﻿ Starting opposing pitchers (Part 1) | The Hardball Times

# Starting opposing pitchers (Part 1)

By their nature, wins are fickle. Yes, a correlation does exist between wins and ERA, but I’ve seen great pitchers finish seasons with few wins and mediocre pitchers rack up wins like bad jokes in a Will Ferrell movie. How else could Joe Saunders finish 2009 with 16 wins and a 4.60 ERA while Randy Wolf gets only 11 wins out of his 3.23 ERA?

 Pitchers with better ERAs will get more wins per game than pitchers with worse ERAs, though since the r = .48, ERA determines only roughly half of whether a pitcher gets a win (Click to enlarge).

There are other factors besides a pitcher’s skill level such as innings pitched per start, bullpen strength, and offensive runs per game that influence how often a pitcher will get a win; however luck still plays a large role in the way wins are distributed. Therefore it is smart to not draft for wins since luck is unpredictable.

There are times, though, in daily Head-to-Head leagues when you need to harness the power of wins to become victorious in a particular week. Such times typically occur on Saturday nights when you are trailing by one in the wins category and are setting your lineup for Sunday. Despite the unpredictability of wins, there is a strategy you can use to increase the chance you will earn at least one win and that is by starting opposing pitchers.

Just to make it clear, opposing pitchers are two starting pitchers who are pitching against each other in the same game. So for example in the first Yankees-Red Sox game of 2010, the opposing pitchers will most likely be C.C. Sabathia and Josh Beckett. The advantage of starting opposing pitchers in getting a win might not reveal itself right away so allow me to dazzle you with some math that will make clear the advantage.

### The mathematics

Starting pitchers as a whole could have earned 2,430 wins in 2009 since there are that same 2,430 total games played in a season and one win is awarded per game. Instead of getting 2,430 wins though, starters earned only 1,706 wins, meaning 724 wins were lost to relievers. What this means is that 70 percent of the time, the win will go to one of the starting pitchers while there is a 30 percent chance a reliever gets it. This 70-30 ratio is fairly stable from year to year. With a 70 percent chance of the starters getting the win, each starter then has a 35 percent chance of getting the win assuming each pitcher is league average.

From a fantasy perspective, starting opposing pitchers offers a unique opportunity to garner wins at a higher rate. When starting both starting pitchers, you have a 70 percent chance of earning a win for your fantasy team. When starting two random pitchers however, you only have a 45.5 percent chance*. Why then would you not always start opposing pitchers if it gives you a extra 25 percent chance to get a win compared to starting two random pitchers?

*For the less mathematically savvy among us, I got to 45.5 percent by first finding the chance both pitchers get the win (.35 * .35 = 12.25%) and then finding the chance both pitchers do not get the win (.65 * .65 = 42.25%). The chance then, that one pitcher gets the win is 100 minus the sum of those percents which is 100 – (12.25 + 42.25) = 45.5 percent.

The answer is that your win potential is capped at one win with opposing pitchers, but with random pitchers there is the chance you earn two wins, a 12.25 percent chance to be exact. Therefore the two-win potential reward of random pitchers balances the decreased chance of getting one win and also the increased chance of getting zero wins.

### Back to fantasy

It is time to take a step back and understand how opposing pitchers can be utilized in fantasy leagues in a practical sense. It is important to note that, over the long run, starting opposing pitchers will not necessarily result in more wins because of the two-win potential of two random pitchers. Starting opposing pitchers can come in handy though in the scenario I detailed towards the beginning of the article, and that is in a Head-to-Head league with daily roster updates.

If all you need is one win and there is a game in which both pitchers in that game are obtainable, theoretically you would be increasing you odds of getting that win by starting both of those pitchers as opposed to two starters in different games. However what’s true in theory is not always true in practice and since all teams, pitchers, offenses, and bullpens are not created equal, the question becomes how much of a decrease in pitcher skill should you accept in order to start two opposing pitchers?

It should be obvious that even if Roy Halladay and Cliff Lee are not opposing each other you would still want to start them since they earn wins at above the average 35 percent rate; however, a point must exist where the difference in pitcher skill is overshadowed by the advantage starting opposing pitchers offers. I understand that this pursuit is limited in its practicality since it can only be used in a certain league type in a somewhat rare situation, but for me, it is pursuits like this that make fantasy baseball so enjoyable.

Print This Post
Guest
Bill

However, starting 2 non-opposing SP gives you the chance at 2 wins and the same net wins

12.25 x 2 = 24.5 + 45.5 = 70

*For the less mathematically savvy among us, I got to 45.5 percent by first finding the chance both pitchers get the win (.35 * .35 = 12.25%) and then finding the chance both pitchers do not get the win (.65 * .65 = 42.25%). The chance then, that one pitcher gets the win is 100 minus the sum of those percents which is 100 – (12.25 + 42.25) = 45.5 percent.

Guest
Ian

Bill, I think his point was that you have a greater chance of getting at least one win, even though your expected value of wins is the same.

Guest
Seth
Yes, starting opposing pitchers gives you a greater chance of getting AT LEAST one win, but the article’s claim that “it gives you a extra 25 percent chance to get a win compared to starting two random pitchers” is incorrect.  You still get “a win” even if you get two wins, and a fantasy owner who needs one win would certainly not be upset if he ended up getting two wins.  The percentage chance of getting at least one win from non-opposing pitchers is the sum of the chance of getting exactly two wins (12.25%) and the chance of getting… Read more »
Guest
Derek Ambrosino
This is kind of like the sabermetric discussion of when to bunt, but a little different. Better chance to score a single run w/ a runner on 2nd, 1 out. Higher overall run expectancy w/ runner on 1st, 0 out. Therefore, in late innings of close games it makes sense to bunt, but not early in games or when you are behind multiple runs. The corollary here would be that this is probably a bad idea to try on Tuesday, but could certainly make sense on Saturday. What is an annoying confluence of skeds is when your two best pitchers… Read more »
Guest
Matt Chelius

Interesting stuff.  The probability of getting losses (whether at least one, two, or none) should be factored in this discussion as well.

Does the probability of an SP loss follow the same 30/70 split?  57.75% at least one?  12.25% two?

Guest
Paul Singman
Seth, Good catch with the percentages. I shouldn’t have been looking for the chance non-opposing pitchers get one win but rather the chance they get at least one win as you pointed out, and that chance is 57.75%. There is, however, another mistake both of us made in interpreting the difference between a 70% and 57.75% win rate for the two types of pitchers. John Burnson pointed to my attention that the 12.25 difference in percentage points between 70 and 57.75 does not mean opposing pitchers are 12.25% more likely to get at least one win. Instead they are [(70-57.75)/57.75]… Read more »
Guest
Paul SIngman

Good question Matt. For opposing pitchers losses are the exact same percentages as it was with wins. There is a 30% chance you get zero losses, a 70% chance you get at least one, and a 0% chance you get two losses.

With non-opposing pitchers the percentages are also the same: A 12.25% chance you get two losses, a 57.75% chance of getting at least one loss and a 42.25% chance of getting 0 losses.