Great XLU/XLB call. Interestingly enough, XLK is much stronger than XLB today. Trust #1 while #2 and #3 are not as dependent/stable and can easily switch places?
The equities trading model I've developed to trade 2006-2011 was based on fundamentals/value filter, then ranked the filtered set by momentum to construct the portfolio. Turnover averaged a quarter. It worked amazingly well in straight up markets but crashed badly otherwise. In retrospect, I'd say the problem was that by the time the momentum pick was selected, it was too "late", ie too close to the point where downside risk was greater than upward potential. I've never tried a counter-momentum ranking formula tho...
Agreed that the problem with momentum is that it is often too late. But there is the added problem that with all of the massive intervention, by the time a momentum model discovers a trend, the Fed (or some other central banker) has discovered it too. So especially lately momentum models are getting whipsawed to death.
My own model isn't so much counter momentum as simply buying dips. The market goes up 2/3s of the time, so buying dips should work twice as often as it doesn't.
Haven't looked into Nasdaq vs XLK. Been under the weather this week and I haven't even done a weekly update yet!
Have you looked at combining position sizing with the sector switching model, e.g. attaching a confidence or magnitude level to the "dip" for the selected XL* component to optimize the projected "bounce"?
Setting an upper bound to the position size is certainly prudent. My idea is actually on the other end whereby if, say, XLU is now in the 1st position but the indicator says the dip is rather shallow, it would indicate establishing a partial position with the expectation that a chance for a larger dip is likely to up the position size later.
That's an interesting idea, but the factor is still too much. If the over-leverage can hit 600%, for instance, then a compensating under-leverage would be 1/6th. That would be on a static price only dip measure.
Using my own internal measures, I've tried to forecast expected returns based on their actual scores, but the score itself was not very significant. Only the relative position rank to the other sectors worked, and so 1st place is just 1st place, without any kind of way to frame a position size.
If this were PART of a portfolio, though, it might work. That is, if I were using different trading strategies in a single portfolio (like effective volume or validea models in conjunction with my own sector model, then I could allocate 1/6th of the portfolio to a sector, and then hold the maximum value of the sector model's dollar size. That is, if I had 60,000, and 1/6th were 10,000, then the next time I rotated I would rotate EITHER to 10,000 (to negate the dip), OR to whatever value the sector portion had reached. So if I sold XLU at a loss of 10%, I'd put 10,000 into the next trade. Or if I sold XLU at a profit of 10%, I'd put 11,000 into the next trade.
That DOES outperform, but only if it's just a portion of the portfolio.
Great XLU/XLB call. Interestingly enough, XLK is much stronger than XLB today. Trust #1 while #2 and #3 are not as dependent/stable and can easily switch places?
ReplyDeleteI've been conveyor belting it myself: buying the first position, holding through the second, and selling in the third.
ReplyDeleteSo, today I'm just in XLU, happily.
Out of curiosity, how is XLI ranked? It's been a rather steady climber as more positive economic news have been coming lately.
ReplyDelete9th place: Dead last, in full momentum mode.
ReplyDeleteThat's why I don't use the model in reverse. You'll get crushed that way.
The equities trading model I've developed to trade 2006-2011 was based on fundamentals/value filter, then ranked the filtered set by momentum to construct the portfolio. Turnover averaged a quarter. It worked amazingly well in straight up markets but crashed badly otherwise. In retrospect, I'd say the problem was that by the time the momentum pick was selected, it was too "late", ie too close to the point where downside risk was greater than upward potential. I've never tried a counter-momentum ranking formula tho...
ReplyDeletebtw Do you find it peculiar that Nasdaq is +0.27% today but XLK (technology) is down -0.38% ?
ReplyDeleteAgreed that the problem with momentum is that it is often too late. But there is the added problem that with all of the massive intervention, by the time a momentum model discovers a trend, the Fed (or some other central banker) has discovered it too. So especially lately momentum models are getting whipsawed to death.
ReplyDeleteMy own model isn't so much counter momentum as simply buying dips. The market goes up 2/3s of the time, so buying dips should work twice as often as it doesn't.
Haven't looked into Nasdaq vs XLK. Been under the weather this week and I haven't even done a weekly update yet!
Have you looked at combining position sizing with the sector switching model, e.g. attaching a confidence or magnitude level to the "dip" for the selected XL* component to optimize the projected "bounce"?
ReplyDeleteYes, but in a crash you would end up over-leveraged.
ReplyDeleteSetting an upper bound to the position size is certainly prudent. My idea is actually on the other end whereby if, say, XLU is now in the 1st position but the indicator says the dip is rather shallow, it would indicate establishing a partial position with the expectation that a chance for a larger dip is likely to up the position size later.
ReplyDeleteThat's an interesting idea, but the factor is still too much. If the over-leverage can hit 600%, for instance, then a compensating under-leverage would be 1/6th. That would be on a static price only dip measure.
ReplyDeleteUsing my own internal measures, I've tried to forecast expected returns based on their actual scores, but the score itself was not very significant. Only the relative position rank to the other sectors worked, and so 1st place is just 1st place, without any kind of way to frame a position size.
If this were PART of a portfolio, though, it might work. That is, if I were using different trading strategies in a single portfolio (like effective volume or validea models in conjunction with my own sector model, then I could allocate 1/6th of the portfolio to a sector, and then hold the maximum value of the sector model's dollar size. That is, if I had 60,000, and 1/6th were 10,000, then the next time I rotated I would rotate EITHER to 10,000 (to negate the dip), OR to whatever value the sector portion had reached. So if I sold XLU at a loss of 10%, I'd put 10,000 into the next trade. Or if I sold XLU at a profit of 10%, I'd put 11,000 into the next trade.
That DOES outperform, but only if it's just a portion of the portfolio.