I have an SDQL and need advice on it
danshan
Senior Member
I created an sdql that I assume is solid.
I ran it on sportsdatabase since 2003 and it has a record of 661-282 with a ROI of 5.9%
I am just curious is this the greatest thing ever or is it typical for sdqls with 1000 games rated to have a 5.9% return?
I was thinking this sdql is like a retirement plan for me but I just wanted others opinions on it.
It is not for sale and I am not offering handicapping services or trying to do anything.
again the question is, is this the greatest thing ever or is it typical for sdqls with a 1000 games rated to have 5.9% return?
I ran it on sportsdatabase since 2003 and it has a record of 661-282 with a ROI of 5.9%
I am just curious is this the greatest thing ever or is it typical for sdqls with 1000 games rated to have a 5.9% return?
I was thinking this sdql is like a retirement plan for me but I just wanted others opinions on it.
It is not for sale and I am not offering handicapping services or trying to do anything.
again the question is, is this the greatest thing ever or is it typical for sdqls with a 1000 games rated to have 5.9% return?
Comments
The thread topic says need advice? Now a question? Ok.
5.9% ROI? Off the charts!.
This however is a common claim on boards without documentation. Professor Ed Horowitz would this consider return "chump change".
My brother he's a scientist, and he found a way to time travel. Seriously!! We already did it a couple of times. It's great! It can only go back a maximum of 10 years though. He's still developing it.
Now my question is, should I go back 10 years and buy Apple and Amazon stock? or should I go back 5 years and buy some Bitcoin? Only serious answers from experienced investors please.
No, because something like this could happen:
https://m.youtube.com/watch?v=43Lk_qxUxVw
I am not sure what is sarcasm and what is a legitimate response. I am now in search of a way to maximize profits with a money management system if that is even possible.
here are the by season ROI's
2004 -0.9
2005 13.5
2006 11.8
2007 3
2008 14.1
2009 13.8
2010 -5.4
2011 -2.3
2012 17.5
2013 2.1
2014 11.8
2015 3.4
2016 2.5
2017 2.6
Rule 7
Unless you are a paid site sponsor, do not refer to any past picks or post any records of picks that were not originally posted on the BettingTalk forum before the outcome was determined. This includes any picks given out in the chat room.
No not striking for a moderator job either.
yeah if your stuff is true that is very cool and not often found or say no dude that is more common than a cold, that is all I am asking for from someone that knows
You sure? It allows you to selectively choose to let some rule violations stand.
so if the past over 13 years the sdql has indicated a roi, that would not be indicative of future results or a similar roi?
Peepsplace.com has the most intelligent level of conversation of any internet forum
Whoever recommended betting full kelly is underselling what you've got (5.9% roi!!1) I would look into 2x or 3x kelly. gl
It is extremely unlikely you will win going forward data mining angles like this. Even if you find something that's statistically significant (p<.05), traditional statistics really doesn't work.
661-282 and the roi is only 5%? What are the average odds?
-198 is average
when you say statistics dont work can you elaborate on that?
Traditional statistics is all about rejecting the null hypothesis. In this case whether the difference between expected return (~66.4% 198/298) and your actual 70.1% is simply due to random chance or if it's actually significant. This is a simple binomial test with 661 successes in 941 trials and an expected return of .664 (probably is less than this due to vig, but it's fine for testing)
you can do this in excel something like this i think =2*(1-pbinom(661,943, 0.66))
R has an easier function
Ok, so with 95% confidence we can say the true return of this system is between 67% and 73%. Average odds of -198, shit fire away we getting rich.
Not so fast. In reality this just doesn't work out. How many systems did you try first before coming up with this one? If we looking for a 5% p-value you'd only have to test about 20 before stumbling into one randomly that looks significant but isn't. Does the system make sense? Ie is there a reason you think the market is systemically under valuing something? LIke say NFL home underdogs the week after getting blown out. That was pretty solid for a long time -- until it wasn't.
That's the main issue with systems that take an angle and suggest betting without projecting our own fair value. You have no idea when the market will start incorporating that angle and probably will only find out once you've gone broke.
I am not picking the night winner of a game a day after it rains and with a pitcher on his 2nd start in June.
My criteria is based on what I consider sound stats, moneyball type factors if you will not some general random coincidences.
I hate sdqls that people post up with 20% ROI based on what I call crazy criteria like
starter with 4 days rest after a 3 game road trip playing at home after being a favorite in the last game and winning by less than 1 run
that to me is garbage and that is not what I have.
My sdql is based on "moneyball" type stats and has averaged 5.9% ROI since 2003 and has almost 1000 games rated
I am basically taking what i consider significant factors in handicapping ball games and seeing how that would do over 13 years of history. I am not gathering some random data that somehow like a rubics cube makes a magic 5.9% ROI
losing the last 3 alone does not give you the odds bettors are on that
racing at home alone does not give you the odds bettors are on that
but together bettors dont like that combo as much and leave a 5.9% gap
Yes, if it's real eventually it will be incorporated into the market. Dr Bob has hundreds of these angles and can't win anymore.
If you have a good theory a better approach is to use the advanced stats you are using to develop a model that produces a fair price for a game. Back test that properly to see what kind of edge you have. You'll have a far better chance of winning going forward.
I'd have severe doubts a major market would be beatable like this, are you testing vs openers? Will you be betting openers