S&P 500 Performance Relative to Gold
Michael Panzner did an article today about whether the S&P's gains mean anything in real dollar value, and I decided to take a quick stab at a way of visualizing this. Therefore, I've charted here the relative performance of the streetTracks gold ETF and the S&P tracker (GLD and SPY). This is the cumulative sum of the daily difference in log-return over the past 100 days, and as you can see, despite the S&P's impressive summer, it has still lost value relative to gold. As an example, yesterday's post-FOMC S&P gain led to a log-return of 0.029. After adjusting for gold as numeraire, the return is nearly halved to 0.0188.
- Michael J Bommarito II's blog
- Add new comment
- 1972 reads
Mike, you wrote: "This is
Mike, you wrote: "This is the cumulative sum of the daily difference in log-return over the past 100 days,"
I implemented this as follows:
1. For SPY, I calculated the log-return for 1 bar, namely LN of price[Bar] minus LN pf price[Bar -1]
2. Same for GLD
3. Subtracted value obtained in step 2 from value in step 1.
4. Summed the value obtained in step 3 over last 100 bars.
Did I do it the right way? If not, can you correct me please?
Because GLD and SPY
Because GLD and SPY often move pre-market and afterhours, using the log-return of a single bar will result in a loss of all of the gap-up and dividend information. I used adjusted close to adjusted close for this comparison. Other than that, your process looks identical to mine.
Actually, I created an
Actually, I created an entire price series along the steps indicated above; my use of the "per-bar" process was merely for sake of clarity. And I do use split and dividend adjustments to end-of-day data.
My values were coming out a bit different, hence my Q. In particular, your chart shows that after the trough in May-06, the highest peak was between -0.05 - -0.1, around Oct or Nov '06, whereas I got a peak at that time around +0.26, which is higher than the highest value I see in your chart (time frame: Sep-05 to Sep-07). I can only assume this is due to differences in data since our process appears to be identical. (I don't use Yahoo data, which I think you use).
I forgot to add in my
I forgot to add in my previous post: If you replace SPY with .SPX (or its Yahoo eqvt ^SPX; S&P500 index), does your chart look any different?
Mike, the Yahoo finance
Mike, the Yahoo finance symbol for S&P500 index is ^GSPC, I just realized. Can you post a chart of that (or whatever symbol you use for the S&P500 index) relative to GLD?