Open Ended MLB Starter Fade System

Updated on September 5, 2015 in Sports Betting Research
14 on September 4, 2015

I’m not sure who is with me on this. I prefer very simple situations with large datasets. Rather that having many trends I just keep a few of these and make a play when 2 or more line up.

I thought I’d share one of these in the hopes that you guys might share different perspectives and show me how you would optimize this one.

The system is very simple: Fade a starter that gave up 1 or 2 runs in his last 2 starts. Note: I exclude shutouts. This is JUST 1 or 2 runs.

That is 3671-3655 +118.96 units
3>os:starter earned runs>0 and 3>oss:starter earned runs>0

works better for the home team…

  • Liked by
  • pcg
Reply
0 on September 4, 2015

Awesome query. I think it is good to have different philosophies.
I am using all kind of league wide trends without specific Team or Player action. Some of them are strongly backfitted. Whatever works.

  • Liked by
Reply
Cancel
1 on September 4, 2015

Here is my opinion on that, and again, just my opinion, and philosophy. I would never wager on that system, for the simple reason it shows a 1.4% ROI. You could almost get that return on a simple savings account interest, and certainly a CD, and I wouldn’t invest in either of those either. Then there is this factor. Your assuming the closing line that the database uses for each play was exactly what you were able to bet the game on. That is certainly untrue. What if the lines you wagered using this system were 1.4 cents worse than what you see in the database that shows? Your left with 0% ROI, and in a different scenario, what if the line wagered vs line used to log the system in the database was off by 5 cents? (very reasonable to assume), you are -3.6%!  When the “profit” threshold is 1.4% that is a huge RED FLAG for me, as represented from above. I never consider anything, or any situation to be deemed reasonable to use unless the ROI is 10% or greater on a high volume of games. The z-score on that system deems it non-predictive. Your concept is very good, and I agree with it, lots of games, simple limited variables, but in this case, I’d stay far away!

Author
on September 4, 2015

Also, do you realize as a 100.00 better you would have had to wager over 3/4 of a million dollars (risked) to return 11,896, and that of course is assuming that you got the same lines used by the database at the time you wagered?

Show more replies
  • Liked by
Reply
Cancel
9 on September 4, 2015

Here is a visual f just what I am talking about:

The last play in this system occurred on 9/2 and the bet was on Minnesota archived in the database at -120. The game opened at -120 (so I assume, at least in this case, it was logged within the database as the opening line).

The game had a closing line from -126 to -130, so certainly the best line was available at opening, and worsened by the time it was bet by the system user, perhaps as bad as -130. That is an 8.3% line change, being wagered into a system that has a 1.4% ROI. Do you see how quickly that can bankrupt you? Now you know why I have a 10% rule!

Admin
on September 4, 2015

The 10% rule is good, but what is the maximum number of parameters you’ll go to to achieve it?

Author
on September 4, 2015

I never view parameters as a reason or not a reason, I mostly use parameters that have stand alone value, so I am stringing together known quantities, not just something that makes something look better.

Admin
on September 4, 2015

Right; you start with a “pinwheel” and you always want that to be 10% or do you want your subset of the pinwheel to be at least 10%?

Author
on September 4, 2015

I start with a winning singular variable, based on the pulse of the sport, each is different. I have about 25 or so variables per sport that win with just 1 variable, or are at least a vig eater. I combine these variables, based the premise I am looking for, knowing the pulse of the sport, and where the value may lie, not string together a bunch of variables and hope to find something. The premise must always come first for me, not the system, my experience from years of work on this, as those tend to die.

Author
on September 4, 2015

here is an example of a positive singular variable in NCAAF:

#dddddd”>

p:points > 49
SU: 1732-846-10 (9.26, 67.2%)   #f8f8f8″>Teaser Records
ATS: 1301-1161-45 (0.81, 52.8%)   avg line: -8.1 #f8f8f8″>+6:  1672-805-30 (67.5%)   #f8f8f8″>-6:  879-1582-46 (35.7%)   #f8f8f8″>+10:  1885-600-22 (75.9%)   #f8f8f8″>-10:  665-1802-40 (27.0%)  

That wins by itself, as NCAAF is a momentum driven sport.

Admin
on September 4, 2015

That is 0.8% roi. So that answers my question. You look for 10%+ by the end of the process; not from the start. I think that was what the original poster was getting at as well. Start with a robust simple concept and do whatever magic you do from there. 

I guess my only argument would be that sportsprediction’s base system isn’t exactly simple. You need 4 things to happen vs. 1 thing for yours.

Subscriber
on September 5, 2015

p:points>49

That is AWESOME weatherwizard. I feel like that simplicity is exactly the kind of thing I’m talking about:

I want to find a big, broad system and as PCG said, THEN do my “Magic” on it after that which is more or less either adding extra parameters and capping it by hand to narrow it all down. I don’t bet everything blindly from the broad system.

I see what you and PCG are saying: I need 4 things to happen with mine and yours only needs one thing to happen. Let me say this though:

THE IDEA WITH MY SYSTEM IS:
I want to fade a hot starter making a contrarian bet. I have a lot of other systems that are similar. Fading a hot starter this way and that way. There’s money in them hills and so since I am aware of this IDEA based on other research I’ve done, I am okay that this one has 4 factors.

And in case you’re wondering; I haven’t done all that bad over the years with this method. I’ve been doing it before I knew about sportsdatabase too.

Subscriber
on September 5, 2015

What do you guys think about this?:

osA(starter earned runs,N=3) < 2 and n:series game = 1 and H and -150 > line > -250 and p:margin < 2

-Fading a Hot starter (ERA L3 games < 2.0)
-Play on team either lost the last game of one by a single run.
-Play on Large Home favorite 
-Last game of series.

125-49 +40.06 units +13.0% roi

Author
on September 5, 2015

I like that!

here is something to consider, and how I build a system:

SYSTEM:

month=9 and STDSERA<2.35 and AF and line>=-140 and p:W

This returns over 10%, but that isn’t the important part, as you will see.

The components of the system are the important parts, and here is why. Each component connector is playing to the contrarian “advantaged side.”

The following are the ROI’s comparison of the components, not profitable, but 100% leaning, and leading to the advantaged side

month=9 and AF                             -3.0%     -0.6%   (advantage dog)
AF and line>=140                            -3.5%     -0.6%     (advantage dog)
AF and STDSERA<2.35                 -2.6%      -1.2%   (advantage dog)
AF and p:W                                     -3.3%      -0.3%   (advantage dog)
AF and p:W and month=9               -5.3%      +2.5%  advantage dog)

there are all the components of the system, mixed and matched in every way. What you can see is, all components, are the advantaged side, not necessarily the profitable side. So when your dealing with a set of variables, it helps to eliminate just using any variable to make the system look better than it really is, and thus allows us a higher degree of confidence that we are working from skill, not accident, and working from meaningful additions, not just additions that makes something look better. This is the biggest flaw I often find in betting systems that I see posted out there, they look good, you can make them sound good, but they have red flags in many cases that suggest they are contrived by accidental discovery, by inserting meaningless variables.

I hope this helps some, and I hope I’m not going over the top of the skill set of most reading this. I can apply the reasoning for each variable in depth, and I’m sure you could figure it out

Baseball has you believe it is all about the pitcher. What is better than taking an elite pitcher on the road and having to lay less than -140? As seen above it is the disadvantaged side (AF and line>=-140)  -3.5% on  -0.4%  against  so we are working from a not profitable 3.1% advantage, which in and of itself is not profitable, but when you learn what the advantaged sides are, many of which are profitable (AF and p:W and month=9               -5.3%      +2.5%  advantage dog), you begin to fundamentally discover how to create a better system, that isn’t based on intentional distortion of a system by selecting a meaningless variable that makes it look good!

There is so much more to write, but for now that should open up some minds I hope.

Show more replies
  • Liked by
Reply
Cancel
0 on September 4, 2015

I have to agree with the mathematical and financial view on this matter. However the mindset is ok. If you take this query as a starting point and have other supporting ones the combination may lead to higher yield and profit either way. At least that’s what I am trying to do. Summative queries with different angles mean as much value as one real good query to me.

  • Liked by
Reply
Cancel