Calling maths experts...
 

Hey guys another set of questions, sorry but you knowledgeable know a s***load more than myself, so I have to ask.

Number of tests to have confidence???

After more research on this forum I found some interesting equations. My results are as follows:

seln 158
ret 217.9
w 52
s/r 32.9%
ave div $4.19
lowest $2
highest $7.7
longest run of outs 9
longest run of ins 4

exp = ((DECSP-1) *W%) -L%
exp = ((4.19-1) * 0.329) - 0.671
exp = (3.19 * 0.329) - 0.671
exp = 1.0495 - 0.671
exp = 0.3785 (37.8% profit for each dollar bet)

MINAP (minimum acceptable profit) I would be very happy with 20%)

ERR = W% - (1+MINAP / DECSP)
ERR = 0.329 - (1+0.20 / 4.19)
ERR = 0.329 - 0.2864
ERR = 0.0426

Confidence level (Z) input 90% confidence = 1.65

TESTS = W% * L% * (Z / ERR) * (Z / ERR)
TESTS = 0.329 * 0.671 * (1.65 / 0.0426) * (1.65 / 0.0426)
TESTS = 0.329 * 0.671 * 38.7324 *38.7324
TESTS = 331.1824

So did I work this out correctly? After 331 bets I can be 90% confident of making 20c on every dollar bet?

Winner Winner Chicken Dinner!

No comments must be onto something?

Winner winner chicken dinner!

Whatever it is, stop using it!

90% is not statistically significant, not that is necessarily wrong mind!

Good luck

If my 90% is insignificant. Then whAt is? Suggestions please?

Winner winner chicken dinner!

Not at all, read what I wrote!
(btw the first part of my reply was a poor jest, no wonder you didn't get it)

No harm mate, call me stupid (gen Y) I'm missing your point? I wanna have a bet do you recommend or not?

Winner winner chicken dinner!

Sure, go ahead!

Something being "statistically not significant" means nothing like "insignificant".

I may put it better by saying does not mean "insignificant"

The reason you probably didn't get too many replies because we do not understand the question.
Or at least I do not, the others may do but decline to comment.

Good luck

Well the question is I'm at test number 158. So after test number 331 can I be 90 % confident that this system will make 20 cents profit per dollar? The main question is did I do the calculations correctly ? Has anyone any experience with the equations... Re: calling maths experts.

If my calculations are correct I'm quite happy with the results

Winner winner chicken dinner!

Hi ,

Do further testing. If you are however confident that you have something that you are happy with. See if it actualy works live. Bet a token ammount even if its $1. More if you actually want to see some results. What this does it cements the system for you. It answers questions, like can I actually bet ? When ? How? How many bets do I miss through mistakes, or when Im missing in action and so on.

Tests are one thing, live betting is another and unfortuntely the two are hardly ever the same. I can assure you that most people cannot achive the sterile results of a test in real live betting EVEN if the actuall results duplicate precisely. Life gets in the way, the bets that you should have you dont because of all sorts of reasons external and also your own mind will play tricks.

With token amounts you will still suprisingly suffer the mental issues (its called having balls) when the next bet is about to be placed. But the token amount betting will show you a real performance of your system.

Do this for a few months untill you are confident and then slowly escalete the bets by increasing your bank. Dont think that you can test with a $1 and then go and earn a fortune betting even a $100 or $200. You have to accustom your self to the bet size mentaly.

One of the reasons dumb staking like the fibernachi dosent work. Either you will be bored stiff betting the lower sums, or you will crumble when asked for a bet that is may be 500% above your comfort level as the last bet of the sequance and watch it go down in a protest after you survived a photo finish.

Trust me it will happen, and others thisng that you can't even dream up.

Then you analyse what you got live vs what you got testing and see !

Most people are brave looking at a test with 30 outs knowing that there is a $30 winner, as they can see the whole test at once , this isnt so in real life.

Further I notice that you have NOT got anywhere near enough data to test anything. The 150 bets how long is that over. (dont answer that, just think about it) 1 bet every day ? I bet every hour ? Or will you steel yourself waiting a week between bets ???? That is a hole new ball game. At 32% SR you will still get somewhere around 10 to 15 run of outs. If you bet one a week you might go 3 months without a winner ! I have been there and its painfull, even if at the end the system shines and delivers what your tests suggest.

Michal

Thanks Michal, exactly the type of response I am looking for.

Now to expand, i am very new to this systematic approach however I've read most of the posts on this forum that the title appealed to me.

In my limited testing I've found that it may be easier to run multiple systems with limited bets.? For example five systems with 100 bets per year simultaneously as apose to one system with 500 bets. In my experience it is easier to get a profitable system by limiting bets... Does anyone agree?

The problem then arises, to get the 1000 tests will take ten years... So I look toward other possibilities to reduce test bets.

So basically at the moment over found the less bets per year = higher POT. Until I can work out something that has 1000+ bets per year and shows a profit. This is all I have and I'm looking for advice because I wanna start making money but at the same time I'm looking for advice because I don't want to lose money...

Winner winner chicken dinner!

In a nutshell Michal !
 

I'm most impressed with these observations Michal. They are I believe the very essence of the difficulty of turning a system into an income. Remember bryzz that even if that system did great all year there's nothing to say it wouldn't all go belly up for a year after that.

Real time betting is critical to system testing, and I also believe a system hasn't proven itself until you're COMFORTABLE putting income earning sized wagers on.

One of the reasons I ended up doing short prices only was because the consistency factor is so much easier to handle.

Ok then there's only one more thing to do befor I go live. What is the cheapest system testing data base that goes back five years? I feel quite confident that it may hold up maybe not at that POT but with the price range, the profit isn't made up of just a few big dividends. And take any section of the sample time and the s/r is very similar. I believe the use of best tote will help too which hasn't been factored in.

Winner winner chicken dinner!

There is a free online one around here some where but not sure how accurate it is, there is also one for sale on this website.
http://forums.ozmium.com.au/showthread.php?t=17950

http://statpages.org/confint.html

This indispensable tool shows your observed strike rate of 52/158 reduces to sustainable one of 25.65%.

Suggesting a sustainable POT of 11% rather than the observed 37.8%.