Note from David: There is much better study of protection out on the net, it can be found at: Grabiner Study

My hope is that you read mine first to see a specific case where protection is false, and then look at David Grabiner's work to see some real meat.

One untested part of baseball's conventional wisdom has been the theory of protection. Whenever a strong batter is in the lineup, baseball people wave their hands and talk about how another strong batter is needed in the lineup in order to "protect" the original batter. When cornered into explaining what protection means, proponents offer one of two versions:

Weak protection - This form is simple to understand. Unless there is somebody with a good bat behind the hitter, the pitcher doesn't want to risk an extrabase hit to the slugger at the plate and will walk the batter. A good batter behind the original hitter will prevent an overload of walks (or at least exploit those walks).

Strong protection - If player A is at the plate, then player A will see "better pitches." The reasoning is that the pitcher does not want to pitch to player B with a man on base, so he will try to get player A out. This involves throwing strikes rather than balls. Since player A is seeing more strikes, he or she will take advantage and will increase his or her offensive performance as measured by batting average and slugging percentage.

1996 was not a highlight in the history of the Giants. Outside of Barry Bonds and Matt Williams, they more or less filled their lineup with the Phoenix Firebirds. And to make a bad situation even worse, Matty spent half the season on the shelf. What was bad for us as die-hard fans is good for us as thinking fans. This year provided a data set to test the theory of protection. Roughly half the time Barry Bonds came to bat, he had the "protection" of Matt Williams. The rest of the time.... Well, you remember 1996.

One more stop before heading for the meat. It has been long known that raw batting average is a poor way of measuring performance. All it reveals is hits per at bat, and hides both walks and power. A hitter has two tasks when at the plate. The first is to reach base and the second is to advance baserunners. These two ideas are captured in On Base Average (OBA) and slugging percentage (SLG). A person's OBA is roughly walks plus hits divided by walks plus at-bats. In other words, when measuring how well a person gets on base, a walk is as good as a hit. The best way to advance baserunners is through power. Slugging average counts bases rather than hits. A double counts as two singles, a home run as four. As an example of these ideas, suppose I came to the plate five times, struck out twice, had two walks and a double. My average for the day would be .333 (1 hit in 3 at bats), my OBA would be .600 (reached 3 of 5 times) and I would slug .667 (two bases in 3 at bats). These numbers give a more complete picture of the player. Take a player who almost never walks and doesn't have a power stroke. He could still have a nice batting average, but it would be "empty." On the other hand, look at a Rob Deer type player. He might have a .200 batting average, but since he has tremendous power and knows how to take a walk, he helps his team much more than his batting average would indicate. Often OBA and SLG are simply added (OPS), and this total is taken as a method for evaluating the hitter. In most years you can count the players with an OPS over 1 without taking off your shoes.

Now on to the data. I freely admit right now that this study has flaws...

A) What I neglected to do is keep track of things besides hits and walks, so PA and OBA are in fact just (AB+BB) and (H+BB)/(AB+BB) instead of the correct values.

B) I made no attempt to correct for situations where Barry would have three "protected" at bats, and then one or two with a pinch hitter batting behind him.

C) A pair of dates is missing from my data... www.fastball.com was good, but not 100% complete

D) There was a stretch in late May where Barry couldn't hit for spit (was he injured?), and I think this may put a real dent into his protected stats.

E) Bonds is one of the best players in the last 30 years. Can anything based on him be used to talk about the average superstars of the world?

F) This was 600 times at the plate, I'd much rather have a data set with 600,000.

OK, one last thing. At the end of the season Barry decided that he was going to go for 40/40. As he had different goals than normal, it may have affected the outcome of the study. To correct for this I've made a data set called No40, which is Barry without protection, but including games only up to his 40th HR (which is about the time he decided to become a thief). [In the end it turns out that this made no difference in his stats]

Ok, now for the raw stats...

Barry Protected

AB	R	H	RBI	BB	SO	2B	3B	HR
279	59	83	59	51	36	17	1	20

Barry No Protection

AB	R	H	RBI	BB	SO	2B	3B	HR
231	59	73	67	97	38	10	2	22

Barry No40

AB	R	H	RBI	BB	SO	2B	3B	HR
205	49	64	56	81	37	9	2	20

Ok, now some rates...

BA:     .297 protected
		.316 unprotected
		.312 unprotected, no 40



"OBA":  .406 pro
		.518 Unpro
		.507 Unpro40



SLG:    .581 pro
        .662 Unpro
        .668 Unpro40

	

"O"PS:  .987 Pro
       1.180 Unpro
       1.175 Unpro40



Walks per 100 "PA"  15.45 Pro
                    29.57 Unpro
                    28.32 Unpro40



K's per 100 "PA"    10.91 Pro
                    11.59 Unpro
                    12.94 Unpro40

For fun, I did Matt Williams, as he spent some time protected by Barry last season.

Protected - 106 18 31 18 6 24 2 0 5
Unprotected - 295 51 91 66 32 67 13 1 17
Ave (.292 .308)  1st # is protected, 
SLG (.453 .532)  2nd unprotected
OBA (.330 .376)
OPS (.783 .908)