Given the moneyline and total, why are NL home teams less likely to score first?

[mathdotcom]

Could just be a 2009 phenomenon, but to put it simply, if you have two games that have identical moneylines and totals, but one is an AL game and one an NL game, the NL home team will be less likely to score the first run.

[chaka]

wouldnt that be due to the run producing DH in the AL batting near the top of the order vs in the NL the weak hitting pitchers batting ninth?

[mathdotcom]

If anything I'd expect the opposite result - by having a stronger top of the order in the AL, the first mover advantage is stronger, hurting the home team's chances of scoring first.

[chaka]

ok i gotcha...i wasn't on same page at first- i was thinking which of the two home teams would score first (like a prop bet)rather than AL Home vs AL Away and NL Home vd NL away scoring first

i think there are too many changing variables in play to have a definite answer.
were two good hitting teams facing top pitchers in one league vs two poor hitting teams vs avg pitching?
Hitters park vs pitchers park.
Is Weather or injuries altering one of the lines where on a normal day the lines would not be the same?
One home team may face a #1 or # 2 starter and the other home team faces a # 4 or # 5


btw what were 2009 results? I am just scratching the surface trying to learn to cap mlb and am curious how many times this situation occured and how low the totals were?

[mathdotcom]

chaka all those things you listed are incorporated into the moneyline or the total

[bztips]

Sounds like a classic case of data mining/dredging/snooping.../ whatever you want to call it.

[mathdotcom]

Sounds like a classic case of data mining/dredging/snooping.../ whatever you want to call it.

Not even close.

[Pancho sanza]

If anything I'd expect the opposite result - by having a stronger top of the order in the AL, the first mover advantage is stronger, hurting the home team's chances of scoring first.

Why do you say the AL has a stronger top of the order?

If the Moneyline/totals are the same, and the NL has 1 very weak hitter (the pitcher), would that not imply that the other hitters as a whole (excluding the 9th hitter) are stronger in the NL side?

[mathdotcom]

You're right, with moneyline and totals the same for an AL game and an NL game, the result I've found would seem to imply that the top of the order in the NL is stronger.

But is it? I thought conventional wisdom was the opposite.

[Data]

Could just be a 2009 phenomenon, but to put it simply, if you have two games that have identical moneylines and totals, but one is an AL game and one an NL game, the NL home team will be less likely to score the first run.

While in practice one must consider individual players strengh and position in the order while betting these props, in theory, not accounting for the players, this phenomena has a simple explanation. If the lines are equal that means that
* the NL home team expected runs are a few hundreths of a run lower than the AL home team expected runs
and
* the NL away team expected runs are a few hundreths of a run higher than the AL away team expected runs

Both these factors contribute to the phenomena observed.

[mathdotcom]

While in practice one must consider individual players strengh and position in the order while betting these props, in theory, not accounting for the players, this phenomena has a simple explanation. If the lines are equal that means that
* the NL home team expected runs are a few hundreths of a run lower than the AL home team expected runs
and
* the NL away team expected runs are a few hundreths of a run higher than the AL away team expected runs

Both these factors contribute to the phenomena observed.

And what is an explanation for these factors? If it is top of the order NL hitters are on average "stronger" than their AL counterparts, then fine, but if that's the explanation I do find it surprising. It could be that AL managers are more likely to occasionally put slumping hitters at the top of the order, etc., etc. I am just curious if anyone has a good theory.

On a side note, where are these hundredths coming from? I didn't say by how much the two leagues differ.

[tomcowley]

If you ignore ML/total, and just look at the leagues as a whole, does the home team score first more often in NL or AL?

[Data]

The discrepancy is due too different average runs expectancy in AL and NL. The lines are just the reflections of the expected runs for both teams AND league average expected runs. If the reflections are identical but the averages are different that means the teams expected runs are different too. How? Explained in the post above.

[mathdotcom]

NL home team scores first 45.97% of time
AL home team scores first 51.62% of time

[buby74]

In the nl the first inning has 15% more runs scored then average with the second inning 15%lower due to the pitcher. So if an al and nl game has the same expected total the nl game will see more runs in the first which might explain the results. Try looking at who scores first in the third inning when the lineup will be scrambled

[djiddish98]

Is there any discrepancy between the average inning when the first run is scored by league?

Probably not, but that just popped into my head.

[Justin7]

Consider this scenario:
Game 1 has two teams with a game total of 8, but both teams have many home-run hitters, but a lower overall batting average.
Game 2 has two teams with a game total of 8, but neither team has many home-run hitters (but have a higher overall batting average).

The teams in game 1 are more likely to score in the first inning than those in game 2. When there is an inning with a score, the teams in game 2 will have a higher average score in that inning than the teams in game 1.

The next thing I would check: does the AL have more home runs than the NL?

[mathdotcom]

So for that to be the explanation, it means NL hitters hit more home runs, or that the top of the order hits more home runs.

I don't think this is true.

Remember we're talking league averages.

[Justin7]

So for that to be the explanation, it means NL hitters hit more home runs, or that the top of the order hits more home runs.

I don't think this is true.

Remember we're talking league averages.

I think it's just the opposite. If two teams have an identical median team total, the one that hits more home runs is more likely to score in any given inning. There was an article on this here:

http://sloansportsconference.com/2010/research-papers/beyond-pythagorean-expectation-how-run-distributions-affect-win-percentage/

[Flying Dutchman]

Firstly, what are the actual numbers? not the %s. Then we can get an idea of how many of these things we're talking about. Calc a z-score.

Did it happen last year? If not, most likely it's transient, even with a high z-score.

If you can't find a decent reason, most likely it's a transient. BUT, keep your eye on it rest of the year, and then next year.

I'd give pretty good odds it'll go away.

[Data]

Justin7, for the benefit of the forum you moderate, how about I explain this slowly to you and then you explain it to the interested masses?

[Justin7]

Go for it, Data. Re-explain it.

I read your explanation. While I agree that the NL has a lower run average per game than the AL (it was about 3/10th of a run per game when I last checked), this should be reflected in the game total. Your typical AL game total is a bit higher than the average NL game.

One of the caveats of Mathdotcom's hypothetical was that the totals were identical. With that caveat, the discrepancy between mean AL and NL scoring should (in theory) be removed.

[Data]

OK, I think I can see where to aim. Just as the leagues have different averages their runs scored distribution curves differ a little bit. The lines match certain values of two (home and away) distributions' integrals. Since the distribution curves are different, the different integral values correspond to different expected runs.

[MarketMaker]

With both totals being equal the NL obviously has to have the probability of producing a run that is lost on the number nine spot distibuted somewhere else in the batting order. This means that the other eight batters would be on average individually stronger than each of the nine batters on the AL team.

A more obvious example would be if each NL team had four pitchers batting in spots six thru nine (or just four instant outs). Then the run production would have to come from the first five batters, giving the team to act first in the inning a huge advantage to score first.

[ThaddeusB]

With both totals being equal the NL obviously has to have the probability of producing a run that is lost on the number nine spot distibuted somewhere else in the batting order. This means that the other eight batters would be on average individually stronger than each of the nine batters on the AL team.

This is the correct explanation. We start with the given that the expected run production for the game is the same. Assuming each player contributes equally to that total (which obviously isn't really the case), the AL batter contribute 11.1% (100/9) whereas the NL batter contributes 12.5% (100/8). That extra 4+% allocated to the top 3 batters means the visiting team scores in the first inning more often in the NL.

[mathdotcom]

Data, I was not looking for a statistical explanation. I was looking for a theoretical (i.e. baseball) explanation - one that an MLB manager might give me. MarketMaker has provided a simple and intuitive explanation .

Data, instead of insinuating that the rest of us are retarded and in need of a Justin7 translation, you might want to learn how to
a) read the question asked
b) learn how to translate jargon into something a layman can understand, let alone someone with an undergrad background in statistics

[mathdotcom]

I think it's just the opposite. If two teams have an identical median team total, the one that hits more home runs is more likely to score in any given inning. There was an article on this here:
http://sloansportsconference.com/201...in-percentage/

I wasn't doubting that the team that hits more HRs is more likely to score in any given inning; was just commenting that for the NL to have the away team scoring first more often, given your explanation, it would mean NL lineups tend to hit more homeruns (or at least the top of their orders).

Justin7 your book is slow to arrive to my mailbox. I bet the mail man stole it and is betting into square lines as we speak

[mathdotcom]

With both totals being equal the NL obviously has to have the probability of producing a run that is lost on the number nine spot distibuted somewhere else in the batting order. This means that the other eight batters would be on average individually stronger than each of the nine batters on the AL team.

A more obvious example would be if each NL team had four pitchers batting in spots six thru nine (or just four instant outs). Then the run production would have to come from the first five batters, giving the team to act first in the inning a huge advantage to score first.

Even without accounting for moneyline/total, the NL away team scores in the first inning 28% of the time compared to the AL's 23%.

[Flying Dutchman]

Data, instead of insinuating that the rest of us are retarded and in need of a Justin7 translation, you might want to learn how to
a) read the question asked
b) learn how to translate jargon into something a layman can understand, let alone someone with an undergrad background in statistics

This from Mr. Manners.

[mathdotcom]

If he had replied that I was a moron and then gave me MarketMaker's explanation, I would have had no problem with that

[Data]

One more time, there are two underlying reasons.

While in practice one must consider individual players strengh and position in the order while betting these props

One is very basic and obvious even for a novice researcher/handicapper. Since it is basic, I only mentioned it above. Later on MarketMaker explained this in more details (the effect is not nearly that strong as in the given example but the logic applies).

The second concept is more advanced and not obvious, it is based on runs distributions. Note, this concept applies to situations where the first concept does not apply (already accounted for).

... in theory, not accounting for the players, this phenomena has a simple explanation. If the lines are equal that means that
* the NL home team expected runs are a few hundreths of a run lower than the AL home team expected runs
and
* the NL away team expected runs are a few hundreths of a run higher than the AL away team expected runs

Both these factors contribute to the phenomena observed.

I must admit I overestimated the amount of thought Mathdotcom has already put into this before asking the question. It turned out Mathdotcom was not struggling with an advanced concept that the line makers already utilize but yet to approach the understanding of its existence.

[mathdotcom]

You're right, I was making the mistake of not holding the total constant even though when I set up the problem I said I was. Given the general low quality of most threads in the think tank, I do not feel guilty making a thread on the spot without much thought.

[Justin7]

This thread has degenerated into dick waving. Moving it out of HTT.

[poochiecollins]

This thread has degenerated into dick waving. Moving it out of HTT.

[HoulihansTX]

This thread has degenerated into dick waving. Moving it out of HTT.

Is that what we call it now?