MLB Capping basics (Video)

[Flyers09]

It wins 89%. You should go deposit your entire paycheck to an online book right now.

well i found another baby here at sbr once again

[Flyers09]

according to this theory the boston/cleveland game is a gift
Boston=69%
Cleveland=40%

Boston=-115
Cleveland=-105

Boston is the winner by a huge percentage...according to this theory

[Justin7]

according to this theory the boston/cleveland game is a gift
Boston=69%
Cleveland=40%

Boston=-115
Cleveland=-105

Boston is the winner by a huge percentage...according to this theory

huh?

What are your projected runs for Boston and Cleveland?

[Flyers09]

i went with the percentage of this year, and came up with that, as for the other games it was pretty even, this is the only game that stood out, as for getting projected runs for a team in its last 100 games, where am i to find that info at?

[Flyers09]

are you checking to see if thats acceptable justin7??

[durito]

[Flyers09]

[suicidekings]

according to this theory the boston/cleveland game is a gift
Boston=69%
Cleveland=40%

Boston=-115
Cleveland=-105

Boston is the winner by a huge percentage...according to this theory

If you were capping the BOS-CLE game using this method and this seasons data (just for example),

BOS: 112 runs scored, 80 runs allowed
CLE: 103 runs scored, 115 runs allowed
*Assume same number of games played for simplicity

E[W](BOS) = 112^1.82 / ( 80^1.82 + 112^1.82 ) = 0.648
E[W](CLE) = 103^1.82 / ( 103^1.82 + 115^1.82 ) = 0.450
E[W] = Expected team winning percentage

Then,

P[W](BOS) = 0.648 / (0.648 + 0.45) = 59%
P[W](CLE) = 0.45 / (0.648 + 0.45) = 41%
P[W] = Probability of winning this game

So Boston has a 59% chance of winning this game. That doesn't mean they won't potentially get blown out in this game, so it's far from being a sure thing. Instead, it implies that if this exact game was played under similar conditions many times, Boston should win approximately 59% of those games.

[suicidekings]

But again, that doesn't factor in the specific starting pitcher matchup, which is a pretty significant element to leave out.

[Flyers09]

wait whats the ^ mean?????

[Flyers09]

If you were capping the BOS-CLE game using this method and this seasons data (just for example),

BOS: 112 runs scored, 80 runs allowed
CLE: 103 runs scored, 115 runs allowed
*Assume same number of games played for simplicity

E[W](BOS) = 112^1.82 / ( 80^1.82 + 112^1.82 ) = 0.648
E[W](CLE) = 103^1.82 / ( 103^1.82 + 115^1.82 ) = 0.450
E[W] = Expected team winning percentage

Then,

P[W](BOS) = 0.648 / (0.648 + 0.45) = 59%
P[W](CLE) = 0.45 / (0.648 + 0.45) = 41%
P[W] = Probability of winning this game

So Boston has a 59% chance of winning this game. That doesn't mean they won't potentially get blown out in this game, so it's far from being a sure thing. Instead, it implies that if this exact game was played under similar conditions many times, Boston should win approximately 59% of those games.

look at the spots i bolded, where r u getting them numbers? r they standard numbers that r used every time?

[suicidekings]

The 0.45 was calculated in the E[W](CLE) line above. The 1.82 comes from empirical testing done on the system. The original calculation used 2 as the exponent, it was determined that an exponent of 1.82 provides a better correlation to actual team winning percentages in baseball. Other sports use different exponents. The link below gives a little more info.

http://armchairgm.wikia.com/Pythagorean_Formula

The ^ means an exponent.

[Flyers09]

ok whats the ^ mean

i think i finally understand this...i appreicate it

[losturmarbles]

exponent

3^2

is 3 squared

[Flyers09]

this math problem
E[W](BOS) = 112^1.82 / ( 80^1.82 + 112^1.82 ) = 0.648
Does not equal .648, im getting .583

??

[whatisit]

this math problem
E[W](BOS) = 112^1.82 / ( 80^1.82 + 112^1.82 ) = 0.648
Does not equal .648, im getting .583

??

You might be doing it wrong bro, it's .648. Make sure you're doing the operations in the right order.

[Flyers09]

i am still getting .583

[Flyers09]

and im getting .472 for cleveland


no way the calculations r right

[Justin7]

^1.82 is better than ^2, but they are close.

I would NEVER use season-long runs scored to predict an individual game.

[Flyers09]

^1.82 is better than ^2, but they are close.

I would NEVER use season-long runs scored to predict an individual game.

so you use last seasons runs?

if so how is that effective? what if players r traded etc etc

[losturmarbles]

and im getting .472 for cleveland


no way the calculations r right

its not advanced math flyers

[Justin7]

so you use last seasons runs?

if so how is that effective? what if players r traded etc etc

As in the video, I'd use the current batting lineup, starting pitcher, and bullpen to set projected runs. I'd use THAT to set the ML.

I've had a lot of feedback and questions on this topic, so I might do a demo video on it.

[Flyers09]

As in the video, I'd use the current batting lineup, starting pitcher, and bullpen to set projected runs. I'd use THAT to set the ML.



I've had a lot of feedback and questions on this topic, so I might do a demo video on it.

if you could that would be highly appreciated

[suicidekings]

Justin:

If you ignore current season data, how do you account for a pitcher when his current season performance deviates significantly from his career stats? Take Zach Duke for example: Career ERA = 4.30, BAA = .305, etc, but has been pitching very well this year so far. If you were capping a game in which he was starting, how would you approach estimating how many runs he will allow?

[Justin7]

Justin:

If you ignore current season data, how do you account for a pitcher when his current season performance deviates significantly from his career stats? Take Zach Duke for example: Career ERA = 4.30, BAA = .305, etc, but has been pitching very well this year so far. If you were capping a game in which he was starting, how would you approach estimating how many runs he will allow?

I'd assume he was lucky so far, and will pitch going forward like he has in the past. And I'd be more right than wrong.