[ This Message was edited by: puntz on 2004-01-11 06:10 ]

Hello Puntz,

It's a little hard to say exactly what I'm looking for in a system. Obviously, it has to make money, but there's a lot of intangibles in it as well.

When I first got Price Predictor, I was happy to find a system with a 15% POT... it was very hard for me to find even those, as I had very little experience with racing beforehand. Now, I wouldn't even consider a system that showed a 15% POT over my data, as I always expect a system to perform about 15% to 20% worse off than it did over my test period.

So, given that I can find a system with at least 40% POT, then I also look at the following....

[*]number of bets[*]strike rate[*]longest run of outs[*]bank drawdowns[*]consistency of strike rate[*]distribution of dividends[/list]
considering all the above together determines whether a set of factors I've isolated become a bettable system or not.

Hope this helps.

Cheers,
Chris.

[ This Message was edited by: puntz on 2004-01-08 11:56 ]

Hi Stebbo,

Interesting topic - something I've been considering recently, i.e., how do I ensure a mechanical punting system will be profitable? You said:
/
Now, I wouldn't even consider a system that showed a 15% POT over my data, as I always expect a system to perform about 15% to 20% worse off than it did over my test period.
/

I'm wondering why do you do this? How long a period do you test your systems over? In my research, I have found the longer the period over which I test, the lower the overall POT. To me, this is because Price Predictor Pro automatically assumes you bet on every selection it hits, regardless of the selections' price (poor or otherwise). Also, the occasional longshot can have a huge impact (on POT) over a few races, but its influence tapers off as the number of races being analysed increases.

From your list of important secondary parameters, I just wanted to comment:

<>number of bets
- per race or in the entire test?

<>strike rate
- the most critical paramter of the system in my opinion

<>longest run of outs
- This will increase over the number of runs being considered (10,100,1000 etc) and may be mathematically estimated quite accurately. I had a program to do this, but in updating to MSexplorer 5.5, the code no longer works. Very annoying! It was from a fellow punter online, Sidsid, who seems to no longer offer his free punting tools :sad:

<>bank drawdowns
- Not sure why this would be a concern

<>consistency of strike rate
- good point - I am currently trying to construct an effective staking plan for "clustering" systems (i.e. wins come in groups between dry runs)

<>distribution of dividends
- unsure why this is a concern

I think any system that turns over even 10% POT is a gem - and the less bets required to scoop off the 10% free and clear is a critical guide to the system's profitability. Better 10% free and clear in 10 bets than in a 1000!

~Lenny



[ This Message was edited by: Lenny on 2004-01-11 05:02 ]

I'm wondering why do you do this? How long a period do you test your systems over?

To me, this is because Price Predictor Pro automatically assumes you bet on every selection it hits, regardless of the selections' price (poor or otherwise).

Hi Lenny,

I am a systems better, so therefore will almost always bet on every selection. I generally don't bet odds-on, and unless a system can show a reasonable strike rate for starting price of $20 or more, I won't bet on those either... but then that becomes "my system"..... I research all these elements carefully before starting to bet on my system.

I have just over 2 years of data, and generally won't consider a system unless it manages at least 100 bets over that 2 years if it's a Saturday system, and at least twice that for an "all day" system....

For reference tho, my main Saturday system based upon the Price Predictor ratings had 642 bets over 20months of test data, and the same system run on weekdays had 1,177 bets over the same 20 months. I've been betting both since October this year, and they are both performing as expected.


<>number of bets
- per race or in the entire test?

in the entire test. I don't mind backing two, sometimes three horses per race, depending upon the selections.


<>longest run of outs
- This will increase over the number of runs being considered (10,100,1000 etc) and may be mathematically estimated quite accurately.

IMHO, LRO is a function of strike rate, not the length of the betting run.


<>bank drawdowns
- Not sure why this would be a concern

I consider bank drawdown as THE MOST IMPORTANT aspect of any system which I research. The MaxDD is a measure of how far the system has retraced in the past, and I like to consider that every system I start betting will go to 1.5 times its MaxDD as soon as I start betting it.

It's a peaks-and-troughs thing. If a system has a profit of 100 units, but a MaxDD of 120 units, then it's likely that the 100 units is a peak, and it's next trough will wipe out all that profit. If it has a profit of 100 untis, and a MaxDD of 25, then it's next trough will still leave the system in profit.

I also set my bank percentage off the maximum drawdown figure. For example, one of my main Saturday systems has an historical MaxDD of 70 units. I multiply by 1.5 which is near enough to 100 units, so I only bet 1% of this particular bank. I know this bank can now withstand a new MaxDD.


<>distribution of dividends
- unsure why this is a concern
A system which returns 100 units of profit, but has 2 bolters paying $55 and $35 is a dangerous system.... I wouldn't bet it.

I think any system that turns over even 10% POT is a gem - and the less bets required to scoop off the 10% free and clear is a critical guide to the system's profitability. Better 10% free and clear in 10 bets than in a 1000!

I'll take the 10% profit from 1000 bets anyday, for two crucial reasons.

1. A system that works over 1000 bets is far more likely to succeed than a system which works over 10, and
2. 10% of 1000 bets is 100 units profit - 10% of 10 bets is 1 unit profit.....

While POT is a great way to compare systems, at the end of the day, PROFIT is what I'm after.

Cheers,
Chris.

Stebbo's polemic

Amidst the layers and layers of cliched dross polluting these Forum pages emerges a post of exquisite clarity.

I'm hoping the many won't take a darn bit of notice - it's probably already sailed right by.

Hey Chris,

You just got a POSITIVE from the DOC.

NOW THATS GOT TO BE THE FIRST TIME EVER ON THIS FORUM :smile:
Doc,
Without the cliched dross this would be an awfully dull place with about 5 posts a week.
Why don't you try it sometime it's actually a bit of fun, and also sometimes that cliched dross is informative for the average person here.

It's a shame we can't all be rocket scientists :roll:

[ This Message was edited by: kenchar on 2004-01-11 19:18 ]

[ This Message was edited by: puntz on 2004-01-12 10:05 ]

Hi Puntz,

can I ask why you've removed all of your postings? I don't think there was anything controversial in them..... ??????

Cheers,
Chris.

staying focused