Filed under:

Taking an in-depth look at the stats behind the game

Each Week FTRS will take apart the traditional box score and present our readers with the key insights on how the game was played. You can view all the Advanced Box Score stories in the Football Analytics Story Stream. For any questions about the metrics listed below please first refer to the Advanced Stats Glossary for definitions.

We will start with probably the most glaring question on every one's mind. Has a team ever had as poor of a performance as Pitt did on their first five possessions? Not in the last 10 years. I only have drive-by-drive data since 2005 and Pitt was the only team, out of 7760 games, so 15,520 team-games, to lose five straight fumbles. Here are some other interesting tidbits from my research:

If we want to have a little fun with it we can do some math (Ok wow, this next paragraph got long, feel free to skip if you don't feel like nerding out, but it's a pretty cool discussion).

Fumbles occur on 3% of all plays. Pittsburgh fumbled five times in their first thirteen plays. If we assume that the probability of a fumble is constant for each play and fumbling once doesn't impact your future chance of fumbling (offseason article note; check if this is true) then we can model the probability of fumbling a certain number of times in a given number of plays according to the binomial distribution. The probability of something happening five times out of thirteen chances when the true probability is 3% is .0000245. That should happen about once in every 40,000 groups of 13 plays, if my brain is working correctly. But that wasn't the only weird thing, Georgia Tech also recovered ALL FIVE of the fumbles. We have already shown that fumble recoveries are random. The probability of us recovering all five fumbles is the same probability of flipping a coin and getting 5 heads in a row (or 5 tails in a row). This is simply (1/2)^5, or 3.125%. A simple way to calculate the probability of Pitt fumbling 5 times in 13 plays and then Tech recovering all 5 of them is to simply assume they are independent, and that is what I am going to do, if you want to argue this point and give me a lesson in probability feel free to chat in the comments. Anyway if two events are independent then the probability of them both happening is simply the product of the two probability which in our case produces an overall probability of .000000766, or a 1 in 1.3 million chance. I'd hate myself if I didn't take advantage of this opportunity:

via gifrific

 Georgia Tech 56 - Pitt 28 Passing Downs Quarter Performance GT Pitt GT Pitt Plays 17 18 Yards / Play Success Rate Yards / Play Success Rate Yards / Play 12.06 6.94 Q1 11.05 65% 10.56 56% Success Rate 47% 44% Q2 4.68 32% 8.83 78% Standard Downs Q3 7.22 61% 7.6 60% GT Pitt Q4 9.05 74% 5.45 32% Plays 59 48 Down Performance Yards / Play 6.9 8.35 GT Pitt Success Rate 61% 58% Yards / Play Success Rate Yards / Play Success Rate Passing Plays 1 8.36 58% 7.66 48% GT Pitt 2 6.5 62% 9.13 58% Plays 12 38 3 11.58 50% 7.08 67% Yards / Play 12.25 8.11 4 1.5 50% 0 0% Success Rate 33% 53% Drive Performance Sack Rate (SD / PD) 0.0% / 0.0% 4.2% / 5.6% Num Drives Avg Start Spot % Methodical % Explosive Running Plays GT 13 59.38 23% 54% GT Pitt Pitt 13 76.54 23% 31% Plays 64 28 % of Possible Yards Gained GT: 75% Pitt: 52% Yards / Play 7.27 7.79 Georgia Tech Hidden Yards 223.08 Success Rate 63% 57% Turnovers GT: 1 Pitt: 6

So, obviously the 6 turnovers by Pitt kind of dominated everything but there are still some things we can look at.

• Our offense is still awesome. We had a little funk in the 2nd quarter but really you have your pick of impressive numbers to quote. 63% success rate on running plays, 75% of available yards gained, an average starting field position of our 40, no sacks. I mean this really was an impressive performance. I do have one complaint, why would we put Tim Byerly in? This isn't a stats discussion or anything, just my personal view. Tim Byerly represents zero improvement over Justin Thomas in any facet of our offense. To take Justin Thomas out for any reason other than an injury or to give him a rest is depriving our offense of it's best player and it throws our team in to a funk. Don't over think this, let JT do his thing.
• Annnndddd our defense looks almost as bad as our offense. This game is tough to analyze correctly though. On one hand forcing 6 turnovers is extremely impressive and the defense absolutely deserves credit for their performance. Also since we had a huge lead for most of the game we probably took our bend-but-break-alot-anyway defense to an extreme and just played conservatively. But forcing 6 fumbles isn't quite a sustainable strategy for every game, so let's hope the defense some how turned a corner based off this game.

Win Probability Chart

What else did you expect?

 Offense Down Dist Spot Q WP* Play Description WPA GT 3 9 GT 21 1 .61 Justin Thomas pass to Charles Perkins for 79 yards, TD .+ 22 Pitt 2 6 Pitt 29 1 .56 Chad Voytik rush for 9 yards, Fumble - .11 GT 1 10 Pitt 38 1 .55 Justin Thomas rush for 33 yards + .09 Pitt 2 10 Pitt 21 1 .84 Isaac Bennett rush for -7 yards, Fumble - .08 Pitt 3 1 GT 4 2 .07 James Conner rush for 4 yards, Touchdown + .05

Drive Chart

Again, the only blemish is really the funk in the 2nd quarter. And for those curious an average kicker would be expected to make a 37-yard field goal 70% of the time.

So what can we learn from a game where we were given a 28 point lead after five minutes? Well I'm really not sure. Our offense did a great job taking advantage of the mistakes Pitt made. Our defense only allowed 4 touchdowns, and that is certainly an improvement. But does anybody actually feel more confident after this game? I'm not sure I do. For the sake of our overall record it is great to get that 6th win. Now 8-4 is more likely than 7-5, which to me is a pretty big deal. At the end of the day, Bowl Eligibility tastes rreeaaalll nice.