Sunday 12 July 2009

Banks, Energy, Nikkei Could Bounce

Looks like some major breakdowns in the markets this past week. Seems almost predestined that we're headed for a retest of the March lows - and maybe lower. But behold the news from Friday afternoon's Commitments of Traders report. It's actually quite bullish on some fronts. Crude oil and gold still look under pressure for now, but the banks, natural gas and, to some extent, the S&P 500 data are looking up. And in crude, my trading setup has turned bullish for mid-August.

Here are a few highlights from the data that I've just updated on my latest signals table.

- S&P 500: The "smart money" commercial hedgers have shot up to 2.43 standard deviations above the moving average in their net futures and options positioning as a portion of the total open interest. They haven't been this bullish compared to the past data since the end of Sept. 2007. The latest COT report shows them at their 12th most bullish positioning since the beginning of the data in 1995.

That's positive for the bulls. The caveat is that the folks I'm fading in this market, the wrong-way S&P 500 small traders, hit an extreme of bullishness the previous week - oops, bad timing, guys! - and they have yet to work off their excessive net long positioning this week. They've got to do that first before my setup can turn bullish in this market.

- Natural gas: This setup is bullish as of Monday's open and will remain so for at least two weeks. Nice timing, what with the massacre last week. People ask about the trade delays built into my trading setups based on the COT data. Both signals in this setup - based on fading the large spec net position and their total open interest - turned bullish on me the Friday before last. But there are trade delays for both signals. If I had acted on those signals this past week, I'd have been toast. The backtesting shows there's sometimes good reason to wait before acting on major moves in the COT data. I think the fact that markets don't react to the data right away - or don't do so right away in the expected direction - leads to confusion about the usefulness of the data. This last week in this market was a good real-time example. Thanks, backtesting.

- U.S. banks: My trading setup for the BKX U.S. Bank Index - based on the three-month Eurodollar contract (the interest rate, not the currency) - has gone bullish for Monday's open.

- Nikkei: This setup is still long - its sixth straight week - and now finds the small traders - whom I trade alongside in this market; no, they're not always the "dumb money" contrary to popular wisdom - boosting their net long position to an astonishing 4.08 standard deviations above the mean as a percentage of the total open interest. That's by far the most relatively bullish they've ever been. Wow. Something big could be brewing in this market over the coming weeks. Note the small traders operate with an eight-week trade delay in my Nikkei setup. So their extreme of positioning could take a little time to make itself felt. But when it happens, it could be a surprise.

-Crude oil: This setup has gone bullish, but not right away. Execution will be on the open of trading Monday, Aug. 10. See more details on the latest signals table.

Be sure to check in for an update of my portfolio page Monday. Good luck this week!

7 comments:

dave said...

S&P 500: "They haven't been this bullish compared to the past data since the end of Sept. 2007. ...

That's positive for the bulls."

But, the stk mkt made a major top on Oct 10, 2007.

Regards,

dave

Alex Roslin said...

Hi Dave,

You are right. In fact, due to the trade delays in that SP500 setup, the signal got actually bearish at that top.

The highly commercial positioning then flipped the setup to bullish as the market moved sideways for a while; the small traders at that time had also hit a major extreme of bearishness. That bullish signal lasted nine weeks and lost 0.9%. A good number of these signals don't make money. It's all about probabilities.

The point in my comment was it's been a year and a half since the commercial traders have been this bullish - not that the signal works out every time.

Regards,
Alex

In Debt We Trust said...

One thing that jumped out at me was the massive increase in short positions for grains - wheat, soy, BUT ESPECIALLY corn between last Friday and the week before that.

I have positions that expire in the fall so I'm not actively trading these days. Still, I couldn't help but notice that the shorts got squeezed the last few days. Argh, if I had access to this data Friday afternoon I would've bought some calls as a contrarian trade.

Alex Roslin said...

Hi In Debt...,

I hope to get to some ag commodities at some point. But just be aware that, as with the other markets, a big shift in positioning in a trader group doesn't necessarily mean anything. It's difficult to combine the COT data with technical trading if you haven't really tested the data. It sometimes doesn't mean what you think. I've often seen a trader group that we expect to be the "dumb money" - like the large speculators - actually be best to trade alongside. In other markets, one trader group may provide highly useful signals, while the others are absolutely useless in any timeframe. Similarly, it's hard to gauge when a position truly has reached a meaningful extreme without really studying the past data in a systematic way. And I don't mean just eyeballing a few recent moves against a price chart. Moves in the data can mislead.

Regards,
Alex

Alex Roslin said...

Hi again,

I should add something else. This also flies against the common wisdom when it comes to the COT data. A single move in one week typically doesn't mean anything for prices the following week. One of the simplest studies you can do with the data is line it up with the price data and test for correlations the same week, a week out, two weeks out, etc. You will find the correlations are virtually non-existent - or at best far too weak to trade off.

In fact, the strongest correlations I've ever found of week-to-week COT-versus-price changes work in reverse: the COT data one week correlated with price changes several months before. So in those cases, it's the price move that could in fact predict the COT data, not the other way around.

Regards,
Alex

In Debt We Trust said...

Can you give an example?

"In fact, the strongest correlations I've ever found of week-to-week COT-versus-price changes work in reverse: the COT data one week correlated with price changes several months before. So in those cases, it's the price move that could in fact predict the COT data, not the other way around."

PS Nice call on the banks. It was also Government Sachs earnings week, so it was little surprise that they blew out numbers.

Alex Roslin said...

Hi In Debt...,

I can't readily find that data. It was a couple of years ago that I did those calculations. I can't recall the exact correlated pairs, but the highest correlations were in to 60-70-percent range. They're easy enough to look for. Just get the net COT positioning and apply the correlation function in Excel to that data and the price data. I'll report back if and when I find what I was doing back then.

Regards,
Alex