Matchup/Game Thread Game # 130 at Pirates

Andy Pettitte	Zach Duke
11 - 13, 4.51	8 - 11, 4.94

Can't help myself. Obsessed with Willy T and his streak.

Had a little mail exchange with Marc Normandin over at SBnation affiliate Beyond the Box Score, in which I said

. . . [Tony] Eusebio had a 24-game streak but it was over nearly 2-1/2 [sic] months, since he was the backup catcher and pinch-hitter. I was pissed when Kent broke his record, because it's nothing for a hitter like Kent to have a record like that, but for a Eusebio it's truly an amazing accomplishment

So I'm glad Willy has taken it back for all us mediocre talents. Actually that's not fair, Willy has one major talent, speed, and then everything else is ordinary. And THAT is much like the rest of us in this world. Most of us ain't five tool players, let's just say this.

And reading back over it this morning, after Willy extended the streak to 28 last night, I thought to myself, "Shit yeah. Willy's only halfway to Joe D, but the odds for a player of Willy's talents to having a 28 game hit streak have got to be equal to the odds for a player of Dimaggio's immense talents having a streak twice as long."

I mean, we all love Willy T, but I'm sure no-one thinks that his hitting skills taken on their own--without the prodigious speed--would have taken him anywhere above High A ball.

So I looked around the net for a methodology to test my conjecture and came across the following page: http://www.johnny-web.com/dimaggio/dimaggio.htm. I won't reproduce his explanation, because that's why I provided the link, but basically, a guy by the name of Greg Wroblewski looks at hits per plate appearance (H/PA), and plate appearances per game (PA/G), to come up with the odds of any particular player starting a hit streak of any particular length on any particular day.

If the methodology is correct (and it seems to be, but I sure ain't a mathematician), then you'd first figure the odds of the guy NOT getting a hit in any one game, based on H/PA and PA/G, then take it to the whatevereth power, to simulate a succesion of games.

He works all this through for Joe D, and based on the fact that Dimaggio had an overall 81.1% chance of getting a hit in anyone game during his streak year of 1941, comes up with .000804% as the percentage chance that a player with Dimaggio's numbers would start a 56-game hit streak on any particular day.

Basically, this is one chance in 124,000. From there, he goes on to figure how many times Dimaggio had a chance to start such a streak during the 1941 season, but I was less interested in that: I want to focus on the idea of Wily Taveras on July 26, a slumping slap-hitter batting .258, and with 5 hits in his last 20 at-bats. What are the odds that THAT hitter would start a 28-game streak the following day? Not very good, it turns out. Not very good at all. In fact, the numbers say that the odds of Willy putting together just the streak he has so far are one order of magnitude more remote than the odds that any hypothetical Yankee Clipper would have been looking at had he started in the outfield, Opening Day, 1942.

That's ten times more remote.

Though I had a conception of the situation, the difference surprised even me. The root of the huge disparity it comes from that Hits per plate appearance number. Dimaggio was much more likely to get a hit in any one game, and that likeliness reverberates through consecutive iterations. I'm not going to post it here, but you can view some numbers over here that show what Taveras' odds of starting, and then maintaining, once he got going, a 28-game streak.

The same numbers say that he's got a 65% chance of hitting his way on tonight, and for daydreamers like myself, a 1 in 130,000 chance of eventually tying Dimaggio.

You go, Willy.

I will append two charts: 1) the chart I started at game twenty, because this post is going to push it off the page, and 2) a chart I made for Mr. Normandin, which gives us a look at Willy's rolling totals in hits/game and total bases/game during the streak.

Astro Hit Streaks of 20 Games or More
Player	Games	Dates	Hits During Streak	Hits per Game	Number of One-Hit Games
Willy Taveras*	30	7/27 - 8/27/2006	45	1.50	18
Jeff Kent	25	5/14 - 6/11/2004	33	1.32	18
Tony Eusebio	24	7/9 - 8/28/2000	36	1.50	16
Art Howe	23	5/1 - 5/24/1981	40	1.74	8
Luis Gonzalez	23	5/26 - 6/20/1997	33	1.43	16
Moises Alou	23	6/22 - 7/19/2001	36	1.57	13
Cesar Cedeño	22	8/25 - 9/21/1977	37	1.68	13
Lee May	21	5/30 - 6/22/1973	35	1.67	11
Dickie Thon	21	7/24 - 8/14/1982	28	1.33	14
Lance Berkman	21	6/17 -   7/8/2001	40	1.90	7
Rusty Staub	20	6/30 - 7/21/1967	33	1.65	9
Kevin Bass	20	7/24 - 8/13/1986	30	1.50	13
---- *Active