I tend to agree with Mark - unless you know who is putting the money on it doesn't mean much. I did some analysis on this last year with quite a few years worth of data by looking at the Tabcorp prices 2 minutes before jump compared to the final approximate. Over approx 40,000 races I found the following:
Approx 7400 winners had shortened on TAB by 10% or more (eg. from 10.00 to 9.00)
Approx 4800 winners had blown out on TAB by 10% or more (eg. From 10.00 to 11.00)
The rest (about 28000 winners) were within 10% of the 2 minute price at the jump.

Based on this I decided that price movement (ie. late money) alone was pretty useless at predicting the winner as less than 20% of winners shown significant evidence of late money being bet.

I seem to remember that Dr Z. was selling a computer program that supposedly tracked this late money and suggested bets based on this info - there were quite a few negative comments about this on another forum last year (sorry - can't remember where).

__________________
"Computers can do that????" - Homer Simpson