Tonight in the Trading Psychology Weblog, I'll post an interesting chart that shows how the SPY ETF is priced relative to the cash S&P index. The gist of that post is that the ETF is much more involved in index arbitrage than is commonly acknowledged.
What triggered this line of research was my trading experience on Friday. I shorted SPY in the AM, testing out a trade idea. To my dismay, the market ground higher for quite a few minutes with unusually strong spikes in the NYSE TICK. Some information that I follow suggested to me that a significant proportion of this buying in stocks, stimulated by upside runs in the S&P emini futures, was being laid off elsewhere. The buying didn't strike me as "real"; it was as if I was looking at only one side of a larger arb trade. As it happened, sticking with the short position was the right strategy, as we steadily moved lower into the mid morning once these buying flurries could no longer bring higher price highs.
Interestingly, the Money Flow for SPY was negative, even as the market was moving higher on Friday. Recall that Money Flow is a cumulative volume measure that adds the volume to the total when a trade occurs at a higher price and deducts it from the total when a trade occurs at a lower price. Normally, of course, you'd expect the Money Flow to be positive during rising markets and negative during falling ones. Not so. As I stress in the Weblog, the daily correlation between Money Flow and SPY price change is essentially zero.
I believe this is the case because large traders are selling SPY in size when buying ES and vice versa. This arb trade is cancelling out the normal positive correlation between Money Flow and price that results from purely directional trading.
Hmmm... If we track Money Flow of SPY, perhaps we can discern when price movement is dominated by an arb trade (and hence unlikely to continue) vs. when it is part of a broader directional trade (and thus likely to continue).
Here's a small piece of confirmation of this reasoning:
Since March, 2003, we've had 98 occasions in which SPY has been up .80% or more on an open-to-close basis. When SPY Money Flow has been strong during this rise (N = 49), the next day in SPY (close to close) averages a gain of .17% (32 up, 17 down). When SPY Money Flow has been weak during the rise (N = 49), the next day in SPY averages a loss of -.08% (26 up, 23 down).
My experience on Friday might not have been a fluke. When buying in SPY is accompanied by weak Money Flow, the odds of continuation are worse than if the buying is accompanied by strong Money Flow. I believe this is because we are indirectly measuring the relative dominance of arb vs. directional trade when we look at Money Flow relative to price.
The buying while I was initially short SPY did not feel real to me. Perhaps it really was only one-half of a larger trade. Whether or not this idea ultimately pans out, I hope it serves as an example of how trading can lead to market insights, which lead to new ideas, which can generate better trading. Understanding markets is as important as predicting them.