MLB Winning Streak Frequency, Sweep Difficulty & More | Mailbag

posted in MLB

Mailbag time! We’ve received several questions recently about how often MLB teams win in certain situations, so we thought it would be useful to compile them all here in one place. Enjoy, and please feel free to ask any questions you might have in the comment section!

How often does the money line favorite win, compared to the money line underdog? –Javier R.

Over the past 13,000 or so game, the money line favorite has won 57.1% of the time. However, that obviously varies based on the actual line. If you bin the favorites according to their money lines, here is what you see:

Fave Line GroupGamesFave Win%
-200 to -415104868.3%
-175 to -19998861.1%
-160 to -174126362.3%
-150 to -159105459.7%
-140 to -149150959.0%
-130 to -139183456.7%
-120 to -129181555.3%
-110 to -119208752.1%
-103 to -109179549.9%

As expected, bigger favorites win more often.

Can you find what the W-L record of a team is during interleague play when the starting pitcher has a win% greater than 75%? –Tee T.

If I just take your question at face value, and don’t add in any requirements for how many games the starting pitcher has pitched in, then teams that fit the bill have won at a 54.3% rate since 2007. If I require the pitcher to have at least 3 wins, then the new win% is 56.3%.

In regular (non-interleague) games, teams starting a pitcher with a 75%+ win percentage and at least 3 wins have a 52.6% win rate. So in this super-simple analysis, the pitcher record seems slightly more important for predicting interleague games.

How often do MLB teams win 2 or more games in a row? I would also like to know often teams lose 2 or more in a row. –Walter M.

There are a few different ways to define this. First, let’s assume you want to know the winning percentage of a team on a 1-game winning streak. That is, if a team has a chance to create a 2-game winning streak, how often do they succeed?

It turns out that teams on a 1-game winning streak have a win% of 50.6% (over about 6500 games). So they fare ever so slightly better than an average team (which by definition wins 50% of the time). Looking at losing streaks, a team that has lost 1 in a row ends up losing their next game 50.2% of the time.

Another way to define this would be to look at every loss, and see if the team won their next two games. In other words, out of all the chances that a team had to begin a 2-game win streak, how many times did they actually do it? The answer to that question is 24.7%. That’s actually slightly less than you would expect by random chance (25%). For losing streaks, the same value is 24.4%, meaning teams follow up a win with 2 straight losses slightly less than you’d expect by chance (25%).

These results make sense — the fact that a team won their previous game means that they are slightly more likely to be an above-average team (and the reverse is true for teams that lost their previous game).

How often does a team trying to complete a 2,3, or 4 game sweep win? This could also be broken up into home/away, favorite/dog, etc. –John C.

As you might have guessed from the answer to the previous question, teams trying to complete a sweep generally win at a bit over a 50% rate (across all scenarios). This table breaks down how often a team that is trying to complete a sweep wins the final game, based on a combination of series length and the other attributes you asked about.

Frequency Of Completing A Sweep, Based On Series Length & Other Factors (2007 to mid-June 2012)

Series LengthAll scenariosHomeAwayFavoriteUnderdog
2 Games51.2%56.5%46.2%54.5%46.5%
3 Games50.9%55.8%44.2%57.6%40.4%
4 Games55.3%56.2%53.6%59.0%48.4%

This shows, for example, that road teams trying to complete a 3-game sweep only managed to win the third game 44.2% of the time. Or that Vegas favorites trying to complete a 4-game sweep did so 59% of the time. Keep in mind that the bin sizes here range anywhere from about 50 games to a few hundred, so these values are not going to be very precise — over the long run, some of these will probably change.

How often does the home team win at least one game in a 2 or 3 game series, or one of the first 3 in a 4 game series?  Also how about at least 2 games or more in 2,3,or 4 game home series? –Brian M.

From 2007 through early June 2012, the home team won at least 1 game of a 2 game series 76.4% of the time (compared to the 75% you would expect if every game was a coin flip). The home team won at least 1 game of a 3 game series 90.4% of the time (compared to 87.5% in a coin flip league). The home team won at least 1 of the first 3 games of a 4 game series 89.6% of the time (again compared to a random value of 87.5%). And the home team won at least 1 game of a 4 game series 94.6% of the time (compared to 93.8% in a totally random world). The same values for away teams were 72.8% (win at least 1 of 2 games), 83.0% (win at least 1 of 3 games), 82.8% (win 1 of first 3 in a 4 game series), and 90.0% (win at least 1 of 4 games).

Over the same time period, the home team swept a 2 game series 27.8% of the time, won at least 2 games of a 3 game series 56.8% of the time, and won at least 2 games of a 4 game series 74.8% of the time. Those same values for road teams were 23.8%, 43.5%, and 61.5%.

  • happy

    Is there software available to buy that would be able to analyze individual team results for this season?

  • David Hess

    happy — Software that analyzes MLB series sweep results? No, not that I’m aware of.

  • streaker

    how does % of winning move after 3 game winning or losing streaks

  • David Hess

    Over the past 2768 times that a team has entered a game on an EXACTLY 3-game winning streak (i.e. lose, win, win, win), their record is 1436-1331 (51.9%).

    Over the past 2671 times that a team has entered a game on an EXACTLY 3-game losing streak (i.e. win, lose, lose, lose), their record is 1294-1377 (48.4%).

    And it’s probably worth mentioning that given the money lines on those games, a blind strategy of betting on or against either of these trends would have been a losing strategy.

  • WildThing

    I would like to know if it would be possible to find out the following:

    In a 3 game series where the teams have each won a game, how often does the Home team win the 3rd game to win the series?

  • David Hess

    Thanks for the question. We don’t have that info handy, but if we do another blog post on streak/sweep data, we’ll see if we can dig up the answer.

  • Jim

    how often do home favorites win in MLB?

  • David Hess

    1) I am going to rephrase the question as “how often HAVE home favorites WON in MLB?” … since past performance does not necessarily mean the same thing will keep happening, and the rate of home favorites winning changes over time.

    2) That depends on what time frame you look at.

    Here are the win% for home favorites by year over the past 10 years:

    2006 58.6%
    2007 58.7%
    2009 59.2%
    2009 59.5%
    2010 60.1%
    2011 58.1%
    2012 58.4%
    2013 58.6%
    2014 57.4%
    2015 57.7% (through May 10th)

  • Jim

    ok first off thanks for the answer but even better questions what is the win percentage for home favorites between -120 and -150

  • Jim

    or even tougher home favorites between -120 and -150 and won their last game

  • David Hess

    Jim, I used the Sports Database Query Language site to get the info in my last answer. Here’s a link to the answer to this question. If you’re curious about more breakdowns, you can edit that query to get other results:

  • Jim

    thanks for that info that helps a ton and at 60.4 with avg decimal odds of 1.73 it looks like a hella of a system

  • David Hess

    Jim, 2 things:

    1) I don’t see 60.4%. I see 54.3% for my linked query: … so you might want to make sure you have the parameters correct.

    2) I wouldn’t read too much into a small edge over a few hundred games. Even when flipping a coin 500 times, it’s not too unusual to see heads or tails hit 55% of the time.

  • Jim

    I modified it and got it to 60.4% over 9000 games. the thing I seen a lot of in playing it is many many small losses and a few big hits but not many big losses, if that makes sense

  • David Hess

    Be careful! Continually tweaking the parameters to improve the performance of your training sample is a recipe for overfitting. When making a model, it’s important to test it on data that was *not* used to create it in the first place. In your case, with the SDQL site, it’s tough to figure out how to make a good training and testing set. But a couple options would be:

    —-Create the model using odd numbered months, by adding this to the end of your queries: ” and (month=5 or month=7 or month=9)” … Then when you have a model you like, test it on even numbered months.

    —-Create a model using odd numbered years, and test on even numbered years (or vice versa)

    —-Create a model using, say, 2009-2013, and then test it on 2014-2015.

    That said, good luck with anything you find!

  • Jim

    The trick is in the win odds avg you have to be smart on which games you are betting. my bet odds avg is like 1.80 and my win odds average is like 1.72

  • Jim
  • Aaron Clutch

    I presume that when someone asks a question such as the one above you must look through a database with all the information to help you figure out these answers as opposed to counting games on, what is this database and is it publicly available to someone like me?

  • Aaron Clutch

    Just scrolled through some more comments and found it, sorry about that.