General Betting
There is currently 1 person viewing this thread.
Will post this on the general betting also, but here we go. Normally thrive on this type of question, but been pulling all nighters all week and brain a little fried.
What I am looking for is system testing help, in regards to variance, and confidence levels.
Firstly, I will say, I completely understand that I have way too few results to do anything, but that is irrelevant at the moment. I am simply looking to place a confidence level indicator in my sheet
Here are the raw stats, for now:
128 races
1120 runners
44 identified bets
.248596 average market share per runner (average odds per runner ~4.02258823)
19 winners
10.93823 needed winners (before commission)
74% edge (edge probably wrong term here, maybe simply profit)
Easy enough to find how to test a system when you know, or at least guess your ROI, or edge, but if you are trying to find your edge, at a certain confidence level to find out just how early you are into your testing.
Not exactly sure of the format, but I would guess something like:
.95 confidence edge >-.5 and >2.5 where I could replace the edge figures to find the confidence level.
Anybody?
What I am looking for is system testing help, in regards to variance, and confidence levels.
Firstly, I will say, I completely understand that I have way too few results to do anything, but that is irrelevant at the moment. I am simply looking to place a confidence level indicator in my sheet
Here are the raw stats, for now:
128 races
1120 runners
44 identified bets
.248596 average market share per runner (average odds per runner ~4.02258823)
19 winners
10.93823 needed winners (before commission)
74% edge (edge probably wrong term here, maybe simply profit)
Easy enough to find how to test a system when you know, or at least guess your ROI, or edge, but if you are trying to find your edge, at a certain confidence level to find out just how early you are into your testing.
Not exactly sure of the format, but I would guess something like:
.95 confidence edge >-.5 and >2.5 where I could replace the edge figures to find the confidence level.
Anybody?
sort by:
Show
per page
Replies: 10
|
By:
Can you have a go at phrasing the question more clearly? It's a bit hard to see exactly what you want to know.
Also, do you mean edge >-.5 and [b] |
|
By:
Oops. That was supposed to say...
edge >-.5 and >2.5 |
|
By:
I give up.
Do you mean edge greater than -0.5 and less than 2.5 ? |
|
By:
I mean is there a way to show the range that the edge would be, based on that information. I meant by that, -.5 as in losing 50% of everything you put on, but it was just a random figure.
Not sure how to explain it any clearer. If you were to use any variance calculator, it would get you to put your edge in, the odds you take, and it would tell you the maximum downturn your bank will likely take, and the maximum upturn it will take, to a degree of a certain % of certainty. I simply wish to reverse the calc to get the edge, using the figures quoted as the maximum of the range. Most of them run a series of bets using the odds required, and give you either longest winning streak, or the 2 mentioned above, and whatever else. But all of them need to know your edge, which is what I am trying to find out. Simply? put, over a series of 44 bets (will of course change in time), what is the lowest edge you could have, positive or negative, that would expect to see a series of 44 bets produce a profit of 174% long term? Would a 0+ve bettor expect to see that profit over that series of bets? Would you need 5% edge, 10%, 20%, 30%? There must be calculations for this. Or put it another way, or the same again? The 74% is the edge of that series of bets. What must the edge be, to hit that level over that series of bets, where that is the highest series you will see, with 95%+ certainty? And what is the maths behind it? Hmmmm |
|
By:
It does for the record have an Archie score of 7.5, suggesting .006% probability that it was luck. But it seems ridiculous over that small a series that would be anywhere near accurate. Only learnt about that today, but don't put much faith in it.
|
|
By:
Think archie is best used in conjunction with a sample size factor. The recommended one is adding together the probabilities until they total at least 10. ie 20 even money shots (with 16 winning would be the equivalent of your 7.5).
|
|
By:
which looking at your OP 44 x .248 = 10.9?
btw are these live results? ie not back tested? Also all archie says is that the "profit" is not due to luck, it doesn't specify what the actual odds are. |
|
By:
|
|
By:
you could mess about with archie by changing the expected win figure.
So you have 49 runs, 21 wins, .246 average expected win probability. If the expected win is changed to .3 (expected wins 14.7) you have an archi of 3.85, so a 95% chance that the actual probability is greater than .3 given 49 runs and 21 wins. My point about sampling is simply that its best to theorise then sample. Rather than sample and theorise. |
|
By:
Thanks for the advice, but been here a decade, and if I wasn't happy to throw money down on a theory, I wouldn't be here.
Your point about Archie, is that effectively could lose the next 12 bets, and it would still tell me I was on a big winner... Is that an effective test, and is there any others? I do note, that the other bot systems I have run, while making money on the whole, although all pulled the pin after a bad run, would have realistically polled very low on Archie, even though I believed in their theory much more than now, which suggests I should have been doing this from the start. Archie tells me I am 1.001 to have stumbled on the holy grail. My head tells me the opposite. I need more tests, and more data. Either way, money down and plan to enjoy the ride either way. Either way, there will be many more ideas, and an effective way to test the possible edge, is required regardless.... |