mercoledì 15 dicembre 2010

Le strategie di trend-following e i mercati post-crisi finanziaria: l'esempio del Decision Moose di Bill Dirlam

I lettori più fedeli di Alfa o Beta? conoscono già il sito Decision Moose di Bill Dirlam: in un post dell'aprile scorso riprodussi un suo commento dedicato al tema dei derivati e della loro regolamentazione. Decision Moose è in realtà soprattutto un sito diventato celebre per una strategia di asset allocation tattica fondata sul momento che ha avuto uno straordinario successo: online dal 30 agosto 1996 ha garantito ai suoi fedeli un rendimento complessivo (teorico, cioè al lordo di tasse, costi di transazione ecc. ecc.) del 2415% attraversando pressochè indenne fasi di turbolenze dei mercati come la crisi dell'estate 1998, lo scoppio della bolla tecnologica, l'11 settembre, la crisi finanziaria del 2008 e così via (qui potete trovare la serie storica di tutte le transazioni di questo portafoglio mentre qui trovate molte più informazioni sul modello e su come è stato costruito).
Il commento settimanale di Dirlam è una delle mie letture preferite della domenica sera: il commento di questa settimana è dedicato alla performance a dir poco deludente della strategia negli ultimi 24 mesi.
Con l'autorizzazione dell'autore lo riproduco qui sotto

Not Fighting the Fed While it Fights Me

Have to admit that the last two years have been very humbling for the Moose. In prior years, it regularly provided a CAGR in the 25-35% range. It caught the 2008 meltdown before it happened and avoided the 45% losses many buy-and-hold equity investors suffered. Lately however, it has underperformed the S&P by a rather frustrating margin.
I've been asking myself why that might be. What, if anything, could I do to improve the model's near term results-- without jeopardizing the aspects of the construct that have provided its long term success? What is different about this cycle-- beyond the fact that government has been roiling the markets for two years?
The three keys to the model's long term success are that it doesn't fight the market; it doesn't fight the tape; and it doesn't fight the Fed. The mathematical reflections of those keys are (1) relative strength among the assets in the model, (2) a technical analysis of each asset, and (3) the Fed Check.

Il Fed Check di cui si parla è un indicatore tecnico basato sul confronto dell'andamento dei prezzi delle obbligazioni del Tesoro USA con l'andamento dei prezzi delle materie prime che cerca di prevedere in qualche modo le prossime mosse della politica monetaria USA. Nelle parole del suo creatore
The Fed Check reveals what the market thinks the Fed should do-- not to be confused with what it WILL do. The indicator is a ratio between Treasury bond prices and commodity prices, examined over the medium term. 
Quando il rapporto si allontana consideravolmente da 1 in un senso o nell'altro è spesso indicativo di interventi sui tassi di interesse in senso restrittivo (quando le materie prime crescono comparativamente troppo in fretta, anticipando futuri trend inflazionistici) o al contrario più accomodanti (quando invece sono i prezzi delle obbligazioni a crescere troppo in fretta, abbassando i tassi di interesse e segnalando la necessità di tagliare i tassi da parte della Fed, un segnale solitamente fortemente rialzista per i mercati azionari)

L'analisi settimanale di Dirlam prosegue con alcune considerazioni tratte dall'esperienza degli ultimi due anni che possono però interessare chiunque cerchi di costruire modelli quantitativi di asset allocation...


 Exogenous government intrusions into the marketplace that truncate trends and increase both unpredictability and volatility are outside the control of any trend-following model builder. I must confine my search to issues that I can address. In that respect, over the past two years, the Fed Check seems to have been the primary culprit in retarding Moose profits. The Fed Check has kept us out of stocks when we should have been in, and put us into bonds when we should have steered clear.
For those of you who are into building your own models, there is a lesson here. That lesson is that every component in a good model requires a very long data history. Problem with exchange traded funds is that the data histories are still relatively short. When the moose was being developed back in the late 80s, the technical component was adopted based on a 70-year time frame, and the the relative
strength component was adopted based on a 50-year timeframe. I only used about a 20-year time frame for back-testing the Fed component, however. That has proved to be a mistake.
During the 70s and 80s, stock prices and bond prices were positively correlated. In other words, they moved in tandem, rising and falling together. They were complementary. Since the 2000–2003 meltdown, however, stocks and bonds have become negatively correlated.
They have become substitutes for one another. When stocks go down, bond prices go up, and vice versa.
Substitution is the traditional view of the stock-bond relationship that devolved during the 30s, 40s, and 50s. The major difference is that the 70s and 80s were highly inflationary compared to the earlier era, during which we were fighting depression-era deflation prior to fighting a world war. Now that we are in another “deflationary” period, the traditional relationship has returned.
The Fed Check's rule to avoid stocks when bond prices are collapsing relative to commodity prices, then, is the result of an initial analysis that only considered an abnormally inflationary environment. In a disinflationary cycle, such as the one that began around 2001, the rule began to work only sporadically, so I adjusted the model, to account for the short term technicals in stocks before avoiding them. In a deflationary cycle-- like the one we've experienced over the past two years-- it hasn't worked at all.
Now that I have figured out the problem and why it may be happening, I guess it's time to go back to the drawing board on the Fed Check.
Meanwhile, the model holds US small-cap stocks this week, even while the Fed Check continues to deteriorate. I still perceive IWM to be the most attractive equity option for reasons expressed in last week's “author's take”. Also, Treasury bonds are extremely bearish. Without some meltdown in China or Europe that threatens the global economy, bonds' bull run is over. Money is exiting bonds and going elsewhere. For those who might prefer a more diversified approach, yes, equities are behaving as a substitute, but so are gold and commodities.

Nessun commento: