Before Saturday’s game against the Rays, Indians manager Terry Francona announced that John Axford will be taken out of the closer’s role and the team will employ a “closer by committee” for the foreseeable future. According to the skipper, any combination of Cody Allen, Scott Atchison, Mark Rzepczynski and Bryan Shaw will see time in the ninth.
The “save” stat has been rebuked by analysts in most baseball circles. That is because the game is usually either won or lost by the time the final inning rolls around. Although the Indians bullpen is deep again this season, the “closer by committee” theory only works when the team reaches the ninth with a lead. This is usually accomplished from innings six through eight.
During the course of a game, some situations are more tense and suspenseful than others. For instance, we know that a one-run lead in the bottom of the ninth inning is more suspenseful than a one-run lead in the top of the third inning. Batting with two runners on and two outs in the eighth inning is filled with more pressure than batting in the same situation in the second inning. Leverage Index (LI) is used by Fangraphs and is merely an attempt to quantify this pressure so we can determine if a player has been used primarily in high-leverage or low-leverage situations. LI is most relevant late in the game, when the reliever is typically brought into the game.
Since relievers pitch only partial games, and in particular because they often come in partway through an inning, ERA is a very poor way to evaluate their performance. Runs scored by runners already on base when they enter the game are charged not to the reliever but to the preceding pitcher. Yet one of the primary roles of an relief ace is to stop a rally– that is, to prevent inherited runners from scoring as well as to prevent additional runs.
A reliever may pitch to just one or two batters before being removed himself to gain a platoon advantage. That reliever may give up the runs to his successor not only his own runners, but also those belonging to the pitcher before him. LI does a fair job of valuing each game situation and assigning responsibility .
From this, it is simple to estimate the probability of a team’s winning from the current score, inning, number of outs, runners on base and relative strengths of each team. To measure the true value of a reliever, we can look at his team’s chance of winning when he enters the game, then again when he leaves the game or the game ends. For example, Allen enters the game with the Indians hanging on to a 2-1 lead in the seventh inning, with none out and the bases loaded. Despite the lead, the situation is high leverage and the probability of the Indians winning is only about 39%. Now suppose Allen gets two strikeouts and forces a groundout, finishing the inning with no runs scored. The probability of an Indians win has now risen to 79 %. Allen is removed from the game and is credited for 40% of a win, or 0.4 wins, regardless of the actual outcome of the game, because what happens from then on is not under his control.
However, that’s not entirely the same for all circumstances. A reliever who pitches in more than one inning will have his team come to bat for at least a half-inning. The change in probability could have resulted from his team scoring additional runs rather than the pitcher getting out of a jam or preventing runs from scoring.
When valuing relievers, leverage (LI) is the key. When the game is on the line, when one play can be the turning point of the game is when leverage is most important. The higher the leverage, the more the chances of winning the game depend on a good performance. Since relievers pitch only partial games, and in particular because they often come in partway through an inning, ERA is a very poor way to evaluate their performance.
From LI, analysts can put together a Win Expectancy (WE) framework as a way to measure how crucial a run is at any point in the game. For example, a leadoff homerun in a game between equally matched teams raises the probability of winning from 50% to 59.8%. A homerun to lead off the eighth in a tie game raises the probability of winning from 50% to 74%, One run is valued differently depending on the game situation.
Thus, the concept of LI, and more importantly, a tangible means to measure it, proves invaluable in understanding usage patterns.
According to the WE chart, the most important situations in a game often occur well before the ninth inning. For games whose highest-leveraged situation is above the mean for all games, the most critical situation is likely to occur in the ninth inning only 40% of the time. Fully 60% of all highly leveraged situations occur in innings six, seven and eight.
With this information in tact, it is important to note that Francona’s use of the “closer by committee” will be most effective if he still uses his best relievers before the ninth inning. As forward thinking as the Indians organization is, I am sure they understand this and will make the most of the “closer by committee.”