So you may have noticed that every now and then, a player’s valuation has changed on this site. You might be wondering, “Why?” or “Hey, does that mean I have to go back and redo that trade idea I had last week?” Or: “Is this arbitrary, or is there a good reason for it?”

All good questions. And yes, there are reasons for the changes. Four of them, typically:

1. New data. Our formulas rely on publicly available services, such as projection systems and prospect rankings, to inform them. So when one of those services updates, we flow in those updates to the appropriate columns in our spreadsheets behind the scenes, which then show up as updates on the site.
2. New player status information. This could be an injury, an injury worse than first reported, or a suspension, for example. It could also be a change of a playing time estimate, such as a when a fourth outfielder gets plugged in for regular duty because the starter got injured.
3. Midseason performance updates. These will occur monthly starting in June, and factor in that player’s performance that year, pro-rated against his preseason projection.
4. Transaction information. If a player gets traded, signed to a new contract, DFA’d, released, etc., we follow it closely, and adjust our numbers if necessary to reflect the new information. For example, if a trade occurs, and we find that our estimate was too high or too low, we will update the valuation after the fact so that it hews closer to reality going forward, because when the real-life market moves, it’s our most important data input. The more this happens, the smarter the model gets. However, we’ve learned to wait a day or two before doing this particular update, for transparency’s sake.

It’s important to keep in mind that we are not talent scouts, nor are we in the projection business. Our estimates should also not be considered predictions. We will be off sometimes. Real front offices are working with much more proprietary information than we have, and that information gap will cause some overs and unders here and there. That’s also why we publish a low, median, and high range. But one of our key goals is to make the model more accurate over time, and we believe that is occurring.

So when you see a number change, it’s for one of these reasons.