Tuesday, January 19, 2016

Green Bay should have gone for two

In last Sunday's football match, the Green Bay Packers scored a late touchdown, making the score 20-19.  They decide to try to kick an extra point (giving them a tie and the chance to win the game in overtime) instead of trying to gain two yards (giving them two points and an outright win).

The NY Times (and me) think this was a mistake:

Teams have made around 48 percent of their two-point tries in the last three seasons. If you factor in that teams made around 94 percent of their extra-point attempts this season; that home teams tend to win overtime games (by a slight margin); that favored teams tend to have an advantage the longer things play out (Green Bay was a 7-point underdog); the math is on the side of boldness.

Extra Point:
Probability of a win=Prob[kicking extra point]*Prob[winning overtime]=94%*45%=42.3%

Two Points:
Probability of a win=48%

NOTE:  I do not know how frequently away teams win overtime games, but assumed that it is 45%.  The article suggests only some number less than 50%.

6 comments:

  1. I also agree that the Packers should have gone for two points because the probability was in their favor but probability statistics seem to have been put aside the further the game went along. The reason for that of course, was the unpredictability of the game itself, not to mention the human element.

    For example:

    With his season on the line, Aaron Rodgers converted a fourth-and-20 play with a 60-yard heave to Jeff Janis (their wide receiver). An illegal motion put Green Bay at first-and-15 with 12 seconds left at the Cardinals' 41-yard line, setting up one final Hail Mary throw. Rodgers nailed Janis on the final play of regulation with a 41-yard touchdown pass, sending the game to overtime.

    Of the 403 Hail Mary situations from 2005 through 2013, only 10 (2.5%) resulted in TDs. Seven other receptions were stopped inside the 10 yard line. Which makes what Rodgers was able to do that much more amazing but even with Hail Mary success , the Packers could not disregard their red-zone play up until that point.

    Now here is where the human element comes in, in regard to red zone play which would have certainly influenced their decision making.

    Aaron Rodgers got his team down into field goal range, but he threw an interception to Patrick Peterson on the goal line that was returned 100 yards for an apparent pick-six. However, the interception got taken off the board due to a defensive penalty on Arizona.Green Bay's drive was kept alive, but it couldn't find the end zone and had to settle for a field goal.

    The Packers' defense forced another quick stop from Arizona, giving Rodgers the ball back at his own 13-yard line. He proceeded to march 81 yards down the field on 16 plays, but once again Green Bay couldn't score a touchdown and went for another field goal. The Cardinals went into halftime with a 7-6 lead.
    Green Bay finally cashed in on a red zone opportunity, with Rodgers hitting Jeff Janis for an 8-yard touchdown to put the Packers ahead, 13-7.

    After the Packers punted the ball away, Arizona struck back with a 22-yard pass to Fitzgerald, putting the ball past midfield. Palmer had a chance to take the lead back, but he threw his second interception of the night on an ill-advised pass to the end zone that was picked off. Again, the Packers couldn't take advantage of the turnover and punted once again.

    Green Bay got the ball back with 3:44 left, but went three-and-out. Rather than punt the ball back, the Packers elected to go for it on fourth down at their own 25-yard line.

    James Jones couldn't hold on to Rodgers' throw, giving the ball back to Arizona.
    In my opinion, the Packers should never have placed themselves in a position to examine the probability of a 2pt conversion..

    References:
    http://blog.cougarstats.com/2015/09/10/hail-marys-just-how-improbable-are-they/

    ReplyDelete
  2. In sports, it is all about probability because there are choices to be made and the Packers vs. Cardinals game was no different whether going for two or the overtime coin toss selection can be evaluated and broken down into probability statistics. The Hail Mary play by Aaron Rodgers is one that almost never works yet he had success twice this season and extra points being moved back have decreased the extra Point Success rate. These are just a few examples at probabilities that sometimes get thrown out the window when it comes time to play a game.

    Watching the Game it is easy to understand why the Packers would go for the extra point and overtime rather than going for 2 at such a critical part of the game/season. While the thought of only having to go 2 yards for the two point conversion seems like a logical reasoning why it should be given some thought, the other side of the argument is that by going for two and not making it essentially ends the season and a chance to go to the Super Bowl. Although the probability to go for two points and win the game was 48%, the probability of being second guessed for not making it if they did go for two would be 100%.

    I am glad that this topic was presented because as a sports fan, it goes along well with the analytic's debate that has become largely discussed in Baseball especially between the old school mentality and new school analytic's view points are put up against each other. The Oakland Athletics have largely been known as the baseball analytic's leader and originators. The gap and discrepancies between baseball lifers and the new “analytical geeks” as they are often called make it a hot topic for debate. As of recently, a baseball front office analytic's guru Paul Depodesta who worked for the Mets recently took a job with the Cleveland Browns. Normally this would not seem unusual, however hiring a front office executive with no football experience especially in a sport like football which rarely uses analytic's to put together teams and choose players makes this both an odd and unique hire. Could this hiring raise future speculations and calculations of the likes of which happened in the Packers vs. Cardinals game and further present the argument of going for two in the NFL among other situations while ultimately bringing about a new experience and way to go about the game?

    Time will tell but it is certainly going to open the door for the debate to further escalate if Depodesta and the Cleveland Browns are able to have success.

    ReplyDelete
  3. Unfortunately I don’t watch football or any sports for that matter, but I do understand where the author is coming from. I have no choice, but to compare this to when I used to watch baseball. I believe it was the World Series of 2001 when the Yankees were a few players from winning the championship against Arizona Diamondbacks and the bases were loaded. The Yankees manager at the time, Joe Torre, had to either bring in the infield players in to guard against a potential bunt or keep them infield the same in case of another hit (potentially a faster one), a land drive. Most managers in the aforementioned situation end up bringing in the infield in and as such the Yankees paid dearly and lost game 7 with a score of 3-2 (OLiney, 2001).

    I do understand that numbers wise it makes the most sense to go with what statistically speaking has proven to be more effective, but what about instincts or that sixth sense that tells you to do otherwise? Can we truly apply psychology in economics or in a numbers game? According to A.B London, “However, there are situations where people would do better by slowing down. And there are cases in which people have far more confidence in their intuitions than is justified”. In other words, in the midst of a game is very likely that the Packer’s head coach made his decision to attempt to try and tie the game based on a “fast thinking” process. If the manager may have made his decision on slowing down and going with his intuition he may chosen to attempt and try going for the win. Of course, I may wrong and if he would attempted to win the game he would have faced the backlash of the media.

    ReplyDelete
  4. Didn't know it cannot be edited - forgot my sources:


    A. L. (2012, February 05). Can we ever trust instinct? Retrieved January 25, 2016, from http://www.economist.com/blogs/prospero/2012/02/quick-study-daniel-kahneman-economic-decision-making


    Olney, B. (2001, November 05). In Final Twist, New York Falls in Ninth. Retrieved January 25, 2016, from http://www.nytimes.com/2001/11/05/sports/world-series-in-final-twist-new-york-falls-in-ninth.html

    ReplyDelete
  5. Making a decision of tying the game or winning comes down to confidence in your team’s ability to defend accordingly. If, by taking your example, we assume that a team is down by eight points with seven minutes left in the game, the decision should be made based on your ability to defend and score again before time expires. If the team decides to go for the extra point, assuming it is good every time, the win probability is 42 percent. If you make the two-point conversion, you're tied, and the win probability is now 47 percent.

    “Given that about 45 percent of all tries are good, we have to see what would happen with the other 55 percent. Your win probability if you miss the conversion is now 37 percent. Thus, we can say that WP(attempting the two) = (0.45 * 0.47) + (0.55 * 0.37) = 0.2115 + 0.2035 = 41.5 percent. So, you have a 42 percent chance of winning after making the extra point, and a 41.5 percent chance of winning if you attempt a two-point conversion. The probabilities are almost the same, so in this case, attempting the extra point or two-point try is dependent on how your defense has performed in the game.” (Fein, 2009). If the time left in the game changes from 7 minutes to 5, the probability drastically changes to 40.7 percent with a two-point conversion, or 32 percent with the extra point.

    Since the probability is virtually the same with seven minutes left, this becomes the break-even point.

    As stated in the movie “The Replacements”, “Winners always want the ball… when the game is on the line.”

    References
    Fein, Z. (2009, January 17). NFL Win Probabilities: Extra Point or Go for Two? . Retrieved from Bleacher Report: http://bleacherreport.com/articles/112292-nfl-win-probabilities-extra-point-or-go-for-two

    ReplyDelete
  6. I think Green Bay Packers made a good decision kicking for extra point instead of going for two points. Yes, based on the probability calculation it seems that going two point conversion is a better option; however there many elements that were not accounted for. When a game that tight, athletes tend to get extra motivated to play their position. Especially if the result determines the fate of the team between elimination and moving up. I played sports in high school (not football). If I was in the position to make the call I would kick for extra point and go to overtime to win it. Here is my reasoning. It is towards end of the game, few seconds left. My chance of scoring a two point conversion is lower than kicking an extra point. I would make that assessment based on my confidence in my team capability as well as opponent’s defense capability. Let’s say I decided to go for two points. I can run the ball and potentially get blocked or fumble to lose the ball. I can go for passing game but chances are my receivers will be heavily guarded. Since it will be a short are to play, it would be easier for the opponent’s defense to cover the ground effectively. Another factor is I will get one shot. If I do not score the team will lose the game. To me the risk is higher. Now if I kick for extra point and go to overtime, I have more than one chance to score and win the game. At over time I will be able to utilize both my offense and defense team to contribute to the winning result. I will have time to strategies. The risk is lower. The NY Times articles talked about “Teams having made around 48% percent of their two points tries. Team made 94% extra points. The home teal tend to win over times.” To me all these information are statistics about previous games. Most athletes will not make calls in the field based on what happened in previous game. Each game and scenario is unique. The probability calculation may work out in papers towards going for two points but it is not reality.

    ReplyDelete