Fri Oct 16, 2015 3:18 am
Just to comment on John's blog entry.
A clutch single is almost as worth as a homerun. If a team has a combined, say, -30, then it's the equivalent for that team of losing roughly 27 homeruns.
In linear weights formula, a homerun is worth around 1.4 runs...1.44 in the NERP formula that Dean uses. Typically, you can interpret this as meaning that, on average, a team will generate 1.4 runs if a homerun is hit.
Linear weights, however, do not consider context---hence, it's not a good way to predict the value of clutch singles.
Expressed otherwise, in linear weights, singles are worth 0.47 runs, but singles are not all equal: some singles have greater chances to generate runs than others. A single with two outs and nobody on will generally not lead to any run. But a clutch single ALWAYS generate runs---at least one, sometimes more, and when one does the whole calculation, based on 24-base-out probability matrix, you get that a clutch rating is typically worth 1.3 runs.
This said, the clutch single is activated only in 12% of all situations---with 2 outs and men in scoring position. So the value of a +1 clutch rating is worth 1,3*12% = 0.16 runs per 108 PA. So clutch has less value than singles (0.47 runs) or walks (0.34 runs). But in my opinion, it cannot be ignored easily.
Since clutch rating has roughly a third value of a single, a simple rule of thumb is to remove a single (hence a chance of on-base and a chance of total base) for every 3 negative clutch rating. For example, Suzuki's most recent card has vs rhp 31 chances of on-base, 33.5 total base and a -15 clutch rating. By using the rule of thumb, you could assume his card to be equivalent of a card with 26 chances of on-base and 28.5 chances of total base and a neutral clutch rating.