Bubble Trouble in Little China

Michael Martin (Broken Symmetry) discussed Didier Sornette group’s prediction that the Shanghai Composite was a bubble and would pop between July 17^th and July 27^th.  The prediction was a few days off, but the lines fit so we must acquit… at least to (by) a degree (insert line about OJ here).

I applaud the effort of attempting to model the path of stock prices reflecting elements of feedback that undoubtedly will rub many traditional economists the wrong way.  Even if predictive abilities are somewhat limited, it is nonetheless highly valuable to be able to identify a bubble as a matter of public policy.  What we shall do at that point and whether accurate prediction can only be done in the final stages of such a bubble (“Minority Report” anyone?) are interesting tangents, but I shall save them for another day.

Sustainability of supernatural growth. 
While the premise of un-sustainability of supernatural growth is correct, the predictive abilities are limited, because the occurrence of exogenous events will limit any historic value of line fitting.  Sornette differentiates between exogenous and endogenous crashes:

“…The existence of exogenous crashes, that is, genuine surprises that can move the market significantly, leads to an intrinsic limitation of the predictability of crashes. This seems to be the unavoidable lot of complex systems that are open to the outside, i.e., that are subjected to a complicated flux of “news.” …” (Source)

 And (empirical data) here:

“Limitations of the LPPL. Using the LPPL model, Johansen and Sornette [2003] have performed a systematic classification of drawdowns in the two leading exchange markets (US dollar against the Deutsmark and against the Yen), in the major world stock markets, in the U.S. and Japanese bond market and in the gold market. They find that, out of 49 significant outliers, 25 can be classified as endogenous (that is, predictable with the LPPL theory), 22 as exogenous and 2 as associated with the Japanese anti-bubble. Restricting to the world market indices, they find 31 outliers, of which 19 are endogenous, 10 are exogenous and 2 are associated with the Japanese anti-bubble.”

In my opinion, the source of such growth has to be capped in order to be predictable.  It is not hard to imagine a microcap stock growing unnoticed in the pink sheets only to go through ever increasing stages of capital infusion as it works its way to the OTC, Nasdaq Capital Markets, and the NYSE.  Using LPPL in the microcap stages to identify a top of the bubble would be a futile effort.  Also, these models might be more accurate predicting an end of a shorter term trend within an opposite longer term trend, but that is just a speculation.

He’s only batting 50/50 or so… or not.
Be careful in calculating the batting average as # right / # of total predictions.  Sornette addresses this issue in chapter 9 of his book.  Calculating probability of predictions in a time series is bit more complicate than that and I will not pretend to understand them (I pulled out of math into finance after taking probability).  Scan a few pages using Google – the results are much more meaningful than 50/50.

China.  A bounce in a secular bear market has a more limited possibility of bullish events.  There will not be the same exuberance flowing through the globe in larger and larger waves raising additional capital from dumber and/or more momentum driven investors.  However even within the confines of a secular bear, Chinese government can theoretically infuse even more capital/credit/leverage into the system and thus blow the predictive value of the Log Periodic Power Law (LPPL) parameters (related to meta-prices).  Although such infusion is more likely, if stocks start breaking to lower lows as opposed to pulling back from the recent highs (it seems like they’re determined to continue pumping money into this thing through 10/1 at least).

Other Predictions
Sornette issued quite a few predictions in the past.  His group miscalled a top in the 1997 stock indices, but nailed the bottom in Nikkei in 1999 and finally the top in Nasdaq in 2000 and the recent oil peak.  In1997, we were in the midst of a long term bull and going against such trend meant risking the occurrence of ever-increasing waves of new exuberance and additional capital.  The additional waves happened and it would have been suicidal many times over to be short.  Ultimately Nikkei, Nasdaq, and Oil reached the final stages of the feedback process without any exogenous interference.  So congratulations on getting these long-term trend reversals right.

Related posts:

1. Chasing the Dragon
2. Radical Transparency
3. Predicting the 2008 Presidential Election
4. Prediction Markets for Valuing Private Companies
5. The Good, The Bad & The Ugly