Tuesday 23 September 2008

Trader Positioning Foretold Problems

How much of a surprise was the latest installment of the Wall St. horror picture show? The Commitments of Traders data has been signaling a huge problem for months. Trader positioning in the 30-day Fed Funds futures contract hit bullish extremes back in May (meaning they expected lower interest rates), just as markets were powering upward, crude was topping $130 (on the march to $145) and everyone was talking about how inflation was the true problem, not deflation.

In the midst of all this euphoria, the large speculators (the big investment firms and hedge funds, including possibly some of the ones now in massive trouble) knew there was a big problem and were positioning themselves accordingly. The large specs in Fed Funds built an extreme net long position as a percentage of the total open interest, and they maintained their extreme positioning for an unprecedented 14 straight weeks. That's longer than the 11 and 12 weeks they were at extreme levels during the dot-com fiasco in 2001 and 2003, as determined by my trading setup for the contract, which trades when the large specs hit extreme levels. At the time, absolutely no one was talking about the possibility of the Fed actually lowering rates. The speculation was about how soon Bernanke & Co. would start hiking. Fast forward to today, when lower rates have become a possibility or at least talk has faded about higher rates. Even now, with the extraordinary $700-billion bailout that's being discussed (more on that here), my Fed Funds setup maintains a slight bullish tilt, with the large specs sitting at 0.3 standard deviations above the moving average for this setup. Which means we're far from an all-clear.

Portfolio update: I've just updated my portfolio page with my latest positioning and closed trades results from Monday. Yeah, the results ain't pretty, but hey, that's trading. Even the best traders lose 40 percent of the time. What's important is to maintain discipline and use proper risk control so I can keep going. Good luck this week!

TAGS: Fed Funds, Bernanke, COT, Commitments of Traders, market timing, trading system development, CFTC, Commodity Futures Trading Commission, COTs Timer, out-of-sample testing, walk-around testing

6 comments:

Anonymous said...

what seems to be happening is that the signals are right, but timing is wrong. it happened two times with the financials already; if it happens also for a third time now, then maybe you should re-think your risk control rules. how could you possibly expect to get the timing right, when you get a new signal each week yet you trade only once weekly? maybe conflicting signals that are too close to each other in time should simply not be traded, in order to avoid risk.

Alex Roslin said...

Hi Anonymous,

Thanks for your thoughts. A couple of missed trades are nothing abnormal in trading. Take a look at the wins/losses column on my latest signals table for any of the setups. Those are pretty normal results for trading of any kind, discretionary or mechanical.

I don't really follow your point about not getting the timing right "when you get a new signal each week yet you trade only once weekly." Yes, the new data comes each week. My average setup trades at a frequency of once every four weeks or so, but more frequent trades are quite frequent in the backtesting. Avoiding frequent signals wouldn't work as it would mean staying in a trade despite an opposing signal. Doesn't make sense.

As for avoiding risk altogether, alas, I don't think this is possible, even if you're in cash (due to currency fluctuations) or money-market funds (as we've seen lately).

Regards,
Alex

Anonymous said...

hi again,

what i meant was that if you got a buy signal for this week and a sell signal for next week for the same security, maybe you should not follow the buy signal because it's not possible to get the timing right. COTs come only once a week so you can only trade on the signals once a week - but when? not necessarily on monday, as we recently saw.

Alex Roslin said...

Hi Anonymous,

Thanks for your message. The rationale behind the trade delays is that it may take time for trader positioning to reflect itself in the market. Regardless, the fact is there appears to be a notable and robust improvement in some results with the delay.

It might be interesting to study extra rules of the kind you're talking about, but doing so adds to the risk of curve-fitting and tends to reduce the statistical reliability of a trading setup, even if it improves performance, which is far from certain. Anyone could apply such a rule yourself to the S&P 500 spreadsheet supplied on my DIY page and see if it does improve results.

Regards,
Alex

Anonymous said...

Thanks genius....why did you not alert us of this in May.

Everyone is an expert in hindsight.

Alex Roslin said...

Well, for one thing, May was when the extreme positioning started, so there wasn't much to report at that point, was there? As the situation developed, however, I posted about it several times here:

http://cotstimer.blogspot.com/2008/07/bernanke-will-not-hike-crude-bubble-not.html

and:

http://cotstimer.blogspot.com/2008/08/trader-data-smiles-on-equities.html

Alex