Anyone have MLB team total push rates?

[IrishTim]

Would be willing to trade for it or dump all my SBR points (220) on you if you're into those.

[username474]

I will post them tonight. I have back to 06.

[IrishTim]

That would be immensely helpful. Thanks username.

[Toples]

these are results from 2004 including playoffs
4,8% of games are push
hope this helps

gamespushesteam110650Athletics112547Astros110056Blue Jays111854White Sox114163Cardinals110053Nationals110154Rangers111458Twins110142Mariners111657Rockies110659Cubs111655Tigers109957Pirates111251Indians113143Angels110852Padres111643Rays110159Royals113353Phillies111056Mets111053Braves110156Giants110950Diamondbacks114053Yankees110456Brewers110257Orioles114762Red Sox110157Reds112551Dodgers109955Marlins

[Toples]

[Toples]

https://docs.google.com/leaf?id=0B_XyknGRbyXmYmVkNDM1YWMtNDQ4OC00NjY2LTljNjUtMzYwNTBhMTg4Yzc1&sort=name&layout=list&num=50

download image

[Peregrine Stoop]

Hrm... I think it's worth more than 220 SBR points

[Pokerjoe]

Actually, I don't think toples has answered the question at all. Most likely Tim is looking for push rates on the 3, the 4 and the 5, not for teams themselves.

[IrishTim]

Thanks for the work, Toples, but Pokerjoe is right. I may have been unclear, but I'm trying to determine how much the "3", "4", and "5" are generally worth.

[Pokerjoe]

Yeah, actually, Toples, I didn't mean to disparage your effort. Anyone willing to share is helping the forum in the right direction. Good luck to you.

Irish, I think you'll find that Poisson provides a pretty good estimate for push frequencies in this case.

[TomG]

5Dimes puts up Team Totals and Alternate Team Totals for big MLB games (usually the Sunday night game). You can infer the push rates from this.

[Data]

Pokerjoe, Poisson is not going to help.

TomG, 5Dimes puts up TT for every game, same as Greek/BetJamaica.

IrishTim, try to be more specific with your questions, it is STILL UNCLEAR what you are asking for. Regardless, use Weibull as the first approximation for making your own TT, if you want to be more precise, adjust for the bottom of the ninth and beyond. No matter what you are currently looking for, that is the stuff that will give you what you really need.

[IrishTim]

Sorry Data - what I'm after is the value in cents or push rate in percentage of each team total. For example, what would be better: UNDER 4 (-125) or UNDER 3.5 (+100)? Since they're so low, I'm sure each one pushes a fair amount, my guess would be 10-15% or so but I have no team total database to check it against. Sure, I could use the drop downs and derive it from there, but I prefer to use that as a back-up to my own push rates, etc. Not that big a deal, I figured that with such low liquidity in the team total markets, most of the think tankers wouldn't bother. But a small-time grinder like me could use some team-total data. Anyway, if people have information that might be helpful you can PM me if you'd rather not have it on the open forum and I'll try to return the favor any way I can.

[Data]

IrishTim, we do things differently. My take on this is once you solve the bigger problem, every derivative market becomes transparent and you realize that HISTORICAL DATA for those market IS IRRELEVANT. If I wanted to know "what would be better: UNDER 4 (-125) or UNDER 3.5 (+100)" I would need to know ml and total. Your question as it is has no correct answer.

[sharpcat]

Accepting what a book suggests as the value of team total push rates would be a huge error I would assume that most here would agree that deriving your own push frequencies would be much preferred if you are hoping to gain any type of an edge. I think what Irishtim is looking for here is to attempt to approximate the value of a half point on team totals.

[IrishTim]

Data, I was expecting that answer from you as it is consistent with what you told me the other week when we talked about something else relating to baseball and it's perfectly reasonable (more than reasonable, it's the "right answer"). However, Sharpcat knows what I'm after, imperfect as it may be.

[sharpcat]

Data, I was expecting that answer from you as it is consistent with what you told me the other week when we talked about something else relating to baseball and it's perfectly reasonable (more than reasonable, it's the "right answer"). However, Sharpcat knows what I'm after, imperfect as it may be.

Data is right considering that game totals play a huge part in team totals.

Lets rephrase the question

Does anybody have push frequencies specific to game totals for MLB team totals? I have a lot of points to share too.

[Pokerjoe]

http://www.baseballprospectus.com/article.php?articleid=342

Might be of use. But it uses the Poisson, which Data dislikes here. Oh, well. Judge its value yourself, Irish.

[Pokerjoe]

http://www.beyondtheboxscore.com/sto.../23/164417/484 is Weibull

[IrishTim]

Thanks Joe, I've read that first article multiple times and used their data even, but hadn't come across the second one. It's helpful enough to get me started. FTR, not a huge fan of Poisson in baseball either.

[Pokerjoe]

Well, poisson as Davenport uses it, it's crudely adjusted, using the average of 3 poissons as you know, but it's been my observation that it's accurate within a few tenths of a percentage. But it is relatively crude, yes; especially, I've found, for the 3.

[Pokerjoe]

http://arxiv.org/PS_cache/math/pdf/0509/0509698v4.pdf

Because you gave me the points, LOL, here's the source paper. But I never have put in the effort to wade through it myself. I'd have to be in a good mood and have nothing to do for the day if I was going to start deciphering such papers. Enjoy, if it's your cup of tea.

[danger]

I believe this is the data you want. These are team total results
for 21,570 games. For example, the away team scored 3 runs in 2929 games
out of 21,570.

away team total, 21570 games
0 1185,1 2141,2 2642,3 2929,4 2726,5 2331,6 1976,7 1617,8 1207,9 917,
10 664,11 442,12 296,13 178,14 125,15 83,16 52,17 21,18 15,19 8,
20 3,21 1,22 6,23 1,24 2,26 1,30 1

home team total, 21570 games
0 927,1 1815,2 2513,3 2966,4 2897,5 2580,6 2209,7 1653,8 1236,9 885,
10 635,11 476,12 296,13 191,14 126,15 60,16 42,17 24,18 20,19 7,
20 8,21 2,23 2

[Pokerjoe]

Danger, that won't work. That data isn't adjusted by expectation. A team's likelihood of scoring 3 runs in all games isn't the same as it's likelihood of scoring 3 runs in games in which it is expected to score 3 runs.

[Pokerjoe]

You'll actually be okay (within a few tenths of a percent, which is not good enough for betting derivatives of course), for on-the-fly comparing lines at different numbers, going with

the 3 lands 16.5%, the 4 lands 14%, the 5 lands 12% and the 6 lands 11%

[IrishTim]

Correct, I already had that data. I don't have any database with a book's team total prices which would be required to determine push rates.

[danger]

MLB home teams
3: 13.0% (12/92)
4: 13.8% (169/1224)
5: 12.2% (81/666)
6: 11% 4/35

MLB away teams
3: 16.4% (44/268)
4: 12.6% (181/1440)
5: 11.7% (41/350)
6: 10% (2/20)

[IrishTim]

Someone privately gave me their push rates and I did a bit of work on these of my own, but I appreciate the help. What source did you use for the data and what radius when calculating the push frequency? I find them all to be a bit low.

[danger]

I have a database of pinnacle lines going back several years and joined that on the retrosheet MLB results. I didn't use any radius, only exact team totals and ignored lines of 3.5, 4.5, etc... I haven't added 2010 MLB results yet, so this is just previous years.

SELECT SIGN(r.home_score - o.line1) team_total_over, count(*) FROM sports_results r, sports_odds o WHERE r.event_id = o.event_id AND r.league_code = 'MLB' AND r.period = 'G' AND o.line_type = 'thm' AND o.site = 'pinnacle' and o.line1 = 4 GROUP BY 1;
+-----------------+----------+
| team_total_over | count(*) |
+-----------------+----------+
|              -1 |      533 | 
|               0 |      169 | 
|               1 |      522 | 
+-----------------+----------+
3 rows in set (5.15 sec)

[Pokerjoe]

Why would you ignore lines of 3.5 and 4.5?
Why segregate home and away results?
Why assume an away 6 lands 10% of the time based on a sample of ... 20?

[danger]

> Why would you ignore lines of 3.5 and 4.5?
Because I could get the answer with a quick database query that way.

> Why segregate home and away results?
It may make a difference, and you can always combine them.

> Why assume an away 6 lands 10% of the time based on a sample of ... 20?
I am not assuming that, just posting the data, including the sample size so people can make their own inferences on margin of error.

[Peregrine Stoop]

it's dangerous to combine AL and NL into one overall MLB number