself, but for ETF traders this might have some significance.

If this trend continues it would indicate that, transaction costs and slippage aside, SPY is becoming a better and better closing price proxy for the S&P 500.

Happy Trading,
ms

Geek note: I’ve used daily dividend-adjusted returns for both the S&P 500 Index and SPY. Dividends for the index were interpolated from quarterly data.

. . . . .

To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.



3 Responses to “SPY vs S&P 500 Correlation Near All-Time High”

  1. 1 Blue

    An interesting point. At first I thought that it might not make that much difference if the ETF didn’t track the index very closely as long as it came out right over a longer period. The error is just as likely to be for you as against you on each day and in the long run it would all come out ok.

    But the problem is that the return on the strategy would have a higher uncertainty. Suppose the ETF were a random walk for n-1 of n days and then jumped to the right price on the nth day. During the n-1 days the strategy would break even (presumably). On the nth day it would do fantastically well or fantastically poorly.

    Over a lot of periods like this the strategy would presumably have more or less the same average as one that actually tracked the index. But the variability would kill the Sharpe ratio.

    With that thought in mind and since you trade on a daily basis it would be interesting to see the correlation on a day-to-day basis rather than just 63 days and 252.

    • 2 MarketSci

      RE to Blue: one small response to your comment above. The correlation is on a daily basis (daily % changes), the 63 and 252-days just specify over what period to look at those daily changes. In other words, I’m looking at the previous 63 and 252 days of correlation between 1-day % changes. michael

  2. 3 Carl

    More interesting would be the correlation to the Emini or full S&P contract (which as we know carry valuation delta’s with the actual cash index), which I think would lead us to speculate on whether this is an arbitrage strategy that is getting crowded, and could in turn help explain part of the large increase in SPY volume over those years of significantly increasing correlation. Would also be interesting to see if the nocturnal correlations are doing anything… yes, geekery!


Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s