Crunching some over/under numbers....

[jellobiafra]

from last season. I wanted to see what the longest streak any team had through the year. I checked about 3/4 of the teams and the longest under streak I found was Tampa at 10. Florida actually had a ridiculous 15 game over streak.

I'm trying to see if a betting system - a chase system - can be worked around the numbers. It's clear that you would have to cap the bets - probably an A-B-C thing like the Morrison system. And you would have to accept that there are going to be losses. You'd have to be willing to lose money every once in a while and walk away. Like I said, the Marlins had a 15 game over streak last year. You just can't keep chasing that for 10 bets.

This isn't going to be something that you'll go 75-1 on through the season. I have a feeling I can work something out that could be profitable based on last year's results. I'm thinking if you started the chase on over bets at about 5 straight unders, you should have a good chance of hitting before you get to 9 - by the 3rd or C bet. I think you would have to start the under chase a little later because it appears that there are longer over streaks than unders for some reason.

I don't know how much action this would give you in a year. I did a couple teams and ended up with 2-3 sessions for each in a season. I'm also not sure if the under bet sessions shouldn't just be tossed out completely because - like I said - at first blush it appears that over streaks last a little longer and the win percentage might not justify playing it at all. I think there might be potential in an over chase system though.

You wouln't be buying any points like the NBA so you aren't paying the extra juice on losses. Lets say chasing $10 -- $11>10; $23>21; $48>44..... possible loss of ~$82. So I need to find a number that shows about 90% success rate to be profitable. I don't know. That's a tall order now that I'm looking at the numbers.

I might end up with something that's just a single bet. Like if I can find a high probability that a team won't go over a certain number of unders or overs in a row (say 65% or so probability) that might be the way to go. Just find that number and play it all year. Not as exciting and not as much action, but again there might be potential.

I'm going to do some more research on this over the next several days. If anyone is interested I'll follow up in this thread. I'll actually probably follow up whether anybody is interested or not.

[Slanina]

Like you said just do over bets. You get to the 4th and 5th pitcher in a rotation you can't expect both of them to pitch good ballgames consistently. One of the two is bound to have a bad outing. So maybe use some filters. Only start the betting if its the 3rd/4th/5th man in rotation due to pitch. And maybe factor in what stadium they are at. If they have 5 straight unders and have a 3 game series at San diego you may want to stay clear of that series. Let alone the padres offense.

[Slanina]

Ranked in order from best to worst hitter friendly ballparks. I left out the mid teams, though. These are based on 2008 stats.


1) Rangers
2) Diamondbacks
3) Rays
4) White sox
5) Red sox
6) Tigers
7) Reds
8) Cubs
9) Braves
10) Orioles


21) Mets
22) Brewers
23) Cardinals
24) Mariners
25) Royals
26) A's
27) Pirates
28) Twins
29) Dodgers
30) Padres

Although, in 2007 the rangers ranked 15th and the red sox ranked 21st. So for 2009 these could be completely different rankings. The only ballpark that consistenly stayed in the top 10 since 2001 is Coors field.

[jellobiafra]

What I'm going to do is go through each team for the past few years and chart every O/U streak. I guess I should note starting pitchers and ballpark whenever an under streak went to an over (maybe vise versa as well). I'll post a synopsis in a few days (might take a week). That should give us a good starting point to discuss filters.

[jellobiafra]

Ok. I've got an update and at first glance it looks to me to be pretty promising.

I decided to start with Atlanta. I printed out the results of their games for the 1999-2008 seasons -- 10 years worth. I've gone through '99 through '03. This is more time consuming than I bargained for but I think it's going to be worth it. I'm taking a break and will finish the 10 year cycle for the Braves tonight or tomorrow and follow up. I just don't know if it'll require doing any more teams. I certainly don't think it'll require doing every team. I'm also not inclined to involve any filters here. First off, the pitchers and ballparks are all factored into the total spread already. Secondly, if these numbers continue to hold up this is going to be a very simple system.

Ok. Where to start. From '99 to '03 the Braves had 384 overs and 381 unders. There were 45 pushes. They had one year (2001) with only 70 overs and 83 unders; and one year (2003) with 90 overs and only 64 unders. The point I'm trying to make here is that I think this is a pretty balanced representation of the league as a whole. Their longest over streak was 9 (twice) and their longest under streak was 8.

Here is how I've broken down the numbers. I have chosen to track every over and under streak of 2 or more games. Over these 5 season the Braves had 93 over streaks of 2 or more games and 96 under streaks of 2 or more games. Here is how the over streaks break down:

2 games -- 47/93 (50.53%)
3 games -- 24/93 (25.8%)
4 games -- 9/93 (9.67%)
5 games -- 2/93 (2.15%)
6 games -- 7/93 (7.52%)
7 games -- 2/93 (2.15%)
8 games -- none
9 games -- 2/93 (2.15%)



Here is how the unders break down:

2 games -- 52/96 (54.16%)
3 games -- 16/96 (16.67%)
4 games -- 16/96 (16.67%)
5 games -- 4/96 (4.17%)
6 games -- 3/96 (3.13%)
7 games -- 4/96 (4.17%)
8 games -- 1/96 (1.04%)



The thing that jumps off that page at me is the percentage of streaks that end before they get to the 4th game. 76.34% of all Braves over streaks between the years 1999-2003 ended after 3 games. 70.83% of all their under streaks ended after 3 games and an incredible 87.5% ended after 4 games. I'm going to work this through the next 5 seasons ('04-'08), but I don't anticipate much differentiation. I think over 800 games is a pretty decent sample size as it stands. Finishing out the Braves will give me over 1600 games. Does anyone think it's absolutlely essential to work through the other 31 teams? Because this took me about 2 hours to do and I don't think I have it in me to go through the entire league.

I think as it stands right now I'm pretty comfortable betting on any over streak to end after 3 games and any under streak to end after 3 or maybe 4 to be safe. Can anyone make a counter argument. I really would like to hash this out and see if I -- we -- could be on to something going into the baseball season. I'm not the most statistically inclined person and I know there are people much more well versed than me on this things posting here. Tell me if I'm missing something obvious, or even not so obvious. I'm open to any suggestions.

[EaglesPhan36]

I know the Pirates had a huge OVER advantage on the road. They only had two road trips all last season where you would have lost a small amount of money if you bet the over in every game.

To me, the key is identifying POS bullpens as early as possible. You can take large advantage of that situation when the combo of bad starters are going on a given day.

[jellobiafra]

OK, for arguments sake I just looked at the Pirates for last season. If you had bet the under after every over streak of 3 games you would have gone 6-4. If you had bet the under after every over streak of 2 AND 3 games you would have gone 20-14. It's not as good going with over bets after under streaks. If you'd bet over after every 3 game streak you'd have gone 4-4 and after every 2 AND 3 game streak you'd have gone 11-12. Not crushing by any means and if you would've played them both, betting the streaks to end after 2 games and if that loses following it with a bet on the next game you would have gone a combined 31-26 on the Pirates. Only 54% and not quite good enough, although the streaks are clearly still mostly 2 or 3 games. I think you'd have to bet after the 2nd and 3rd game in order to gain any benefit from this though. If over half the streaks are ending after 2 games, you probably can't wait around for the 3rd game to bet.

We know from above that the the Braves over streaks during that span had a 76% chance of ending after 2 or 3 games. What I'm wondering from the more statistically inclined is what kind of winning percentage would you get if you had bet every one of those streaks after 2 games (50.53%) and then the ones that lost you bet on the next game as well (25.8%). Because now that I'm looking again I'm wondering if this is really going to be of any advantage. On that 2nd bet (betting after 3 game streaks) I think you take out the 50.53% (chance the streak ends after 2 games) from the equation and now are dealing with 25.8% (chance after 3 games) chance of winning out of the 49.47% (100% minus 50.53%) remaining. Still just better than a 50% proposition and not worth any more of my time.


So I think all I'm going to come up with here is that it is a better than a 70% proposition that a streak will end after either 2 or 3 games. So maybe you can use this knowledge to find good prospects to play. You can use filters on that after 2nd game bet or the after 3rd game bet - filters like pitchers, parks, bullpens..... basic handicapping tools, knowing that there is a high likelihood that the streak is going to flip either after that 2nd or 3rd game. Look at the two games together and figure out if one of the two is more likely to go the other way. If you can handicap which one it's going to be then you might be ahead of the game. Just remember that betting either one individually is still basically a 50-50 prop. If you can look at them both and figure an advantage for the streak breaking on one game or the following game though - either or - you might be able to find a winning strategy out of this.

[SOXROCK]

over/under system

have been looking at teams against l or r handed pitchers. Just started with Baltimore and found out that last year the O's scored under 83% of the time against right handed pitching didn't know if anyone has looked at this type of angle for an A-B-C style betting. Lets say we wait for 2 right handed pitchers to go over (in a row) and we bet the next 3 for an under. Still working on it lots of data to input.

[jellobiafra]

Thanks Sox. You might have something. I gave up on this....perhaps too early. The 55% offers the slimmest of fragment of an edge and I'm gonna use it if the pitchers,park, lineups, weather etc warrant. The 70% after 2 OR 3 games brings you right back to coin flip odds. I like where you are taking this though. Keep us posted please.