clock menu more-arrow no yes

Filed under:

Evaluating one Giants evaluation

New, 73 comments

One projection system has evaluated the Giants in every game this year, so let’s evaluate it

San Francisco Giants v Chicago Cubs
Most Giants photo of the year? Matt Moore sad in the foreground with Anthony Rizzo’s home run trot in the background.
Photo by Jonathan Daniel/Getty Images

As a writer for a prominent Giants blog (which one? I’ll never tell you), I often get unsolicited emails from people and companies hoping I’ll mention their product. Often, I will ignore them — while I like free things, I probably don’t really need a McCovey Cove Dave bobblehead, just to cite one example — and then never respond to the email and leave them hanging on my response forever. Take that, marketing people trying to do your jobs!

All year, I’ve been getting series preview emails from some site called Numberfire — I debated not including its name in this article just because it would be funny but the responsibility to cite my sources won out — about what the percent chance is of the Giants winning each game in each series they’ve played this year. I thought it might be fun to look at how those percentages have played out in the first half? Why? Because I’m a butt.

Reality Vs. Odds

Date Opp W L Season W Season L Odds W L Season W Season L Right Wrong Season Right Season Wrong Accuracy Season Accuracy Pct off per game
Date Opp W L Season W Season L Odds W L Season W Season L Right Wrong Season Right Season Wrong Accuracy Season Accuracy Pct off per game
04/02/17 @ARI 1 0 1 0.564 1 1 0 0 1 0 1 -0.564 -0.564 -56.40
04/04/17 @ARI 1 1 1 0.588 1 2 0 1 0 1 1 0.588 0.024 1.20
04/05/17 @ARI 1 1 2 0.542 1 3 0 0 1 1 2 -0.542 -0.518 -17.27
04/06/17 @ARI 1 1 3 0.5 3 0 0 0 1 2 0 -0.518 -12.95
04/07/17 @SD 1 1 4 0.453 1 3 1 1 0 2 2 0.453 -0.065 -1.30
04/08/17 @SD 1 1 5 0.544 1 4 1 0 1 2 3 -0.544 -0.609 -10.15
04/09/17 @SD 1 2 5 0.542 1 5 1 1 0 3 3 0.542 -0.067 -0.96
04/10/17 ARI 1 3 5 0.602 1 6 1 1 0 4 3 0.602 0.535 6.69
04/11/17 ARI 1 3 6 0.568 1 7 1 0 1 4 4 -0.568 -0.033 -0.37
04/12/17 ARI 1 4 6 0.564 1 8 1 1 0 5 4 0.564 0.531 5.31
04/13/17 COL 1 4 7 0.586 1 9 1 0 1 5 5 -0.586 -0.055 -0.50
04/14/17 COL 1 5 7 0.583 1 10 1 1 0 6 5 0.583 0.528 4.40
04/15/17 COL 1 5 8 0.571 1 11 1 0 1 6 6 -0.571 -0.043 -0.33
04/16/17 COL 1 5 9 0.619 1 12 1 0 1 6 7 -0.619 -0.662 -4.73
04/18/17 @KC 1 6 9 0.464 1 12 2 0 1 6 8 -0.464 -1.126 -7.51
04/19/17 @KC 1 6 10 0.595 1 13 2 0 1 6 9 -0.595 -1.721 -10.76
04/21/17 @COL 1 6 11 0.551 1 14 2 0 1 6 10 -0.551 -2.272 -13.36
04/22/17 @COL 1 6 12 0.549 1 15 2 0 1 6 11 -0.549 -2.821 -15.67
04/23/17 @COL 1 6 13 0.57 1 16 2 0 1 6 12 -0.57 -3.391 -17.85
04/24/17 LAD 1 7 13 0.529 1 17 2 1 0 7 12 0.529 -2.862 -14.31
04/25/17 LAD 1 7 14 0.476 1 17 3 1 0 8 12 0.476 -2.386 -11.36
04/26/17 LAD 1 8 14 0.542 1 18 3 1 0 9 12 0.542 -1.844 -8.38
04/27/17 LAD 1 8 15 0.501 1 19 3 0 1 9 13 -0.501 -2.345 -10.20
04/28/17 SD 1 9 15 0.612 1 20 3 1 0 10 13 0.612 -1.733 -7.22
04/29/17 SD 1 9 16 0.539 1 21 3 0 1 10 14 -0.539 -2.272 -9.09
04/30/17 SD 1 9 17 0.601 1 22 3 0 1 10 15 -0.601 -2.873 -11.05
05/01/17 @LAD 1 10 17 0.422 1 22 4 0 1 10 16 -0.422 -3.295 -12.20
05/02/17 @LAD 1 10 18 0.425 1 22 5 1 0 11 16 0.425 -2.87 -10.25
05/03/17 @LAD 1 11 18 0.454 1 22 6 0 1 11 17 -0.454 -3.324 -11.46
05/05/17 @CIN 1 11 19 0.492 1 22 7 1 0 12 17 0.492 -2.832 -9.44
05/06/17 @CIN 1 11 20 0.513 1 23 7 0 1 12 18 -0.513 -3.345 -10.79
05/07/17 @CIN 1 11 21 0.557 1 24 7 0 1 12 19 -0.557 -3.902 -12.19
05/08/17 @NYM 1 11 22 0.476 1 24 8 1 0 13 19 0.476 -3.426 -10.38
05/09/17 @NYM 1 11 23 0.533 1 25 8 0 1 13 20 -0.533 -3.959 -11.64
05/10/17 @NYM 1 12 23 0.527 1 26 8 1 0 14 20 0.527 -3.432 -9.81
05/11/17 CIN 1 12 24 0.568 1 27 8 0 1 14 21 -0.568 -4 -11.11
05/12/17 CIN 1 13 24 0.584 1 28 8 1 0 15 21 0.584 -3.416 -9.23
05/13/17 CIN 1 14 24 0.573 1 29 8 1 0 16 21 0.573 -2.843 -7.48
05/14/17 CIN 1 15 24 0.592 1 30 8 1 0 17 21 0.592 -2.251 -5.77
05/15/17 LAD 1 16 24 0.478 1 30 9 0 1 17 22 -0.478 -2.729 -6.82
05/16/17 LAD 1 17 24 0.481 1 30 10 0 1 17 23 -0.481 -3.21 -7.83
05/17/17 LAD 1 17 25 0.572 1 31 10 0 1 17 24 -0.572 -3.782 -9.00
05/19/17 @STL 1 18 25 0.428 1 31 11 0 1 17 25 -0.428 -4.21 -9.79
05/20/17 @STL 1 19 25 0.441 1 31 12 0 1 17 26 -0.441 -4.651 -10.57
05/21/17 @STL 1 19 26 0.405 1 31 13 1 0 18 26 0.405 -4.246 -9.44
05/22/17 @CHC 1 20 26 0.429 1 31 14 0 1 18 27 -0.429 -4.675 -10.16
05/23/17 @CHC 1 20 27 0.451 1 31 15 1 0 19 27 0.451 -4.224 -8.99
05/24/17 @CHC 1 20 28 0.393 1 31 16 1 0 20 27 0.393 -3.831 -7.98
05/25/17 @CHC 1 20 29 0.538 1 32 16 0 1 20 28 -0.538 -4.369 -8.92
05/26/17 ATL 1 20 30 0.542 1 33 16 0 1 20 29 -0.542 -4.911 -9.82
05/27/17 ATL 1 21 30 0.556 1 34 16 1 0 21 29 0.556 -4.355 -8.54
05/28/17 ATL 1 22 30 0.619 1 35 16 1 0 22 29 0.619 -3.736 -7.18
05/29/17 WAS 1 22 31 0.479 1 35 17 1 0 23 29 0.479 -3.257 -6.15
05/30/17 WAS 1 22 32 0.516 1 36 17 0 1 23 30 -0.516 -3.773 -6.99
05/31/17 WAS 1 22 33 0.472 1 36 18 1 0 24 30 0.472 -3.301 -6.00
06/02/17 @PHI 1 23 33 0.473 1 36 19 0 1 24 31 -0.473 -3.774 -6.74
06/03/17 @PHI 1 23 34 0.524 1 37 19 0 1 24 32 -0.524 -4.298 -7.54
06/04/17 @PHI 1 23 35 0.49 1 37 20 1 0 25 32 0.49 -3.808 -6.57
06/05/17 @MIL 1 24 35 0.49 1 37 21 0 1 25 33 -0.49 -4.298 -7.28
06/06/17 @MIL 1 24 36 0.458 1 37 22 1 0 26 33 0.458 -3.84 -6.40
06/07/17 @MIL 1 24 37 0.446 1 37 23 1 0 27 33 0.446 -3.394 -5.56
06/08/17 @MIL 1 25 37 0.5 0 37 23 0 0 27 33 0 -3.394 -5.47
06/09/17 MIN 1 25 38 0.54 1 38 23 0 1 27 34 -0.54 -3.934 -6.24
06/10/17 MIN 1 25 39 0.561 1 39 23 0 1 27 35 -0.561 -4.495 -7.02
06/11/17 MIN 1 26 39 0.54 1 40 23 1 0 28 35 0.54 -3.955 -6.08
06/13/17 KC 1 26 40 0.445 1 40 24 1 0 29 35 0.445 -3.51 -5.32
06/14/17 KC 1 26 41 0.492 1 40 25 1 0 30 35 0.492 -3.018 -4.50
06/15/17 @COL 1 26 42 0.458 1 40 26 1 0 31 35 0.458 -2.56 -3.76
06/16/17 @COL 1 26 43 0.522 1 41 26 0 1 31 36 -0.522 -3.082 -4.47
06/17/17 @COL 1 26 44 0.434 1 41 27 1 0 32 36 0.434 -2.648 -3.78
06/18/17 @COL 1 26 45 0.474 1 41 28 1 0 33 36 0.474 -2.174 -3.06
06/19/17 @ATL 1 26 46 0.484 1 41 29 1 0 34 36 0.484 -1.69 -2.35
06/20/17 @ATL 1 26 47 0.478 1 41 30 1 0 35 36 0.478 -1.212 -1.66
06/21/17 @ATL 1 27 47 0.505 1 42 30 1 0 36 36 0.505 -0.707 -0.96
06/22/17 @ATL 1 27 48 0.449 1 42 31 1 0 37 36 0.449 -0.258 -0.34
06/23/17 NYM 1 27 49 0.563 1 43 31 0 1 37 37 -0.563 -0.821 -1.08
06/24/17 NYM 1 27 50 0.516 1 44 31 0 1 37 38 -0.516 -1.337 -1.74
06/25/17 NYM 1 27 51 0.516 1 45 31 0 1 37 39 -0.516 -1.853 -2.38
06/26/17 COL 1 28 51 0.531 1 46 31 1 0 38 39 0.531 -1.322 -1.67
06/27/17 COL 1 29 51 0.499 1 46 32 0 1 38 40 -0.499 -1.821 -2.28
06/28/17 COL 1 30 51 0.581 1 47 32 1 0 39 40 0.581 -1.24 -1.53
06/30/17 @PIT 1 31 51 0.45 1 47 33 0 1 39 41 -0.45 -1.69 -2.06
07/01/17 @PIT 1 32 51 0.449 1 47 34 0 1 39 42 -0.449 -2.139 -2.58
07/02/17 @PIT 1 33 51 0.478 1 47 35 0 1 39 43 -0.478 -2.617 -3.12
07/04/17 @DET 1 33 52 0.395 1 47 36 1 0 40 43 0.395 -2.222 -2.61
07/05/17 @DET 1 34 52 0.44 1 47 37 0 1 40 44 -0.44 -2.662 -3.10
07/06/17 @DET 1 34 53 0.424 1 47 38 1 0 41 44 0.424 -2.238 -2.57
07/07/17 MIA 1 34 54 0.474 1 47 39 1 0 42 44 0.474 -1.764 -2.00
07/08/17 MIA 1 34 55 0.529 1 48 39 0 1 42 45 -0.529 -2.293 -2.58
07/09/17 MIA 1 34 56 0.503 1 49 39 0 1 42 46 -0.503 -2.796 -3.11

I had the table all set up nicely with the word “Real” above the first six columns, “Projected” above the next five, and “Analysis” above the last seven, but apparently I’m not good enough at this to use merged columns, so you get slightly crappier formatting. Sorry! Also, there were two games that the system saw as 50/50, so I just said both of them were neither system wins nor losses for the sake of the analysis. This could be bad form, and I don’t really care because it doesn’t matter.

Anyway, here’s what this shows: the projections system was basically 50/50 on its ability to project any given game, but the level of confidence it showed in each game was overstated. That’s what the last two columns, the “Accuracy” and “Season Accuracy” measure. For example, in the first game, which the system got wrong, it said the Giants had a 56.4% chance of winning, so it got a -.564 on accuracy. For the second game, it said the Giants had a 58.8% chance of winning, so it got a .588. Then Season Accuracy aggregated that, so after those two games, you had a sense of how the system was doing overall, and then “Pct off per game” divides that aggregate by the number of games and comes out with an overall percent error. I’m explaining this process in detail because I’m expecting someone to tell me why I shouldn’t do it like this and then I’ll learn something for free, you suckers.

Anyway, here’s the main takeaway: the system has been off by just 3.1% so far this year. There is a 3.1% difference between the 50-40 team that the system projected the Giants to be (even with a massive course correction it made about a month into the season) and the 36-56 crapfest that we’ve been watching for months. It seems like Jonah Keri was wrong. The extra 3% is really the key to building a baseball team.

Of course, there’s one other way to look at it: The Giants are so bad that the system can predict their losses with astonishing certainty. So even though the overall record of the system in terms of its wins and losses is not very good, it has now so thoroughly learned that the Giants are irredeemably horrible that it has a shot of erasing the massive mathematical deficit it built up for itself by thinking they were good.

In conclusion the Giants are bad when they were supposed to be good, and you already knew that, and here’s a big old table that says it anyway.