Over/Under strategies in Major League Baeball

**InTheHole** · 05-03-09, 11:44 AM

And By the Way, I am similarly illogical when it comes to following traffic data. When the sign says take the upper level because it has less traffic I take the lower level (for those living in New York City and taking the GWB)

**suicidekings** · 05-03-09, 01:42 PM

Originally posted by InTheHole

If multpiple "experts" are suggesting a strategy wouldn't it make sense that the experts setting the lines already factor in such data.

I think that while the linesmakers are aware of the YTD O-U record, it wouldn't make sense to forecast too far in advance, and "experts" saying that we're due for a correction in May is completely arbitrary. I would feel more comfortable continuing to bet the trend as opposed to trying to predict it's reversal.

Maybe certain pitchers should be avoided (high achievers on low scoring teams), but I like the over trend to continue.

**MonkeyF0cker** · 05-03-09, 01:44 PM

No expert is doing any blanket forecasting like this.

**twister** · 05-04-09, 12:35 PM

Here's a O/U tip:

When the away team wins, the over comes through more often than the under.

For the last 12 seasons combined, when the home team wins the O/U is 47.4%/52.6%, whilst when the away team wins, the O/U is 51.5%/48.5%.

Now, this information seems useless as you don't know who wins until AFTER the game. However, the wise way to use this information would be to not only cap the O/U, but to also cap the ML. Then only play the over when you think the away team will win, and only play the under when you think the home team will win.

That itself should increase your profitability on totals. (Unless you are rubbish at capping ML

).

**Arilou** · 05-05-09, 08:04 PM

twister, this effect is small enough that trying to use your own ML instead of the actual ML is likely a waste of time; if you have a big enough ML edge to have an impact, bet the ML instead! On top of that, I'd make sure that games with say home dogs actually go over more often to make sure the line doesn't simply adjust.

**twister** · 05-06-09, 07:49 AM

Originally posted by Arilou

twister, this effect is small enough that trying to use your own ML instead of the actual ML is likely a waste of time; if you have a big enough ML edge to have an impact, bet the ML instead! On top of that, I'd make sure that games with say home dogs actually go over more often to make sure the line doesn't simply adjust.

This wasn't an "and/or" type tip, in terms of ML or Totals. It was a combination. i.e if you are a +EV ML capper (which means your ML are better than the actual ML), and a -EV/breakeven Totals capper, use your ML talent to increase your profits on totals (or reduce the number of games you bet on, thus increasing your ROI).

For example, I'm new to capping MLB Totals so I've been "paper betting" to see how good/bad I am. Capping almost every single game on the board (aside from game 2 of any double headers), from 04/15-05/05 I am:-

152-109-13 +29.1 units.

Now, the results where I capped the total to go under, but only "placing the bet" if I had capped the ML value to be with the home team, or where I capped the total to go over, but where I capped the ML value to be with the road team, I am:-

110-69-9 +30.99 units.

See, althought my total units has barely increased, I have reduced the number of games (and thus amount staked) by around 85-86 games. This drastically increases my ROI, my win %, and therefore I could safely use slightly bigger units to increase my actual $$ profit without my risk-of-ruin increasing (in comparison with situation 1).

**Data** · 05-06-09, 10:39 AM

twister, that does not refute the point that Arilou made. The problem with your reasoning lies in the contradictory view on correlation between side and total. While you reasoning is based on the right assumption that such correlation exists, your betting scheme does not account for that correlation at all. There should be no "improving" changes in your totals betting numbers because all of the additional edge you'd found due to moneyline capping would be accounted for in the bets you'd placed on the moneyline. Were you to place a bet on total due to moneyline capping then that bet would be correlated to the moneyline bet you already placed and therefore placing that bet would be incorrect.

**twister** · 05-06-09, 12:07 PM

Originally posted by Data

twister, that does not refute the point that Arilou made. The problem with your reasoning lies in the contradictory view on correlation between side and total. While you reasoning is based on the right assumption that such correlation exists, your betting scheme does not account for that correlation at all. There should be no "improving" changes in your totals betting numbers because all of the additional edge you'd found due to moneyline capping would be accounted for in the bets you'd placed on the moneyline. Were you to place a bet on total due to moneyline capping then that bet would be correlated to the moneyline bet you already placed and therefore placing that bet would be incorrect.

Data, having read some of your previous posts I know you are quite well informed in most things statistical, so I will hold your view in high regard.

If you are consistantly more correct than the bookmakers as to who wins and who loses, should it not stand to reason that when your ML capping and Totals capping deduce that the play is Away/Over or Home/Under, the 'Totals' have more +EV to hit in those combinations than a Away/Under or Home/Over?

If the bookie thinks the away team should hit 4200/10000, and the away/overs occurs 51.5% of the time, then that is 2163 overs. If however, I have an edge and I hit 4500/10000, than 51.5% of that is 2317.5 overs.

This means the bookmaker has underestimated the away side, and overestimated the home side. Given that home/overs occur ~48.5% of the time, than 48.5% of the bookmakers 5800/10000 for the home side, is 2813 overs. In reality, the home team only wins 5500/10000. 48.5% of 5500 is 2667.5 overs.

Does this not mean that when the home side win (against my prediction of away), that the chances of 'overs' hitting are 145.5/10000 (1.45%) less than what the bookmakers predict.

Ofcourse, this is all dependent on the factor of having a consistant edge over the bookmaker, but in the case that you do, is my understanding not correct that given that you are consistantly better ML capper over the bookies, taking the 'over' when your ML capping denotes 'home', or taking the 'under' when ML capping denotes 'away'decreases the probabilty of you hitting?

Or is my maths completely wrong?

**Data** · 05-06-09, 12:51 PM

Originally posted by twister

Or is my maths completely wrong?

twister, I skipped through the math and I assume it is fine. The problem is that it is only fine under the assumption that you do not make moneyline bets and that must be wrong as your edge in moneyline bets is bigger than in totals. Yes, you do have an edge in totals but that is not the reason to make bets if you can make bets with higher edge.

**twister** · 05-06-09, 01:01 PM

Originally posted by Data

twister, I skipped through the math and I assume it is fine. The problem is that it is only fine under the assumption that you do not make moneyline bets and that must be wrong as your edge in moneyline bets is bigger than in totals. Yes, you do have an edge in totals but that is not the reason to make bets if you can make bets with higher edge.

Got you.

Essentially it boils down to which has the bigger edge, which is obviously ML.

**u21c3f6** · 05-06-09, 01:04 PM

Originally posted by Data

twister, I skipped through the math and I assume it is fine. The problem is that it is only fine under the assumption that you do not make moneyline bets and that must be wrong as your edge in moneyline bets is bigger than in totals. Yes, you do have an edge in totals but that is not the reason to make bets if you can make bets with higher edge.

Data, I would agree with you 100% if the scenario was that you could only make one wager, then of course wager on the one with the higher edge. But these are two different wagers which while correlated are not necessarily dependent on each other. I am sure there is some complicated math formula to figure out how to wager on both but I would look at these as two separate wagers and would wager an amount based on the edge of each independently. I don't say this is right or optimal, just the way I do it.

Joe.

**Data** · 05-06-09, 02:51 PM

Originally posted by u21c3f6

Data, I would agree with you 100% if the scenario was that you could only make one wager, then of course wager on the one with the higher edge. But these are two different wagers which while correlated are not necessarily dependent on each other. I am sure there is some complicated math formula to figure out how to wager on both but I would look at these as two separate wagers and would wager an amount based on the edge of each independently.

Joe, we do not need complicated math here, just a common sense. What you are saying is applicable to the bets twister made before moneyline adjustment which were supposedly uncorrelated to his side bets. However, under given scenario, the additional edge in totals was due to correlation to the big edge in sides. Therefore, there was no additional edge in totals to justify separate (additional) bets on totals.

**reno cool** · 05-06-09, 06:16 PM

there's a reason why road winners should have higher totals. However, it seems that historically overs are bad bets in general. So I tend focus on home/unders.

**Pancho sanza** · 05-06-09, 06:42 PM

This overs in April phenomenom is something that came about during the roid era, scoring got higher and higher and it took the books a month or so into the season to adjust the totals.

I backtested this for about 10 years worth prior to the roids period and it was a loser.

**kokky** · 05-08-09, 10:11 AM

When I look game totals in previous games I see that there isn't many series like OOOOOOO or UUUUUUUU. Its much more often OOUOUUOUO or something similar. Thats logical, because bookies predict outcomes relatively accurately, its hard to find edge.
So I'm wondering is there a rule like this perhaps: after 4 over games there is 65% probability that fifth game will be under, or after 2 overs/ unders, or any regularity in this?
I think that resoults in one or two seasons would be enoguh to get significant resoult, but that is lots of analising.
Does someone have this kind of data?
What do you think about this?

**twister** · 05-08-09, 10:24 AM

Originally posted by kokky

When I look game totals in previous games I see that there isn't many series like OOOOOOO or UUUUUUUU. Its much more often OOUOUUOUO or something similar. Thats logical, because bookies predict outcomes relatively accurately, its hard to find edge.
So I'm wondering is there a rule like this perhaps: after 4 over games there is 65% probability that fifth game will be under, or after 2 overs/ unders, or any regularity in this?
I think that resoults in one or two seasons would be enoguh to get significant resoult, but that is lots of analising.
Does someone have this kind of data?
What do you think about this?

You can get this data for free from here http://www.atsdatabase.com/Stats/MLB...t-Results.html

Just click, "<team name> records played last 5 years" - then copy the entire table, then paste into MS Excel via [right-click, "paste special", "text"].

At the top of the link, you will see all the other MLB teams. Do the above for all the other teams, and you will have a database of all results, closing lines, results, scores, closing totals, and totals result for the past 5 years.

From this, you can write some excel functions to determine when 'x number of overs' have occured in a row, and thus how many of the following totals results are either over/under.

**kokky** · 05-08-09, 11:44 AM

Thanks twister :-) There are all stats that I wanted on that site.

But I don't know how to make that function. I'm not good in programing.
Can you or somebody else please explain how to do that? Then I'll calculate it all and post it here. I'm sure these data will be helpful to many readers.

**idontlikerocks** · 05-09-09, 02:29 AM

home teams usually do not bat in the bottom of the ninth inning in games they win. perhaps this accounts for the corelation in question?

**Arilou** · 05-09-09, 01:30 PM

idontlikerocks, that is exactly why it happens.

Kokky, remember when you search for such things that you're never going to find a 65% (or even a 53%) with a search like this.

**reno cool** · 05-10-09, 03:07 AM

half inning is worth about half a run. I would think that's a considerable amount. Maybe I'm missing something, but I don't see how a book avoids this correlated parlay.