Zeev,
I have used much of IBD's approach to help identify (not predict) tops and bottoms, and their approach is typically pretty good. It has been awhile, but when I was a subscriber, they considered 4 distribution days (rising index volume with either a decline or failure to advance near the top) in a 2-3 week window, followed by a 5th day of major distribution (1%+ decline on rising volume) as a market breakdown.
Of course, because their approach is not predictive, it misses the first week or two of a rally or decline. However, it is also pretty quick to correct any false signals.
Here is the data I track as of yesterday's close, using a 3 week window (major distribution days are shown in square brackets). Let me know if you think it's useful. (Uh-oh. I can feel the heat of the flames already).
-David
---------------------------------------------------
NASDAQ: 6 distribution days
[4/13], [4/15], [4/20], 4/27, [4/28], [4/29]
S&P 500: 5 distribution days
[4/13], 4/14, [4/20], [4/28], 4/29
DJIA: 5 distribution days
[4/13], 4/14, [4/20], [4/28], 4/29
Current Index status:
NASDAQ: 4/21 rally attempt BROKE DOWN on 4/28
(Recent Breakdowns: 3/10, 3/11, 3/15, 3/22, 4/20, 4/28, 4/29)
S&P 500: 3/25 rally BROKE DOWN on 4/20
(Recent Breakdowns: 3/15, 3/19, 3/22, 4/20)
DJIA: 3/25 Rally-attempt BROKE DOWN on 4/20
(Recent Breakdowns: 3/10, 3/11, 3/15, 3/19, 3/22, 4/20)
AI
Market Data 
Markets 
