Help with Math for a 3 Game Series in MLB. Im in over my head!

**danshan11** · 07-31-18, 09:16 AM

I think you are trying to connect things that have no connection.
if you get 10 blacks in a row on the roulette table will the next spin be black and the answer is the implied percentage is the same.
Streaks of any kind are a long time fallacy that people bet wholeheartedly. Any correlation of a streak pattern is noise. Each game is very independent of each other and presents its own unique implied.
if the closing line is +100 assuming no vig there is a 50% chance and the previous 1 2 or 3 games have no involvement in the probability.
Basically streaks are noise. Tons of people will argue differently but reality and historical data proves it

**danshan11** · 07-31-18, 09:24 AM

you can play with this query to check it yourself but even if you think you see something it is just deviations not edge

500: Internal Server Error

http://sportsdatabase.com/mlb/query?output=default&su=1&ou=1&sdql=p%3AW+and+series+game%3D3+and+pp%3AW&submit=++S+D+Q+L+%21++

p:W is previous win, pp:W is previous previous win (2 games back) and ppp:W is 3 games back hope you understand so you can play with it and any perceived edge is just deviation usually and Sportsdatabase adds in a -105 -105 margin for the line and roi

**BestBoyMike** · 07-31-18, 09:49 AM

Originally posted by danshan11

http://sportsdatabase.com/mlb/query?...+S+D+Q+L+%21++

p:W is previous win, pp:W is previous previous win (2 games back) and ppp:W is 3 games back hope you understand so you can play with it and any perceived edge is just deviation usually and Sportsdatabase adds in a -105 -105 margin for the line and roi

Thanks for the link! I'll take a look and play around with it.

**BestBoyMike** · 07-31-18, 10:01 AM

Originally posted by danshan11

I think you are trying to connect things that have no connection.
if you get 10 blacks in a row on the roulette table will the next spin be black and the answer is the implied percentage is the same.
Streaks of any kind are a long time fallacy that people bet wholeheartedly. Any correlation of a streak pattern is noise. Each game is very
independent of each other and presents its own unique implied.
if the closing line is +100 assuming no vig there is a 50% chance and the previous 1 2 or 3 games have no involvement in the probability.
Basically streaks are noise. Tons of people will argue differently but reality and historical data proves it

Ok, each game is independent of one another but my %...

When a team wins Game 1 (in a 3 game series)...
they win 1 more game (in the series) 48.9% of the time and
they win 2 more games (sweep) 31.2%

are based on 1,648 Total (3 Game Series)
Seems like a decent sample size

All that being said i was just looking for the pure math and %. If i can find a way to bet that, is a whole separate question.
So my question still hangs out there if someone likes to crunch numbers.
Thanks

**Waterstpub87** · 07-31-18, 10:10 AM

You won't be able to use this win. Anything that is a very simple angle like this won't work.

The naive way to solve this, assuming a team is 50/50

Win 2nd Game .50
Win 3rd Game .5

To sweep .5*.5=.25

But you have to take into account if a team is a favorite, it is more likely to sweep

Consider if we have a 60/ 40
Win 2nd: .6
Win 3rd: .6

To sweep .6*.6= .36

So if you consider 60% teams are more likely to win the first game, your sample is going to be biased towards those teams, pushing your sweep number up, making it look like its correlated. But it isn't.

**danshan11** · 07-31-18, 11:03 AM

Originally posted by BestBoyMike

Ok, each game is independent of one another but my %...

When a team wins Game 1 (in a 3 game series)...
they win 1 more game (in the series) 48.9% of the time and
they win 2 more games (sweep) 31.2%

are based on 1,648 Total (3 Game Series)
Seems like a decent sample size

All that being said i was just looking for the pure math and %. If i can find a way to bet that, is a whole separate question.
So my question still hangs out there if someone likes to crunch numbers.
Thanks

5902-5517 (0.13, 51.7%)

avg line: -112.6 / -100.1

on / against: -$12,648 / -$43,989

ROI: -0.9% / -3.2%

as you can see the win % correlates with the line

now lets look at this same scenario where the line was -160 or basically the implied was 61%, they win 61% of the time

147-94 (0.84, 61.0%)

avg line: -160.0 / 150.0

on / against: -$340 / -$600

ROI: -0.9% / -2.5%

**danshan11** · 07-31-18, 11:04 AM

here we see the same scenario on a dog as you can see the team wins 45% of the time and that is very close to the line impied of 120 is 45.5%

153-183 (-0.47, 45.5%)

avg line: 120.0 / -130.0

on / against: +$60 / -$1,597

ROI: +0.2% / -3.7%

**danshan11** · 07-31-18, 11:07 AM

there is no connection that is creating an edge with a previous game series win and game 2
look at wednesday verse tuesday, with this mentality you would only bet it on wednesday's now you got to know that makes no sense

1115-967 (0.26, 53.6%)

avg line: -114.6 / 102.0

on / against: +$6,455 / -$16,609

ROI: +2.4% / -6.7%

1691-1627 (0.04, 51.0%)

avg line: -112.9 / 100.3

on / against: -$9,969 / -$7,627

ROI: -2.4% / -1.9%

**danshan11** · 07-31-18, 11:12 AM

whenever you think a trend showing positive signs think of coin flips
if you flip a coin 1000 times (decent sample size right)
if you did it 5 times
2 times heads wins 550 times
2 times heads wins 450 times
1 times heads wins 500 times
this is a good sample size but deviations make it seem good when it is just noise, DONT BET NOISE!
these deviation numbers are just out of my head, the true deviation numbers are more precise but I am trying to just give you the idea of how it works and you would not bet on heads because the last 1000 rolls it won 54% of the time

**BestBoyMike** · 07-31-18, 11:20 AM

Originally posted by danshan11

5902-5517 (0.13, 51.7%)

avg line: -112.6 / -100.1

on / against: -$12,648 / -$43,989

ROI: -0.9% / -3.2%

as you can see the win % correlates with the line

now lets look at this same scenario where the line was -160 or basically the implied was 61%, they win 61% of the time

147-94 (0.84, 61.0%)

avg line: -160.0 / 150.0

on / against: -$340 / -$600

ROI: -0.9% / -2.5%

Danshan11 - Great explanations, Thank you! It all makes scenes, sometimes i just need to see things from a different view point.
Dont bet noise is a very simple but powerful concept.
Thanks again for the examples!

**danshan11** · 07-31-18, 11:28 AM

good luck, anytime you need anything let me know, you would be amazed at the number of bettors who believe in weird streaky stuff, unfortunately for Nerfs they learn them and for sharps they eat from the Nerf bets on these.

**VeggieDog** · 07-31-18, 12:15 PM

"When a team wins Game 1 (in a 3 game series)...
they win 1 more game (in the series) 48.9% of the time and
they win 2 more games (sweep) 31.2% "

What if you broke it down between Home teams and Road teams? I would guess the Home team has a higher win rate.

**danshan11** · 07-31-18, 01:24 PM

Originally posted by VeggieDog

"When a team wins Game 1 (in a 3 game series)...
they win 1 more game (in the series) 48.9% of the time and
they win 2 more games (sweep) 31.2% "

What if you broke it down between Home teams and Road teams? I would guess the Home team has a higher win rate.

Any team home away old ugly would have a win rate based on the game in particular nothing to do with the past games.

**BestBoyMike** · 07-31-18, 06:10 PM

Originally posted by VeggieDog

"When a team wins Game 1 (in a 3 game series)...
they win 1 more game (in the series) 48.9% of the time and
they win 2 more games (sweep) 31.2% "

What if you broke it down between Home teams and Road teams? I would guess the Home team has a higher win rate.

Those stats are for home teams, i have the stats for away teams also. Not as much difference as you would think.
However as my new friend Danshan11 has pointed out, there is no connection that is creating an edge. Sorry