This entry was posted on Monday, November 19th, 2007 at 8:53 am and is filed under Education, market commentary, system trading. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.
In my post 2 weeks ago about the large drop in the banking sector I had made a few comments as to what I thought this drop might mean going forward. I had used a hypothetical holding period of 10 days to see what sort of forward results we could see. I also think it shows quite succinctly the power of backtesting.
Here is what I said 2 weeks ago:
Over the next 2 weeks I expect to see what our small snippet of history has shown us.
- Massive volatility in the market and banking in particular
- A tendency towards a upward reaction after a panic low is put in
- I would expect the range in the Bank Index to have a range of >7% low to high over the next 2 weeks
All of these “predictions” were not based on opinion or my personal bias, but were based on what the market had done on average in the past when this condition triggered.
Here is the chart showing that post 2 week period. The blue dot shows the trigger day and the bands denote the high to low range after that period:
Let us review each of the statements.
“Massive volatility in the market and banking in particular” - although most would have thought this statement to be an obvious one it actually was not. If you look at high volatility periods in general these spikes tend to see a lessening in ATR (Average True Range) in periods that follow. Yet we saw an increase in the 10 day ATR for the SP-500 jump 24% and in the BIX.X a 41% increase.
“A tendency towards a upward reaction after a panic low is put in” - on November 7th we had a 6.8% drop in the BIX.X, the largest one day drop in the sector since the selloff began in February and the 8% upward reaction after that massive down day within 4 days.
“I would expect the range in the Bank Index to have a range of >7% low to high over the next 2 weeks” again this was pointing to volatility but also zeroed in on just how volatile. The average for the periods I tested was for post trigger range of over 10% and so I tried to be a bit conservative yet still give an idea of the magnitude of the potential 10 day range. We ended up with an 11.73% range - right on in terms of the historical norm.
This little example gives us a snapshot of the power of backtesting data. It not only gives me trade ideas and how I can frame them but also can give a broader view of what the context of the broader market might be. I would highly recommend you find some way to test and analyze past market data in order to make educated guesses at forward looking market analysis.
Have a Great Week! A shortened one at that!
Dave Johnson
November 19th, 2007 at 2:02 pm
Months ago you published a study showing results of buying ES when McClellan closed below -200. McClellan closed -211 on November 7 and ES closed 1482.75. Interesting that only 1 day since then (Nov 13) has closed higher and McClellan still extremely oversold -254 as of this writing. If I remember correctly the average holding period for McClellan return to > zero was 10 days. Shows me how unrelenting the selling pressure has been this month.
November 19th, 2007 at 3:38 pm
No the T2106 indicator published by Worden is different the the published McClellan number. It still has not crossed -200 although both bounces were when it crossed -160.
http://thetradingdigest.com
/blog/2007/06/07/what-does-a-200-mcclellan-cross-mean-buy/
Worden measures using all NYSE stocks so the absolute value differs.
http://www.worden.com/TeleChartHelp/
Indicators/Worden_s_Market_Indicators_T2s.htm
Scroll down to T2106