For the past week the following system has shown about a 10% POT on Place betting using Unitab divvies. Unfortunately I haven't bet them.

1) Default neural settings.
2) Outright top in the Dist category.
3) Max distance of 1,400 metres.

Unfortunately I haven't recorded the size of the fields.

Five selections today. We'll see how they go.

Geelong
3/9

Ipswich
1/7
6/2

Wagga
5/1
6/3

If interested I could check it for you before investing too much?
Cheers

lomaca, if you would please.

There's been two selections today that have already run - a first and a fourth.

Here you are Michael

Races tested 6646
total return $5986.72
POT -11%
S/R 40.26%

All races all tracks all race type included.
Does look like it has potential.

The last time I ran this programme was about two years ago and now it seems to be uncommon sluggish so I have to look at it before I can test this idea in more detail.
I included the test file printout that you can put into excel and see if you can find any clues.

The attached zip is a textfile that will easily import into excel as a "fixed width" text file.

Good luck

Attached Files

TopPL.zip (76.1 KB, 875 views)

Ran it on about 1400 Metro races and the loss is around 20% so leaving those out almost makes it profit neutral, which is not bad at all.
Although some tracks seem to do better than others.
All it needs is an other filter maybe size of field?

Have to fix this routine before I do any more testing, it used to take about 10 minutes to do 50 thousand races and now it takes over an hour to do just 5 thousand, strange?

Thanks, lomaca.

It's the usual story with the neurals - they don't perform too well with metro tracks.

Unfortunately I know zero about computers so I can't offer any suggestions about the problem.