MLB Run Line System

Quote Originally Posted by nfl_huskers:

A few posts by w1s3k1d and rivermonsterz got me interested in researching possible positive ROI run line systems.  Found one using SDQL, so I'm going to see if the great results continue into 2017.

Bet the -1.5 Run Line for ...
     Away Favorite
     Total >= 8
     .500 to .550 season winning percentage for Favorite.
     .410 to .510 season winning percentage for Underdog.

Query: AF and total>=8 and 50<=WP<=55 and 41<=o:WP<=51

Based on SDQL results, the system is up 109 units over the past 10 seasons, and has an overall ROI of 12%.

Games     RL Record         RL Avg Line    $ RL On    Season
150          70-79 (47.0%)        109.1           -$440        2007
103          49-54 (47.6%)        122.3          +$620        2008
101          48-53 (47.5%)        128.3          +$805        2009
52            25-25 (50.0%)        120.5          +$503        2010
67             5-32 (52.2%)        134.2        +$1,436        2011
84           42-42 (50.0%)        124.4           +$898        2012
48           27-21 (56.2%)        128.8        +$1,326        2013
98           52-46 (53.1%)        130.3        +$2,158        2014
61           35-26 (57.4%)        129.0        +$1,949        2015
115         59-56 (51.3%)        123.5        +$1,667         2016

I broke it down by month to see the best entry point. Definitely performs best starting in June when the winning percentages are more meaningful. However, the results are positive for April, so I'm going to go ahead and start tracking now to see what happens.

Games      RL Record                     RL Avg Line    $RL On    Month
74             29-24 (0.16, 54.7%)             122.9       +$1,112       4
119           41-49 (-0.31, 45.6%)             124.5           +$92       5
196           77-72 (-0.10, 51.7%)             126.7      +$2,487       6
229           82-87 (0.05, 48.5%)              124.1      +$1,518       7
287           104-105 (-0.30, 49.8%)         123.3      +$2,267       8
275           102-94 (-0.09, 52.0%)          121.5      +$2,885        9
11             7-3 (1.70, 70.0%)                 117.8        +$561       10

Just going to flat bet one unit per play.

Quote Originally Posted by Scrivero:

Still sounds good.

A question: How is the variance in this system? E.g. how much did each season go negative before ending up positive?

I am asking this because I want to tail one MLB system but I would like it to be a rather low variance one (because I can not estimate the variance as I am new to MLB), or basically one that does not normally go that much on minus before reaching a positive result in the end.

And that is because with a lower variance system I can have a higher unit size or bet more units per game or even just have more systems that I can play at the same time with the smallest possible chance (out of all possible chances, not gonna say that there wont be any risk) of actually losing my whole BR if things go really awry.

Quote Originally Posted by Scrivero:


As you already know, I have never run SDQL queries, but I am eager to learn. So, I tried the query you used for this but with a little tweak:

Your query was: Query: AF and total>=8 and 50<=WP<=55 and 41<=o:WP<=51

And the results I got with that, which were very close to yours:
RL:

	442-435 (-0.11, 50.4%)	avg line: 123.6 / -136.4	on / against: +$10,822 / -$16,098	ROI: +11.9% / -13.2%

I made a tweak and made this query: AF and total=8 and 50<=WP<=55 and 41<=o:WP<=51

And the results were:

RL:	106-87 (0.01, 54.9%)	avg line: 125.2 / -136.9	on / against: +$4,397 / -$5,594	ROI: +22.4% / -21.0%

So apparently, if I havent gotten this all wrong, betting on games that have an O/U line of exactly 8 brings a ROI of +22,4 % compared to the over or exactly 8 that brings a ROI of +12 %.
Of course the amount of games goes down a lot so its not as trustworthy as with the >=8. Maybe the variance is less though. I dont have the skills to check that.

Please correct me if I got it all wrong. If its right though, should we try to use it and tweak the system more? Should I try to make it into my first MLB system ever without knowing anything about the game or about betting it :D

Could you possibly check that exactly 8 O/U by seasons? I dont know how to do it. I could just take the results to excel but there seems to be a faster/easier way.

Quote Originally Posted by nfl_huskers:



Hey Scrivero!

I'll crunch the numbers to get a better picture of the variance like I did for the NHL system. I'll have more time over the next couple days.

Quote Originally Posted by nfl_huskers:




You are absolutely correct about the query with total=8, and it looks like a good one to track. I usually just use an inequality to include multiple totals, and take a hit on ROI to lower the variance a bit. A nice increase in ROI for a specific value is usually caused by one or two seasons being better than normal, but in this case, it's spread very well across seasons.

If you want to group by season, just add the word "season" to the query with no values associated with it ...

Query: AF and total=8 and 50<=WP<=55 and 41<=o:WP<=51 and season

I removed some columns from the result, but here it is:

games           RL                   Avg Run Line    $ RL On   
14         12-2 (1.93, 85.7%)       111.1              +$1,295           season = 2007
17          8-9 (-1.44, 47.1%)      121.2               +$70               season = 2008
16         6-10 (-1.12, 37.5%)     116.2                -$390               season = 2009
10         7-3 (0.50, 70.0%)       124.8               +$590              season = 2010
20        10-10 (0.30, 50.0%)     133.8              +$349     season = 2011
22       12-10 (0.68, 54.5%)     121.8               +$427     season = 2012
18       10-8 (0.22, 55.6%)       125.9              +$404     season = 2013
31       16-15 (-0.76, 51.6%)    125.4               +$420     season = 2014
27       16-11 (0.24, 59.3%)     131.5               +$996     season = 2015
18       9-9 (0.28, 50.0%)        132.3               +$236     season = 2016

You can see that only one season (2009) finished negative. Nice!

For my system, I just found the best range of totals and filtered the winning percentages based on that range. Ideally, we would look at each total individually and tweak the winning percentage for each one to get a much higher ROI. Then, basically, combine the best of each Total into one system.

My goal for now is to see how this system plays out independent of the lines and data in the SDQL database, but please feel free to track your research here. I'm curious how high you can get the ROI by filtering at each specific Total value.

Quote Originally Posted by Scrivero:



I tried this to get the totals for all seasons by months (e.g. all seasons May stats):
AF and total=8 and 50<=WP<=55 and 41<=o:WP<=51 and season and month
- but it didnt work, what am I doing wrong..

I tried to get the lowest point in negative for each season:
AF and total=8 and 50<=WP<=55 and 41<=o:WP<=51 and season and low
- but did not work, cant figure it out.

Quote Originally Posted by Scrivero:

Still sounds good.

A question: How is the variance in this system? E.g. how much did each season go negative before ending up positive?

I am asking this because I want to tail one MLB system but I would like it to be a rather low variance one (because I can not estimate the variance as I am new to MLB), or basically one that does not normally go that much on minus before reaching a positive result in the end.

And that is because with a lower variance system I can have a higher unit size or bet more units per game or even just have more systems that I can play at the same time with the smallest possible chance (out of all possible chances, not gonna say that there wont be any risk) of actually losing my whole BR if things go really awry.

Quote Originally Posted by nfl_huskers:




If you want the May results consolidated for all seasons, use the following:

AF and total=8 and 50<=WP<=55 and 41<=o:WP<=51 and month=5

If you want each month's results across all seasons, use the following:

AF and total=8 and 50<=WP<=55 and 41<=o:WP<=51 and month


I currently use a spreadsheet to get the lowest point in each season. I'm not sure if it's possible via query. There is a "profit" value, but it's only for the money line bet.

Quote Originally Posted by nfl_huskers:



Assuming each season was played start to finish ...

Season          Low Point          High Point          End Of Season
2007                -15.50u                 +7.20u                  -4.40u
2008                  -7.35u                 +6.20u                 +6.20u
2009                  -0.75u               +10.95u                 +8.05u
2010                  -4.63u                 +8.40u                 +5.03u
2011                  -2.13u               +15.36u                +14.36u
2012                  -3.48u                 +8.98u                 +8.98u
2013                  -6.62u               +13.26u                +13.26u
2014                  -3.18u               +21.63u                +21.58u
2015                 +0.84u               +21.49u                +19.49u
2016                 +1.42u               +19.12u                +16.67u

Not too bad after 2008.

Quote Originally Posted by Scrivero:


Thanks for these man. I will also use a spreasheet for my future systems lowest point calculations.

You have given me so much with teaching this stuff, I cant thank you enough. I probably cant even return the favor before I learn more. I do have some innovative system ideas and sometimes a LOT of spare time in my hands so what I will do is that I will at least offer some reflection and conversation on this system stuff. I can also calculate some stuff that needs a spreadsheet if you wouldnt have the time to do it. Just ask whenever it would be needed.

Quote Originally Posted by Scrivero:


Not bad for sure, though certainly more variance than in the MLB under, which is of course expected too with just can 50 games per season in the Under system and ca 85 per season in this RL system.

If I have the time, I will try to check how each season did during the first 2 months (individually, because you already calculated the total for each month). Maybe there is some freak season with a huge profit for April and otherwise all would be negative. Same with May. If so, then one could start betting this from June.

I am also very intrigued by the aspect of trying to start tailing systems with the lowest point possible. That is because if one can start tailing any system at its lowest point, then all systems, even the not profitable ones, suddenly turn into profitable systems.

The more plausible way would be to start tailing a profitable system when its halfway to its lowest point. E.g. a +10 units system (lets assume 10 % ROI, 100 games, 1 unit bet per game) is down 3 units (lets assume in 15 games, 1 unit bet per game). We start tailing. It goes all the way down to minus 6 units, climbs all the way up to +15 units and then comes down to the +10 units in the end.

So by waiting for it to go down 3 units I have increased my ROI from 10 % to 15,3 %. How? Well, like this:
- Normally we would bet 100 games (100 units) to profit 10 units, a ROI of 10 %.
- Now we only bet 85 games (85 units) to profit altogether 13 units, a ROI of 15,3 %.

And what if we figured out the point of bailing out from the system before it ends? Well, the highest point was +15 units. Lets say we again bail out in the middle of the final +10 and that +15, so at +12,5 units. Lets assume that the final +2,5 and then -5 units happened in 20 games. The calculations would be:
- We bet 65 games (65 units) to profit 15,5 units.
- A ROI of 23,8 %.

THAT is where my goal is after a few years. Well, its the goal even now but I probably wont get there yet as I am too inexperienced with this stuff.

What I will do now is that I will check good +ROI MLB systems and see which one of them goes down a lot and then start tailing it. The risk is that it never goes down, it just goes up and I can never start tailing it. But that is ok as there are so many systems available and Im not gonna do more than 1 or 2.

Quote Originally Posted by nfl_huskers:



Interesting idea with tailing the system towards its low point before starting the betting.  You definitely miss out on those seasons where it never goes negative, but gain the extra ROI if it goes down and rebounds. Same for bailing on the positive side. Makes for some interesting analysis to find strategic entry and exit points.

Quote Originally Posted by Scrivero:


For sure. I would also love to find more than one entry and exit point per system. That is, to start it when its down, exit when its up, get back in when it gets lower, and end it as high as possible. Probably very difficult but still.

(By the way, I did now two entries to two MLB systems that were down -6,1 and -31 units. Lets see where they go.)

You basically had that idea (of entry and exit and re-entry and exit) already in the idea of the NHL system in the other thread, with skipping specific months here and there. It would be of course even better to try to find more specific time frames to skip, such as in NHL from the middle of January to the first week of February. Because a month is still such a long period with a season that has 8 months when there are games.

Still, skipping full months is still the easiest way to to go because, well, its easy. All the data from SDQL can be taken monthly, I have no idea if one could try to tweak the query to make it another, different period of time during a season.

Anyways, lets see when the first games of this system come up :)

By the way, I still have that "to check how each season did during the first 2 months (individually, because you already calculated the total for each month). Maybe there is some freak season with a huge profit for April and otherwise all would be negative. Same with May. If so, then one could start betting this from June." to do. Not easy to find the time but I will try. It doesnt even take that long but just need to find uninterrupted time (wife, kid, you know).