# Projecting Game 1 of the Giants vs. Braves NLDS

Trying to predict the results of any MLB playoff series isn’t much different than guessing heads or tails. Supposed experts that are employed by ESPN, MLB, Yahoo or any online sports website made for the casual fan often overstate the chances for any given team and do the opposite for the weaker of the eight teams playing in October. For instance, 9 of 10 of ESPN’s experts predict that the Giants will beat the Braves and not one is picking against the Phillies. And you can’t really blame them. Both the Giants and the Phillies are the superior teams at this point in the season; no explanation for the Phillies, and the Braves have scuffled in the 2nd half and are weak due to injuries to the likes of Chipper Jones, Martin Prado and Takashi Saito. However, they’re simply predicting what they see happening, and aren’t giving any statistical odds in their results.

The fact of the matter is that in a very short series (5 or 7 games) no team is a statistically clear-cut favorite to win. There’s simply way too much random variation that occurs in any given baseball game. For instance, what are the chances that a .600 baseball team beats another .400 team in a game (not considering pitching matchups; just consider that the pitching perfectly represents the team’s entire staff for that game)? A perfect example is the Philadelphia Phillies (.599 W% in 2010) playing the Arizona Diamondbacks (.401 W%). We’ll ignore homefield advantage and current rosters for simplicity and only look at the percentage of games that these two teams won this season. We’ll use the Odds Ratio to calculate the expected result:

W%(A v. B) = W%(A)*(1 – W%(B))/(W%(A)*((1 – W%(B)) + (1 – W%(A))*W%(B)))

So, after plugging in .6 for A and .4 for B, the expected result would be for the Phillies to win 69.2% of the time. So even one of the worst teams in the league still has a 30% chance of beating the best team. Intuitively, it makes sense; bad teams win plenty of baseball games as well, and there’s a good chance some of those wins come against good teams.

At Walk Like a Sabermetrician, a quick rundown (based on W% and pythagorean W%) was done to calculate the various odds of some of the possible playoff match-ups. The Phillies had the highest chance of advancing before the Division Series started, yet only at 59%. The best team using their rankings was the Yankees, with the worst being the Reds; if the two were to meet in the World Series the Yankees would have a 66% of being World Series champions, still manageable though for the Reds.

For the individual games it’s a bit trickier than that when factoring in pitching matchups, lineups, platoon splits, high-leverage bullpen arms and any thing that makes a team stronger or weaker than their total season numbers suggests (due to injuries, trades, roster changes, etc.). First, we’ll compare the expected lineups against each other and calculate their expected R/G based on true-talent levels:

Player | wOBA | rv600 | ePA | Player | wOBA | rv600 | ePA | |

Andres Torres |
.362 | 18.5 | 4.8 | Omar Infante |
.350 | 11.0 | 4.8 | |

Freddy Sanchez |
.335 | 3.3 | 4.68 | Jason Heyward |
.400 | 40.8 | 4.68 | |

Aubrey Huff |
.370 | 23.4 | 4.54 | Derrek Lee |
.371 | 22.0 | 4.54 | |

Buster Posey |
.376 | 26.5 | 4.46 | Brian McCann |
.378 | 25.7 | 4.46 | |

Pat Burrell |
.343 | 4.6 | 4.34 | Alex Gonzalez |
.311 | -11.7 | 4.34 | |

Juan Uribe |
.335 | 2.3 | 4.23 | Brooks Conrad |
.320 | -2.4 | 4.23 | |

Jose Guillen* |
.327 | -3.5 | 4.1 | Nate McLouth |
.331 | 1.1 | 4.1 | |

Pablo Sandoval |
.364 | 18.9 | 3.98 | Rick Ankiel |
.328 | -1.0 | 3.98 | |

Pitcher |
.228 | -56.7 | 3.86 | Pitcher |
.255 | -43.7 | 3.86 |

** **

**wOBA – **Weighted On Base Average

**rv600 – **Run Value (Above Average) per 600 PA

**ePA – **Expected PA per lineup slot in the NL

*There’s a possibility that Guillen won’t start (hopefully), and Cody Ross would likely take his place. Ross would add approximately .03 runs to the Giants offense per game (4.69 rather than 4.66)

The Giants lineup is expected to score 4.66 runs per game (.33 above the NL average), while the Braves would be expected to score 4.71 runs per game. That might be a red flag to anyone considering the Giants averaged 4.30 runs during the year while the Braves only averaged 4.56 and have lost Martin Prado and Chipper Jones for the year. Well, part of it is that a full-time player plays 90% of the time (barring injuries and position, as both Heyward and McCann have missed time due to each scenario, respectively) providing a good amount of plate appearances to average to even replacement level players. In the post-season however, one would expect the starting lineup to be the same for almost every game barring an injury or platoon situation. Both teams also have players whose true talent level suggests regression towards a higher level of production (Sandoval, Lee, McLouth), as well as the Giants having given plenty of playing time to players who are no longer in the lineup.

So the Braves lineup would be expected to score .05 runs more per game, which in a short series is almost completely insignificant, making both offenses virtually identical, without considering platoon or home/road splits.

Here we look at the Game 1 starters, Tim Lincecum and Derek Lowe, and again look at true-talent levels using tRA:

ERA | FIP | tRA | IP | |

Tim Lincecum |
2.91 | 2.79 | 3.12 | 6.7 |

Derek Lowe |
4.03 | 3.79 | 4.42 | 5.9 |

tRA comes from StatCorner’s tRA.

Bullpens are a fickle thing, and trying to eliminate the inferior pitchers from each team, as well as adjusting for injuries and such would simply create more noise than necessary. Instead, we’ll just finish off each starter’s expected line with the bullpen performance this season (again using tERA). The Giants sit at 3.46 with the Braves at 3.37.

Starter IP | Starter RA | Bullpen IP | Bullpen RA | Total RA | |

Giants |
6.7 | 2.32 | 2.3 | 0.88 | 3.20 |

Braves |
5.9 | 2.90 | 3.1 | 1.16 | 4.06 |

I excluded defense because after a few quick calculations, the difference between the two teams in any given game is less than 1/10th of a run; simply unnecessary noise.

Using the Odds Ratio combined with the pythagorean records expected from these numbers, we get these results:

Team | eRS | eRA | x-W% | % Victory |

Giants |
4.66 | 3.20 | .665 | 60.25%* |

Braves |
4.71 | 4.06 | .568 | 39.75%* |

**Expected Final Score:** Giants 4.39, Braves 3.58

*I have not factored in homefield advantage, something that I would like to do but was a bit unsure of how to apply in playoff situations. So the Giants chances are probably even better than the given numbers, holding everything else constant.

In conclusion, if these are indeed the lineups utilized tomorrow night, given the pitching match-ups, the Giants have a 60% chance of pulling out a victory in Game 1. I think even the average fan would expect some type of advantage like this given a Lincecum vs. Lowe duel.

The Giants really need to win the game where the odds are in their favor, because the next few games will be on a much more level-playing field given the projected pitching match-ups. We’ll see what happens tomorrow night. Tune in to TBS at 6:37 PM (Pacific Time) for the game!

This is confusing the shit out of me

Are you familiar with sabermetrics? If not, I could see how this would be a tad confusing as this is a bit more advanced.

Thanks for the legwork on this–very cool to check out as we lead up to the game.

No problem. Originally I was going to predict the series, but with pitching matchups changing I figured that would be too difficult. There was enough work anyways, so I can just do it game by game now.

Hopefully I’ll be doing this for a few weeks 🙂