Hi all,

My first post here.

Firstly, I'd like to congratulate Uselessbettor on creating such a superb web-based analyzer. UB, it's great that you're still working to improve it, and imho the ability to distinguish metrop from non-metrop as suggested above would be a major leap forward. (Distinguishing provincial from country would also be terrific, but I don't know if that's feasible for you?)

A few other things I've found to be important when it comes to database research:
-Type of race. Nuanced class differences aren't important, but Black Type races and Maidens need to be considered separately from everything else... 2yo/3yo/Set Weights/Handicaps are also useful distinctions.
-Day of the week. (Saturday/public holiday metrop really is a different proposition from everything else.)
-Ability to distinguish metrop/provincial/country for a horse's last start as well as today's race makes an enormous difference. For instance, there are filters that show long term profits based on horses running metrop-to-metrop that do not work with any other permutation.

Hope it's not presumptuous of me to make these suggestions with my first post. Obviously I've no idea how feasible or dfficult it might be to implement any or all of these parameters, but I thought I might as well pass on what some extensive past research has shown to be important.

You would think so wouldn't you, but every bit of research I've done suggests that this is a myth. I've even spoken to a professional punter in Britain who says that this is true of UK racing as well. Finishing position believe it or not is far more important when looking for profitable angles than beaten margin. And if you use ratings among filters, then beaten margins data will already be built into those figures anyway.

The only things I've ever found to be useful regarding beaten margins are at the extremes. Horses that win by big spaces >2 or 3 lengths do have high next start strike rates, but some pretty sly filters are required to overcome the tight prices that eventuate. I've seen a couple of angles on horses beaten a long way also, but only in very special situations and even then it's not that great.

When I was running the GTX data base , I found one can get too bogged down with the finer points to a level where one ends up with 5 bets a month.

Thus needing a bunch of systems so one does not die of boredom.

I found generally , beaten lengths had less advantage than Last start placing.

The lengths came into its own when working out how many lengths a horse won by.

Hoses winning by 5+ lengths did show some advantage

__________________
Cheers.

I reached the same conclusions also using GTX. The pro punter for whom I was doing some work also had a customised database alongside GTX on his computer and it showed the same thing.

Suggestions are great and I'll need to work on seeing if I can work out the different data sets. I would like to add a lot more to help out my fellow punters.

Given that you have such a great attitude to making further improvements: not as important as the suggestions in my original post, but certainly valuable if doable...

-If you changed Track Condition and R&S Pace Position to numerical data (fast=1... heavy=5) / (leader=1... backmarker=6), you could use "greater than"/"less than"/"between" operations as per the other variables. A legend could be included in parentheses for clarification.

-Weight-over-limit (or a limit weight variable) would be more useful than the absolute weight variable in isolation.

-Price rank as well as actual odds is useful. For instance, it would be great to be able to check closing price favourites only.

-SP last start is also useful.

Thought I'd mention that if it is possible to incorporate this feature, an excellent list can be found by googling "Australia racetrack grading". (Not sure I'm allowed to post a direct link?). Very handy for differentiating provincial and country meetings.

Also, if even a metrop/non-metrop variable is added, I would strongly suggest classifying Darwin, Hobart and Launceston separately from other metropolitan meetings, even though they are officially "metrop" venues.

The result from the website now is :



There were 33019 horse form lines in the database at this point in time. There were 1638 selections for the System
There were 1384 which Lost (did not win race) for the System for a strike rate of 84.49%
You had to pay out $1382.34 but brought in $1556.1 after 5% commission. This means a Profit or loss of $173.76 or a percentage profit/loss of 10.61%

Test Another System


The profit has increased over the last few weeks.

this one has failed so far since this was devised.



There were 43652 horse form lines in the database at this point in time. We update when we get time.There were 2256 selections for the System
There were 1891 which Lost (did not win race) for the System for a strike rate of 83.82%
You had to pay out $2023.67 but brought in $2143.2 after 5% commission. This means a Profit or loss of $119.53 or a percentage profit/loss of 5.3%

Test Another System
The Rules used were : WinPriceLay < 10 and barrier >= 9 and donScottRank >= 2 and careerRaces >= 5

A slight variation on this one is showing a profit:



There were 97885 horse form lines in the database at this point in time. We update when we get time.

There were 2621 selections for the System
There were 2190 races for the System
There were 2245 which Lost (did not win race) for the System for a strike rate of 85.65%
You had to pay out $2301.35 but brought in $2450.635 after 5% commission. This means a Profit or loss of $149.285 or a percentage profit/loss of 5.7%


Test Another System


The Rules used were : WinPriceLay < 10 and barrier >= 9 and donScottRank >= 5 and careerRaces >= 5