How about that - the highest ranked stocks for 4th gen (a long system) got the worst returns, and the ranking designed to pick short candidates actually had positive returns in the top buckets!
It seems to be a topsy turvy world where everything is operating back to front. Has anyone else noticed this?
---------------------------------------- recent4thgen.png
(24679 bytes)
(Download Count: 189)
Sadly, yes, I have been noticing this for a few months. There seems to be, indeed, a perverse behavior in almost every system I have been tracking (many, many). I am not very surprised. This is quite typical for factor based systems. They almost always miss the next market turn, and then crash and burn. After quite a few years researching (and investing) these systems, I concluded that they are little more than linear extrapolators. In other words, they just tell us what has been working in the past. Then the market turns . . . In the very, very long term (multidecades), it is clear that some factor provide outperformance, say, momentum and value, on the average. From a practical trading point of view, the catch is how to survive the times when they do not. A few weeks of suffering is one thing. Months, perhaps a year or more, is kind of difficult to survive. Martin
Are you saying this is a short term phenomenon like bad weather, or is this a siesmic shift in the market? Basically, is it time to completely rethink the strategies or should one ride out the storm?
One hypothesis I have: Perhaps we are witnessing the effect of hedge fund liquidation:- have they been using similiar strategies as us? Does their unwinding of positions now cause share prices to react in the way we observe? In which case, would it be useful to know the aggregate flows into and out of hedge funds as a possible indicator of trouble?
According to Pimco, the "shadow" banking system will continue to delever. This delevering will cause a liquation of every asset that can be easily marked to market instead of marked to model.
Liquid investments will continue to be liquidated regardless of whether or not they are good investments.
I would expect the factor models to continue showing perverse behavior for as long as the deleveraging continues.
I think that the Eurodollar / OIS or the spread between corporates and treasuries may be good indicators of the crisis or deleveraging pressure.
It is a change in market behavior, call it the "price production mechanism". These changes happens with some frequency (once every few years) at random times. Are they seismic ? Well, yes, if we are invested. Completely rethink the strategies ? That would imply going agains the factors we know have worked for a long time. Take momemtum, for example. You will notice it has not worked for the last few months. Does it mean it will never work again ? I don't think so. How long will it take to work again ? I have no idea, and that is the problem. I guess the only viable answer is to ride the storm. But we do not know how long the storm will last . . . or even if we will survive the storm. It has been advocated many times to follow some kind of indicator that takes one out of the market at times like this. I still have to see one that works with any reliability. Additionally, value momemtum stocks tend to take off before the general market. This market shift has, of course, some reason behind it. Yes, it can be hedge funds reversing their carry trade: long commodities, short financials. Or dumping their holdings if they were not fast enough. The point is that at every market shift there is always some reason that explains the shift . . . in retrospect. No way to know the reason (or the timing) of the next one. I know I am sounding very agnostic ( = impractical ). Unfortunately, I have never been able to find a factor based system that can ride these market shifts/turns or whatever we want to call them. Marting
Some interesting comments, I have decided to deconstruct the ranking and test some factors individually over the last 3 months.
A while ago I posted about factor reversal. This seems to be the same effect on a massive scale. I wonder if the original assesment was a "canary in the mine".
I have posted the pictures that show the result of short interest, momentum, and value over the last 3 months. In each case, you can see a clear reversal from "normal" behaviour. One wonders to what extent this is short covering. For sure the "short" factor has shown strong reversal, but the momentum has reversed very strongly as well.
Finally, I have appended the picture showing the Price2Sales factor, that I originally mentioned had reversed. Interestingly, it seems to have performed the "right way round" over the prior 3 months. hmm.
---------------------------------------- recentshorting.png
(24327 bytes)
(Download Count: 159)
I truly beleive this is delveraging. It's the only thing that makes sense. It would be one thing if the factors that have worked for so long had a period of underperformance where they only match or slightly lag the overall market, but this is a complete reversal. This is evident from Olikea's charts. All the garbage stocks out there, homebuilders, financials etc. are up too. Longs are being sold, shorts are being covered. At some point fundamentals will matter again. There is a lot of leverage to be worked out of the system, so it may take time. But to suggest that buying companies with low multiples of cashflow with good prospects and momentum is not going to work in the future sounds a little ridiculous to me.
p.s. I think Olikea's observation about P/S still working is because the short covering in some beaten down financials with seemingly low P/S ratios. Try running that factor with financials restricted.
pps. my short portfolio, which has done well in the past year was up 11% Thursday. I.E. I lost 11% there. GM was one stock, up almost 30%. GM is not a buy it is massive short squeeze!
Assuming that the factor reversal is caused by delevering, what factors either stock or market can act as leading or concurrent indicators of a crisis?
Is this data available in Port123 or do we need to look elsewhere for the market data?
An alternative would be to look at factor reversals. Is the factor reversal itself the signal?
Does anyone have a method that works?
Personally the only timing that has helped me until now has been the market timing system of Vector Vest. I consider Port123 as superior in every respect except this one aspect. The timing signals have kept me out of the market or encouraged me to short.
We had to get out of our systems and models around mid year and i have been tracking them in real time since then. They are inverse. If we had gone long the short model and short the long model our performance would be off the charts.
I suspect this is due to massive de-leveraging of quant based funds but don't know for sure.
Volatility has been so high there were days our long models under performed by 500 basis points on 20 stocks.
