19th September 2005, 11:36 AM
|
|
Member
|
|
Join Date: Jan 1970
Location: Bundy
Posts: 292
|
|
Quote:
|
Originally Posted by jfc
Use the number of winners rather than number of runs to guide your confidence in the results.
An average system with a 10% strike rate and 2,000 runs would yield 200 wins.
Your sample has 213 wins and therefore is probably just as accurate as that 2,000 run test.
|
I agree,
Someone suggested a test of 3 times 150 consecutive races,that is all well and good for a system that has a high strike rate but it is unreliable if the system only provided 15 winners or so in a 150 race stretch.
A better idea would be 3 consecutive tests of 35 to 40 winners.
Wesmip has enough data to start having a crack,aslong as he hasnt unwittingly backfitted the results all should well. |