WAR Is A Flawed Stat

If you have been paying attention to baseball over the last a couple years, you probably have come across the statistic, WAR.  In the ever ending search to find a player's value or impact on a team, the sabermetric people came up with WAR. WAR stands for Wins Above Replacement. It basically shows the worth of one player over a replacement player(usually a call up from the minor leagues). For example, if a player has a WAR of a +6, it means he will give the team six more wins over course of the year while he is in the lineup.
WAR takes in account the players all around impact of the game- hitting, fielding(not only fielding %, but also gives a guessed value to range of balls played in field) and baserunning(stolen bases is included but also advancing extra bases on hits also). Plus a different formula for pitching.  There is no one defined formula for WAR and different sources uses slightly different formulas. Here is the calculations and some more background information on it: http://en.wikipedia.org/wiki/Wins_above_replacement.

The sabermetric people uses a total analytical approach when using this stat. It does not consider how clutch a player is, or leadership qualities, a team's actual Win-loss record as a part of a player's value. For example, it doesn't look at if a player comes through in late innings down the stretch of pennant stretch that allows a team to win. Basically a game or plate appearance in April is just as important as in September and a player on a last place team should be equally compared to one on a divisional winner.  WAR also ignores the RBI stat as the proponents of WAR believe that the batter has no control who is on base when he is up and should have no bearing on his value. 

The above is the problem for me for using WAR as the defined stat to show a player's worth or case for an award like an MVP.  Games and seasons are decided by players coming through in high pressure situations and knocking in runs with men on bases.  I will give you here a game scenario to show you the flaw in the statistic. Yes, one game is a very small sample size, but it proves my point.

Let's say the Angels and Tigers have come down to the last game to decide who goes to playoffs and who goes home. The individual MVP race is too close to call between Mike Trout and Miguel Cabrera. The voters are forgoing their picks until the regular season is all over. The last game is a matchup between aces in Justin Verlander and Jered Weaver. Because of this runs are at a premium.
Mike Trout leads off the game with a single, steals second and advances to third on out. He is left stranded. Game remains scoreless and Trout comes up with two men and and two out in the seventh inning.  He fails to score the runners as he strikes out. In the eighth, Miguel Cabrera who has not gotten a hit in the game so far, comes up with a runner on 3rd with less than two outs. He knows this run can be the game and season. He also knows just a fly ball to the outfield could score the run. He does just that and Tigers hold on to win the game,1-0.
So who had the better game? Who should win the MVP? According to WAR, it's Trout. He had the hit and stolen base even though he failed to knock in the runs when his team needed and while Cabrera had no known value to the game. But to the people watching, it was the Tigers winning and Cabrera was a big reason for it.

The above illustrated to you that can not put a number on the human side of the game. Baseball is still played by humans and not by names and statistics on a piece of paper. In order to see a player's true value, someone still has to sit down a watch him play.  You can't just plug numbers into a formula and spit out a name of a player who has the greatest WAR and award him a MVP award.
After saying all of this, I still think WAR has a place in the game. It is a great tool used to decide between two potential free agents for a general manager to consider, for an example.  But, just like any other statistic in the game of baseball, it has it's flaws. He should be used alongside other stats and not the end all be all statistic.