Reverse Middle Scalps: Putting the House Percentage in Your Favor.

[AussieVamp2]

very interesting, thanks, have to have a look at last year's numbers now

[AussieVamp2]

Cons,

A middle is winning both sides of a bet - not something going to happen very often with what you bet on

i.e. say I bet Chicago +12.5

And Los Angeles -10.5

or something like that that I backed earlier in the year and lucked out

if Los Angeles wins by 11 or 12, that is a middle, it has landed in the 'middle' and both bets pay

a scalp is backing Los Angeles at 2.10 and Chicago at 2.15 and you win no matter what happens, but only a percentage, not whole bets.......

[AussieVamp2]

ok, here's a whole season sample :-

1 34 1.4%
2 41 1.7%
3 120 4.9%
4 95 3.9%
5 240 9.8%
6 150 6.1%
7 256 10.5%
8 172 7.0%
9 243 9.9%
10 172 7.0%
11 206 8.4%
12 114 4.7%
13 132 5.4%
14 87 3.6%
15 121 4.9%
16 54 2.2%
17 64 2.6%
18 38 1.6%
19 29 1.2%
20 24 1.0%
21 19 0.8%
22 9 0.4%
23 6 0.2%
24 6 0.2%
25 8 0.3%
26 1 0.0%
27 1 0.0%
28 1 0.0%
29 1 0.0%
30 1 0.0%
33 1 0.0%


have another one, but not quite complete

[AussieVamp2]

seems that the even number total run numbers happen less than the odd, around the general total range - which makes some sort of sense without analysing/thinking about it perhaps, as have to have a winner in baseball - you score in units of 1 run - 1 run wins are very common, by all accounts

and looking at the above numbers, overall, around 61-39 split in favor of odd totals for baseball game results overall that season.......

[AussieVamp2]

can't help with favorite for the above at the moment, but here are the numbers for that season with home team winning

1 18 1.4%
2 22 1.7%
3 67 5.1%
4 63 4.8%
5 127 9.7%
6 85 6.5%
7 150 11.4%
8 85 6.5%
9 136 10.4%
10 93 7.1%
11 109 8.3%
12 48 3.7%
13 70 5.3%
14 37 2.8%
15 61 4.6%
16 26 2.0%
17 37 2.8%
18 16 1.2%
19 18 1.4%
20 13 1.0%
21 14 1.1%
22 5 0.4%
23 3 0.2%
24 2 0.2%
25 5 0.4%
27 1 0.1%
29 1 0.1%
30 1 0.1%

[chance_of_rain]

From April 3 - May 29 of this year, 224 of 738 games were 1 run games. That's 30.3%.

The following stats are over-under-push for specific O/U lines this year Apr 3-May 29.

O/U# O-U-push

7.0 1 over - 2 under - 1 push(landed on 7.0)
7.5 13-10
8.0 12-11-1
8.5 25-21
9.0 55-45-8
9.5 61-65
10.0 48-51-5
10.5 80-80
11.0 34-43-4
11.5 12-15
12.0 4-0-0
12.5 1-1
13.0 2-1-1
13.5 4-5
14.0 5-2-0
14.5 3-4
15.0 2-0-0

TOTAL 362 over - 356 under - 20 push

49.1% over - 48.2% under - 2.7% on the number


Of the 20 games that landed on the number, 14 (70%) were odd O/U numbers, 6 (30%) were even O/U numbers, similar to the 61-39 that AV2 quoted for last year.

[pay2play]

AV, the stats you post (let me first thank you for taking the time) aren't quite what we need, I don't think. The chance of scoring exactly 10 runs in a game where the average is expected to be 10 runs is probably greater than the chance of scoring exactly 10 runs in games where the average is expected to be, say, 14. It's the top of the bell curve in the first sample, on the left hand side in the second (I'm not spelling this out for you, I know from your other posts you're not math illiterate, but we Americans often are, so its for the other readers) It might not be a very steep slope, but there's some slope nonetheless. Do you have a such a breakdown? I just think that as a total tends to be smaller, and if it's odd, it's more likely to land on the exact # than if it's large or even, and otherwise sample results can be averaged to increase their predictive value. Maybe all we need is the average chance of even #'d totals landing, and the same chance for odd#'s. I think this would be more accurate than an historical sampling of the chance alone of 11's or 13's or whatever hitting, unless someone can suggest a reason why 11 totals are more likely to hit than 13's. It's going to depend on the distribution, obviously. If we see that the totals land flat in 9's and 11's more than 7s and 13's, for example, we might presume that scores tend toward the league median even in games with off-median numbers.

For the purposes of bucking the middle, which is what BOOMER started this thread about, the chance of YOU getting middled is your cost, and you'd better be pretty sure you've got it right, enjoy hours in front of your computer, and be ready for some swings. Nonetheless, I think it probably does provide an arbitrage opportunity.

To whomever asked how I developed my formula for figuring totals value, I really don't remember. I don't bet totals. I just did it one day, it's hardly advanced math, because a friend who bets seriously asked me to. I suppose I just took a small sample and estimated a run's impact on winning percentage. I suppose further I must have figured 1 run's increase to your chance of beating a total from 50% to 58%? Amazing how fogged my brain gets when I'm tired. (let me see, 1=1=2, okay, 2+2=, uhhh). Its just a simple formula, adequate for hand calcs by non-math types, which is what my friend needed. I would warn you, as I did him, that to keep it simple some accuracy is lost, most obviously to me in that a 2 run variance is not twice as valuable as a 1 run variance, a 3 run variance not 3 times as likely, etc. But for comparing value between #'s with money lines, which is what you sometimes need to do betting bases totals, then it does come in handy. The value difference between 10,-110 and 10,-120 is obvious; between 10,-110 and 9.5,-125, not so clear. But again, I'm not a tout, I don't bet totals myself, so do your own work, or, hopefully, lets get someone to post here with a better approach. I hardly expect my 30 minutes of work on a Sunday afternoon a few months back to be the end-all to the discussion.

I've only really put effort into developing a run-line/money-line value formula, which I know is accurate, but also less valuable, because a dime money line usually (but not always) has more value than a 20 cent run-line.

[AussieVamp2]

pay2play - I realise the limitations of the above, and unforunately no, don't have the conditional type probabilities you are after at my fingertips - have to see what I have lying around in other places for a season's worth of stuff

[pay2play]

Looking at Rain's sample, which is too small to figure a good percentage, we can nonetheless see that, yes, odd #s hit more than even ones, and lower #s hit more than bigger ones, which makes sense.
Now we just need the largest possible sample, and I don't think going back to before the league's scoring increase will hurt the sample's validity because we're controlling for totals.

[Boomer]

A couple of things. First of all, I know where to come when I have a math problem. Some of you guys blow me away with your math knowledge. I don't even know what calculus is.

Second of all, don't you agree that you have to break down American versus National League because of the Designated Hitter Rule when creating your statistical data.

And it's really interesting to figure how many games land on the number when the BM make it that the number. That seems like a more intelligent way to analyze the data as that is exactly what you're concerned about.

I've put away a hypothetical percentage of bankroll and will keep real tight logs on how this strategy works over the course of this season (provided this designated bankroll doesn't go broke first).

I will segregate these wagers from my other wagers and report on my progress as the season advances.

Okay, now some of you money management gurus, let's say I start with a hypothetical bankroll of ten dimes, how much specifically should each reverse middle scalp attempt be. Based on my experience, I hope to have two to six wagers a day on this.

[mikren]

This is an intriguing idea, but it looks to me like there may be an error in the calculation of one's edge in Boomer's original post. Assuming a 10% chance the game will land on the number, a 20 cent difference will not give you a 10% edge, in fact, you will come out slightly behind.

For example, say you bet

100/110 at 10 Over +110, and
100/110 at 9 Under +110

If you bet 10 games, you will push 1, losing $100, the other 9 times you will win $10 for +$90, for a net loss of -$10.

Therefore, using your original assumption that you wish to be conservative and have double the edge, you would need at least 40 cents difference rather than 20. This would give you the 20 cent edge you are looking for no matter if the under or the over occurs. Did I miss something here? Actually, the only reverse/middle I ever bet was by accident where I bet Over 8.5 at +120 and Under 8 at even. Both places had read me 8.5s when I originally called and unfortunately I had loaded up on Over 8.5 when the second place told me their line was 8 and not 8.5. The game was 2-0 in the 8th, but of course ended 5-3.

[bsky2]

Chance of rain,

Where do you get your baseball stats? Do you subscribe to a service or have you found a site for free?

[chance_of_rain]

I keep my own stats using Excel.

[Boomer]

Mikren,

You confused me with your math in your post.

The way I figure it is if I am collecting 10 cents every 10 games that equals one dollar and that is my break even point.

Therefore, if I am collecting 12.5 cents every ten games, after 10 games, theoretically I lose one game, and have a net profit of 25 cents.

By ascertaining I receive a total of 25 cents which ensures me 12.5 cents on each side, my break even point is 1 in 8 games, because 8 multiplied by 12.5 equals 100.

Right?

[reno]

Boomer, calculus is math used to compute the area under a curve. It was invented by Playtex so they could make bras that would properly fit women.

Scalpers/middlers ARE 2nd(or middle)-class citizens. Accept your place in the sportsbetting hierarchy. Winners are 1st-class citizens, and losers/squares are 3rd class. Until Mahatma Gandhi reincarnates as a sportsbettor and attempts to destroy this sportsbetting "caste" system, it is what it is.