How Much to Weigh 2020: Hitters Edition

The 2021 draft season may be the most unpredictable one … ever. While several rule changes (e.g. NL DH) can cause some evaluation inconsistencies, the biggest monkey wrench to deal with will be the lack of previous season stats. Projection systems will weight the 60-game 2020 season (37% of a normal season) to historical averages but what are those who don’t use traditional projections systems? Or read or listen to people who don’t care for them in order to create a compelling narrative? It’s time to anchor some historical context around those narratives.

All analyses “should” start with at least some set of weights. For those who start with one, they’ll have an advantage. For those who guess the correct one, they’ll have a huge edge. The issue with setting weights is how to create a short season one that fills all the narratives.

First, most of the players were rushed to get game-ready and were not 100% ready when the season started. So the beginning of the season stats could be used to replicate this ramping up period, but the weather was still warm at the season’s end increasing offensive output. Maybe the entire first month should be ignored with everything in flux. Possibly the last two because that were the 2020 regular season months. played. So, I decided to look into all the possible options including a horrible first attempt.

While I ended up taking two-month samples to create the rest of the analysis, I thought I’d be smart and just use hitters who had two normal (>500 PA) and one limited season (between 185 PA and 254 PA). What I found were a bunch of broken-down hitters. These hitters had limited playing time because of an injury and never recovered. Here are the averages for several stats from the broken down group and hitters with 180 PA combined in an August and September sample.

The injured group saw a near 80-point drop in OPS with no rebound. The Aug/Sep group just saw a steady age-induced decline with a total of ~40 OPS points. I knew I needed to use two-month samples. I created three.

• April and May: Includes the beginning of season ramp up time.
• April and September: The first and last months with a four-month gap in-between.
• August and September: Close to the 2020 months played (similar temperature) and proximity to next season.

Also, I ran the weights on just September because some fantasy managers discuss the noisy first month of data. The theory is that some players were still adjusting to the situation.

Now, take a moment and figure out the time frame that should be the best sample. Try not to start with the following results and work backward with a biased mind.

Here are the weights for OPS, AVG, HR/PA, and SB/PA. I know there are a ton more stats to consider, but I’m starting the discussion with an overall talent metric and the talent-based roto categories.

Here a quick look at each category

• Overall: The closer the short-season stats are to the next, the more weight they have. The Apr-May data has the same weight as just Sept. While the Aug-Sept sample is closest to the next season.
• OPS: Only one sample, the Aug-Sep, weighs the recent data more than that of two seasons ago.
• AVG: It takes so long for a player’s AVG to stabilize and the 2020 season should almost be thrown out. In all instances, a hitter’s AVG from two seasons ago has more weight.
• HR/PA: In a couple of instances, the recent data outweighs that from two seasons ago.
• SB/PA: Some deviation from the previous stats shows up. In the Aug-Sep sample, the last season (i.e. 2020) needs to be weighed more than the other two.

After seeing the results, I’m sure those who may use a weighting will still pick one that fulfills their bias. Oh well, to each their own. Personally, I’m torn between the Aug-Sep and just the Sept samples. For reference, maybe I’ll split the difference between the two.

The key from this study is to focus on 2019 when preparing for 2021. Now, there could be exceptions and the whole offseason is in front of us to prove or make up those exceptions. I bet (and hope for the sake of my backroll) the narratives win out.