Monday, May 13, 2013

Are tops really a process? Analyzing 25 years of $SPX topping patterns

Anytime we approach new significant highs in the stock market, the Twitterverse lights up with "This is THE top" calls.  Well, I'm somewhat guilty of this, too.  I'm sure it has much more to do with ego -- trying to prove how right you are.  However, I've learned the hard way that being right is far from being positively correlated with profitability.

So let's look at the data.  There's a saying that "tops are a process and bottoms are an event", so I've done a little hindsight-based-K.I.S.S.-back-of-napkin type analysis to take a closer look at stock market top formations.

There's a zillion different potential approaches that could have been taken to performance this analysis, and I just did something I felt comfortable with.  This is not a statistically significant analysis nor an actionable trading system, but for me, this homework helps me to get a "feel" for the data from my perspective and also generates some interesting takeaways.

Since "tops are a process" is often repeated as an trading axiom, does there really need to be a big rush to exit a long position (or to enter a short)?  And how much time do I realistically have to act on a topping pattern?

  1. Started with a 25 year monthly chart of the $SPX to identify key highs during an uptrend
  2. On the monthly chart, down arrows were placed at pivot swing highs based on a high that had 1) two preceding highs that were lower and 2) two succeeding highs that were lower
  3. The highs from the monthly chart were then identified and marked on the weekly chart with a down arrow
  4. On the weekly chart, starting from the pivot swing high arrow, the next pivot swing low was identified, and the retracement was measured
  5. If the retracement to retest the pivot high was greater than 50%, then that point was identified on the chart with a smaller down arrow
  6. The number of weeks between the "Potential Top" and the 50%+ retracement was identified
  7. And finally, the shape of the 3 legs of the swings, starting from the high, was also drawn on the chart
Here's the high level SPX monthly chart going back 25 years (January 1998 to May 2013), with each significant high marked with an arrow:
The following charts drill down to the WEEKLY charts, with each arrow corresponding to the arrow on the monthly chart above.

July 1998 - March 2001

 January 2004 - September 2006

October 2006 - August 2009

February 2011 - December 2012

July 2010 - Current (May 2013)
[Note: significant overlap with chart above]
Doesn't look like a topping process is forming yet.

  • After a "Potential Top" was established, ALL examples had a bounce from the succeeding pivot swing lows of at least 50% that retested the highs (based on weekly chart)
  • After "THE Top" (2000 and 2007) was been established, there was adequate time to exit on a retest bounce to highs (a 50+% retracement) within a 2 week period.
  • After a "Potential Top" was established (13 incidents), there was adequate time to exit on a retest bounce to the highs (a 50%+ retracement) within an average of 4.5 weeks
  • During the last 2 run ups to new all time highs (2000 and 2007), there were 6 significant swing highs (false tops) before the "THE Top" was established
  • For both the 2000 and 2007 tops, once the pivot swing low that preceded the high was broken, it potentially indicated "THE Top"
  • The 2000 pivot swing low was broken after about 8 months, and 7 months after "THE Top"
  • The 2007 pivot swing low was broken after about 5 months, and 3 months after "THE Top"

Long term investor:
Over the very long term perspective (i.e. years/decades) which is the timeframe for the long term investor, the stock market has an overall bias to go higher. 

So unless you have a highly accurate method of picking tops, just wait for some confirmation that the trend has changed before exiting.  Otherwise, you're likely to get many false signals until you get it right.

Something as simple as waiting for the pivot swing low that preceded the high to be broken (blue lines on the monthly chart), in conjunction with other signals seems to makes the most sense.  You get several months to make your decision.

So yes, don't need to rush to the exits.  Based on a longer term horizon based on monthly charts using the break of the pivot swing low criteria described above, the topping process lasts 3 to 7 months based on 2 incidents, over the past 25 years This is loosely identified as "Topping Formation" on the weekly charts.

Longer term trader:
If you're looking for a trade lasting up to a few weeks and don't really care about picking "THE Top", then every potential top had a bounce that exceeded 50%. 

That could be a good area to potentially exit long positions or even go short, with a reentry or max stop on short above the "Potential Top."  But all the losses from getting stopped out could likely negate the profits you gain from eventually catching "THE Top."

You can also keep an eye on the monthly/weekly charts to see whether a potential top is lining up so that it can give some bias to the daily chart timeframe.

In the case of the shorter term timeframe, the topping process, where a bounce up retests the 50+% mark to the highs on a weekly chart, lasts on average almost 4 weeks.

Day trader/Swing trader:
Don't even bother reading this, except for entertainment and cocktail party purposes.  (And yes, I essentially fit into this category).


Michael said...

Great research, Thanks!

Grove Under said...

Hi Michael,

Thanks for the feedback, much appreciated!