Could some test this idea and see how it went in the past?

The main rule is:

The horse must be a last start winner and won by more than a length.

What was the profit/loss?

Benny

Results for Sydney metro from 1/1/2003

Bets = 1803
Winners = 312
Return = 1405.6
S/R = 17.3%
POT = -22%

What sort of filters should I aply to try and improve this system?

Benny

Here is a system that I have found on the Internet, and which I have adapted to Australian racing conditions.

Rule 1: Operate only on handicap races.
Rule 2: A contender must have won by more than one length in its last start.
Rule 3: Eliminate any fillies and mares from consideration.
Rule 4: A contender must have had its last start within the previous 21 days (inclusive).
Rule 5: Must have started at 10/1 ($11) or under at its last start.

What were the results after appling the filters?

Was there a profit.

Benny

For the period 12/2/2005 to 5/3/2005, the results were as follows (Sydney, Melbourne, Brisbane and Adelaide meetings on Saturdays only):

Selections: 37
Wins: 11 (29.73%)
Places (including wins): 21 (56.76%)
Win return: $38.10 (SuperTAB); $38.96 (SP)
Place return: $38.70 (SuperTAB); $45.00 (SP)
Number of runners: 358 (10.34%)
Number of winners: 34 (32.35%)
Relative frequency factor: 3.1304 (32.35/10.34)
Expected number of winners: 8.6558
Actual-to-expected ratio: 1.2708

So, a little profit (2.97% on SuperTAB and 5.29% on SP) was obtained during the period covered.