Hot and Cold, August 6

Form by the rates: 16 heating up, 16 cooling off across hitters and pitchers.

Form, measured as a rate rather than a total. Comparing the week of July 27 to August 2, 2026 against the whole of July 1 to July 31, 2026, per plate appearance for hitters and per inning for pitchers, because a player who simply started more games last week has not changed, he has just played.

Hitters

Heating up. The top tenth of the group by change in points per plate appearance, a bar of +0.36 set by this run's own spread rather than by a fixed number.

Cooling off. The bottom tenth, at -0.45 or worse.

Pitchers

Heating up. The top tenth of the group by change in points per inning, a bar of +1.55 set by this run's own spread rather than by a fixed number.

Cooling off. The bottom tenth, at -1.65 or worse.

What this means

Start with Foster Griffin, because he is the one line in the table that should stop you. A week at minus 1.31 points per inning is not a bad start, it is a start that actively cost you, and it came off a month at 1.81. Yet the forward rank reads 57. Nothing in this data explains that gap, and I would rather say so than invent a reason: we can see the rate collapsed and we can see the engine did not care, but the driver is not visible here. What we can say is that a projection holding a top-60 arm through a minus week is telling you the model's read is built on something the last seven days did not touch. The strongest case against trusting it is simple. A pitcher ranked 57 is a pitcher you are starting most weeks, and if the engine is anchored to a role or a workload that has quietly changed, you eat the losses one start at a time. The falsifier is the next outing: if he goes short again and the rate stays negative, the rank is stale and you cut bait. Until then, hold and start him.

Sanchez and Webb are the opposite case, and they are the easy ones. Sanchez jumped from 0.78 to 4.00 points per inning and sits at 23. Webb went 0.73 to 2.33 and sits at 126. Neither number moved because of the week. These are ranks that were already there, and the hot week is the projection catching up rather than the other way around. Nobody in your league is selling either one, so there is no action here beyond starting them.

Luzardo is where I would actually take a position. He fell from 3.03 to 0.38, a drop of 2.66, and the forward rank is 44. That is the engine saying one bad week inside a window this short cannot move a season's worth of evidence. Wheeler makes the same point louder: down 1.69 and still ranked 8th. If a manager in your league is spooked by Luzardo's week and will take a discount, buy. The cost of being wrong is real, though, so be honest about it. The two windows overlap by five days, which means the drop you see understates how bad the last seven days were. If the next two starts land in the same place, this stops being a cold week and becomes a change in what he is.

The batters are quieter and mostly noise. Hoskins went from 0.37 to 1.67 and the engine still has him 649th. That is not a projection ignoring a hot week, it is a projection saying the week was a handful of plate appearances and the underlying player is unchanged. Same shape for Manzardo at 604 after a jump of 0.74. Rate stats over a week are almost all sample.

The one worth a second look is Del Castillo. He fell 0.75, from 0.50 to 0.25 below zero, and the rank is 312, the best of the cooling batters. A catcher who was producing at 0.50 and is now negative is either sitting more or hitting into bad luck, and the data here cannot tell you which. That is the observable: if he is still getting four or five starts a week ten days from now, the cold spell was noise and 312 is right. If the starts dry up, the rank is wrong and you drop him. Manzardo and Warren, both cheap, are the speculative adds if you need the roster spot to do something.

How to read this

Everything here is a rate. Points per plate appearance and points per inning remove the schedule from the comparison, which matters more than it sounds: the single most common reason a player "broke out" last week is that he started six games instead of three.

The two windows overlap, the week sits inside the month, so every player is partly being compared against himself. That pulls each number toward zero, which means real form changes are understated here and none of them are manufactured by the method.

The cutoffs are the tenth and ninetieth percentiles of the group being shown, recomputed every time this runs. A pitcher and a hitter are never held to the same bar, because their spreads are not the same shape.

A player needs 5 plate appearances or innings in the week and 20 in the month to appear. A rate built on two at-bats is not form, it is arithmetic.

All articles · Redraft rankings · Dynasty rankings