How much of an 'edge' is needed?
 

Just been playing round on an online system builder I found online. The prices and results it shows are based off highest of either official on course, NSW TAB and VIC TAB starting prices.

What I'm curious about is how much do you need to improve on say for example those horse paying $3 or less, to turn a profit either way?

I only have very limited BFSP figures to compare this against , but seems you would be break even/small loss after commission backing all these at BFSP. Someone with a BFSP may be able to correct me on this?

The results I have from the system builder are for all horses $3 (benchmark data) and under:

Selections - 17332
Winners - 3700
Strike rate - 40.66%
Ave Divi - $2.33
Loss on Turnover 5.251333%

Now lets add a simple filter over the same data period and I get:

Strike Rate - 38.9%
Ave Divi - $2.37
Loss on turnover 7.93% (2.68% greater LOT v benchmark data)

Is there enough there to crimp out a small POT (ie 1% POT) laying these?
How much is realistically required either side of the 'benchmark' figures to eek out a small profit on turnover?

I'm not interested in debating the filter added and the validity of the filter , more simply the idea of how far you need to move away from the benchmark to have a chance at making a profit either backing or laying.

Anyone care to share there experiences thoughts?

If its puntingform your using it probably wont work. I tried laying a few systems through my bot over the past couple of years but the commission and difference in price beats me everytime. Without commission would be more likely successful.

If your going to be system testing then you can't rely on figures after the fact, or best tote, or anything like that as a filter as they will not be the same when you look at them pre race.

The second thing is you can't compare returns from BFSP with best tote. They are completely different beasts.

Now to why your edge won't work.. The layers/backers on betfair do not have as much mug money thrown in with them so the prices are slightly more accurate. You will find regardless of what you find your slight edge will be adjusted for on betfair by slightly longer odds and you will lose around the 5% anyway (commission).

I had a look at your query and using the much maligned TAB prices ....

Taking 395,909 horses in races of 8 to 14 starters (and NO other criteria) we find that :

Prices of $1.50 to $5.00 produce an AWSR of 26.17%, and Av.Div. Of $3.31 and a POT of (13.39%).

Favourites in these same races produce a WSR of 30.98%, Av.Div. $2.83, and a POT of (12.35%).

Prices of >$5.00 generate a (price graduated) AWSR of 5.8%, Av.Div. of $12.95, and a POT of (24.83%).

So, looking at selections that are rattling around $5 and less, the “edge” you require from your system must exceed +/- 13% just to break EVEN (over time). In other words .... your method must identify horses which have been mispriced by the market to this extent (by under estimation).

Be extra careful of methods/systems/results which rely on samples of 30 or less or those that indicate they can predict short term trends in (for example) runs of Favs. Anybody that can do this is already making a fortune on ASX.