We did something new with today’s roster update and treated it like the updates we’ve typically only done between yearly releases. I thought I’d take some time to explain what you can expect and how it differs from our usual weekly updates.
Here’s a three things to note out the gate:
- Overall Rating is just a made-up number that exists to give a high-level idea of a player’s ability at a given position. There are unique formulas for each position that value certain skill sets differently depending on a player’s primary position. This is why a position change can cause a pretty substantial shift to a player’s Overall even though their attributes haven’t changed at all, like Ben Zobrist from 2B (86) to RF (81) earlier this year. The Overall Rating is not something we look very closely at when we are attempting to accurately represent our players.
- Attribute Ratings are what drive the simulation results in our game and set players apart from each other (Hits per 9, Contact vs Left, Fielding, Speed, etc). Unlike the Overall Rating, Attributes are HUGELY important to the way the game plays. They are NOT just a reward for a player performing well recently, they are the fabric of how results are decided when you make a pitch, swing the bat, track and dive for a ball, make a throw, and so on.
- Though we have a couple drastically different methods for constructing or reconstructing our players attributes, our foundation is a sabermetrics approach similar to what you might find in the yearly baseball prospectus. We love advanced stats, particularly the stats that are good for analyzing a player’s performance independent of the team or stadium he is working with.
Before Today… All of the previous weekly updates were the result of rigorous analyses from a couple of stat-head designers that manually identify players whose expected future production may have changed drastically. The designers then tweak any questionable Attributes by hand. Obviously this can be quite subjective, sometimes leading to some spirited debate within the studio when it comes to our cherished homers.
Today… Like I mentioned above, today’s update was an evolved version of what we do between yearly releases. It’s an algorithmic approach that assesses all players and almost every one of their attributes based on predictive statistics. What’s great about the algorithmic approach is that it covers a much, much larger pool of players and does so in great detail, considering only objective statistical data. Situations like positions changes or playing through injuries can be misleading out of context. For these, our stat-heads comb through and straighten them out. What’s unfortunate is that the first step in the automated rebuilding of our rosters is to overwrite the very specific manual changes our guys spent significant time on in past months, all of which was carried out while balancing development on the features they are working hard on for MLB ’16.