09 April 2014

The Impact of Losing Hardy and Machado

The Orioles were disappointing after their first seven games. They lost a series against both the Tigers and Red Sox and lost the first game of the Yankees series. The only starting pitcher that has been effective is Chris Tillman and the offense had mostly been ineffective. All in all, a 14 to 5 win against the Yankees in the eighth game was just what fans wanted to see.

The eighth game showed it is simply too early to panic about these early results. It’s a long season and ups and downs should be expected. Players that have underachieved so far this season will eventually get going.

What it isn’t too early to be worried about is the fact that J.J Hardy and Manny Machado have been injured. Hardy has suffered from back spasms while Manny has been unable to recover from a knee injury. Both players are expected back eventually but until they return players like Ryan Flaherty, Steve Lombardozzi and Jonathan Schoop are forced to take their place. I used two methods to quantify the impact these injuries have had on the 2014 Orioles.

The first method is by calculating the current rosters individual stats from 2011-2013 against lefties and righties and plugging the numbers into a lineup simulator. The second method is by determining each player’s wRC+ from 2011 to 2013 against lefties and righties and seeing how many runs each player would be expected to score. For each method I compared how many runs a lineup with Hardy and Machado would score to a lineup with just Hardy to a lineup with neither.

The first thing I did with the lineup simulator was test its accuracy in past years. I took the Orioles that played the most in 2011, 2012 and 2013 and put their data into the simulator to see whether the projections were close to reality. It turned out that the numbers were reasonably close. They projected the Orioles to score 667 runs in 2011, 698 in 2012 and 707 in 2013 while they actually scored 708 in 2011, 712 in 2012 and 745 in 2013 which is more or less accurate. One reason why the projections were low is probably because the Orioles play in a hitter friendly stadium and this isn’t factored into the projections. The projections were off by roughly 4.5% each year.

When I plugged in each starters statistics from 2011-2013, the lineup simulator projected that this team would score 711 runs against righties with Machado and Hardy in the lineup, 690 runs with Hardy but without Machado in the lineup and 678 with neither Machado or Hardy in the lineup. The lineup simulator also projected that the club would score 695 runs against lefties with Machado and Hardy in the lineup, 649 with only Hardy and 610 with neither. Multiplying that by 104.5% means that the club should be expected to score 711 runs against righties and 695 against lefties with both Machado and Hardy in the lineup, 690 against righties and 649 against lefties with only Hardy in the lineup and 678 against righties and 610 against lefties with neither in the lineup.

Next I took the wRC+ for the Orioles that played the most in 2011, 2012 and 2013 and determined how many runs they would score. It projected that the Orioles would score 694 runs in 2011, 692 runs in 2012 and 700 in 2013. The stat wRC+ attempts to determine how many runs a team would score in the average ballpark. Since the Orioles play in a hitter friendly ballpark it makes sense that wRC+ would underestimate the amount of runs the Orioles score. This method was off by roughly 3.8% per year.

The average team scored 700 runs in 2013. If you divide that by 9 then you’d expect the average player to be worth roughly 77.77 runs. A basic way to determine how many runs a player will be worth is to simply multiply the runs scored by an average player by a given players wRC+/100.

This stat suggests that this team would score 706 runs with Machado and Hardy in the lineup, 691 runs with Hardy in the lineup but not Machado and 675 runs with neither against righties. Against lefties, the team would be expected to score 775 runs with both in the lineup, 723 runs with Hardy but not Machado and 695 runs with neither. Multiplying that by 3.8% means the team should be expected to score 706 runs against righties and 775 against lefties with both in the lineup, 691 against righties and 723 against lefties with only Hardy in the lineup and 675 against righties and 695 against lefties with neither Hardy nor Machado in the lineup.

I think the reason why the lineup simulator expects the Orioles to score fewer runs against lefties than wRC+ is because it takes into account that the Orioles have a number of bad offensive players against lefties. These players are projected to hit into a lot of outs and reduce the number of chances that the better players receive.

From 2011 to 2013, the Orioles had roughly 72% of their plate appearances against right handed pitching. According to the simulation they should be expected to score 715 runs with both Manny and Hardy in the lineup, 687 runs with only Hardy in the lineup and 667 runs with neither in the lineup or that replacing Manny and Hardy with Flaherty and Lombardozzi should cost the team 48  runs. According to wRC+, the Orioles should be expected to score 734 runs with Hardy and Machado in the lineup, 708 with only Hardy in the lineup and 688 with neither in the lineup and that replacing Manny and Hardy with Flaherty and Lombardozzi should cost the team 46 runs. Both methods return roughly the same results.

Now that it's been determined that the team loses roughly 47 runs offensively due to the replacement players it is necessary to determine how many runs this team would lose due to defense. Machado and Hardy are excellent defenders at their positions while Flaherty isn’t a shortstop and Lombardozzi and Schoop are decent defenders at best. Even if Flaherty, Lombardozzi and Schoop can be average defensively, which is highly unlikely, according to UZR Machado and Hardy were worth 47 runs above average defensively last year.

With the limited amount of defensive data available it is hard to do anything but speculate. Flaherty has played a limited number of innings at shortstop. He's above average at second base but who knows whether he can play an adequate shortstop. Schoop has been in the major leagues for a limited amount of time. He doesn't have a reputation for being an above average defensive player and it isn't clear whether he can provide average defense. Lombardozzi has played 774 innings at second base before this year. His defense is considered to be average in a limited sample. These players could be decent defensively or they could be very bad. 

It wouldn’t be unreasonable for Machado and Hardy to be worth 45 runs more offensively and 45 runs more defensively or about 90 runs better total. The standard ten runs equals a win rule indicates that Machado and Hardy could be worth 9 more wins than the Flaherty, Lombardozzi, Schoop trio. If defense is worth less than UZR claims then this number would be smaller and if Flaherty, Lombardozzi and Schoop are worse than average defensively then this number could be larger. If one expected this team to win 88 games with Machado and Hardy then this team should only win 79 games without Machado and Hardy. Their injuries are a huge blow to the Orioles playoff chances.

Yet it also provides hope for the rest of the season. The reason why the Orioles have looked bad at times during the first eight games is because without Hardy and Machado this team is much weaker. A hard schedule to start the year doesn't help. Hardy and Machado are expected back in the near future and the Orioles should show massive improvement when they return.

If this team can stay around .500 without them then when they return we should expect to see a much more potent offense and defense.


Unknown said...

Great analysis to help quantify the situation. Heartening, too, for the short term, since that 9-run difference over a full season (88-79 wins) works out to only .05 wins per game: a difference of only 1 win over, for instance, the first 20 games.

Anonymous said...

the statistics work best with large sample size, but not for small sample size, say 7 or 8 games, so you are kind of saying the obvious, although it's fun to do the calculations.

Jon Shepherd said...

I think the take home is that the variability observed in a 7 or 8 game data set is quite wide, but that the presence of Machado and Hardy would result in a higher probability of run production. These numbers should not be seen as specific values, but points to which the data spread in anchored by.

In other words, I think it is more than just fun to do this. There are real world applications in projected performance. What you do not want to do is take that performance over 7 or 8 days and extrapolate it. Data application from the deep to the shallow is more dependable than from the shallow to the deep.

Matt Perez said...

Thanks Nowick.

The other thing is that a 79 win team isn't going to look as good as an 88 win team. If you're expecting 88 win talent then you may be dismayed to only see 79 win talent. This is the case over a seven game sample or over a 162 game sample. You'll notice the difference in ability even if it doesn't show up in the standings.

Even if they're looking bad at points now they'll look much better when Machado and Hardy come back. Fully agree that provided they come back soon it should have a relatively small impact on the final result.

It's just another reason not to panic. Provided that they do return sooner rather than later.

Unknown said...

What about the runs they will take away on defense. Machado is worth as much on D as O, and Hardy is money..

Jon Shepherd said...

Much of the article is about defense.