Pitching Probabilities

prohornblower · #1 08-10-2007, 04:44 PM

I'm currently doing a little study, but I need the assistance of people more knowledgeable than myself here.

I'm trying to figure out what the hypothetical probabilities (I say probabilities but really looking for %)
of each starting pitcher to win a ballgame. (Or I should say, give his team a chance to win).

I'm guessing they look like this:
Ace: 66%
2: 58%
3: 50%
4: 42%
5: 34%

The range may be more like 36%-64%. 66% is pretty high and 34% seems low.

Also, what kind of deduction can we expect if they pitch on short (1 day less) rest? I guestimate that they are 10% less effective on short rest.

And finally, what increase in chance can we expect if they pitch on extended (1+ day more) rest? I guestimate that they are 8% more effective on extra rest.

I also believe there is likely no real noticeable additional increase for 2 days extra rest, as opposed to 1 day extra rest. And, my guestimates may be off, but certainly short rest is more damaging than extra rest is improving.

Thanks for the help, guys! [img]/images/graemlins/grin.gif[/img]

If you do not have/know the real data, please by all means speculate what you think the numbers are.

Rubeskies · #2 08-10-2007, 04:48 PM

A % would depend hugely on the team he plays for and his opponent. A blanket % would be misleading on any given day.

Are you looking for a league Average?

prohornblower · #3 08-10-2007, 04:55 PM

Yeah, hypothetical averages.

slothinator · #4 08-10-2007, 05:12 PM

I can pass on to you what I use for projecting W/L for my fantasy projections each year. I project a pitchers W-L record based on his average innings per start, his ERA, and his expected run support.

I assume that a pitcher who averages 6 innings per start will factor in the decision 66.7% of the time (I'm not sure what the exact number is, or where I got this from. But it has always worked pretty well). Basically it is IP/start divided by 9 to get "decision %". I multiply that % times the number of starts I project the pitcher to make, and then run that through Bill James W-L forumla, using the pitchers projected ERA and and his teams expected runs/game. That gives me that pitchers expected W-L record.

Like I said, it's not an exact science (I don't have a team of statisticians like BP, I'm a one man gang), but it usually gives me a relatively good idea for the season. I'm not sure how it would apply for a single game application. Let me know if you find this helpful in anyway with what you are doing.

Rubeskies · #5 08-10-2007, 05:13 PM

So I guess all you'd have to do was rank all pitchers over a time period, a few years maybe, into categories from Ace through 5th starter.

Then just see how their team did when those guys pitched. Is there something I'm missing?

prohornblower · #6 08-10-2007, 05:54 PM

Ehhh, I probably need to clarify here.

Mainly what I'm looking for has more to do with how much better is an ace, and #2 starter than average? And how much worse is a #4 & #5 starter than average?

I'm not so concerned with the win-loss results, per se. It's more like, how much more valuable are the top 2 compared to league average, and how less valuable are the bottom 2 compared to league average? It's more about giving the team a chance to win, and less about what has happened.

Also, if y'all can chime in on the quantifiable effects of pitching on short rest as compared to extra rest, that would be awesome. These deductions/increases should affect each pitcher in a 5-man rotation identically, hypothetically.

prohornblower · #7 08-10-2007, 05:57 PM

Maybe what I can do is take the top 30 (starting) pitchers from last year in terms of ERA+ to find the ace, and take pitchers 31-60 in ERA+ to get the #2 guy, and then just mirror the numbers about the 50% "axis" to project how bad the #4 and #5 guys are?

EDIT: I've already thought of a bug. ERA+ takes into account relievers, so the average starters ERA+ may be more like 102 or so. Crap.

Benholio · #8 08-10-2007, 06:00 PM

That sounds like as good a strategy as any.

Thremp · #9 08-10-2007, 06:05 PM

Sportsbetting is a better place for this question. You'll get higher quality responses from those that choose to answer.

prohornblower · #10 08-10-2007, 06:13 PM

[ QUOTE ]
Sportsbetting is a better place for this question. You'll get higher quality responses from those that choose to answer.

[/ QUOTE ]

Are you implying that there are posters in Sports who would read this and choose to not answer, even if they have a good answer? That is a pretty loaded comment and I'm not sure what your angle is. Or maybe you are talking about the extra traffic on this forum and higher probability for less-thought-out responses?

But I will consider cross-posting in that forum.