Archive for Meta Analysis

I was wrong. About seven years ago, I wrote on how hitters may overperform their projections since they played through an injury. The injury hampered their production in the season in question, lowered the future projection, and created a buying opportunity. For years, I believed this steadily until last season when I re-ran the numbers and found “jack squat”.

Earlier this week, I examined some of this past season’s hitters who fought through the pain and felt a deeper analysis was needed. I dove in and the results were backwards. I found no bounceback should be expected from hitters who played through injuries, but there is more. For those hitters who play through the discomfort, their future production will take a major hit.

The key to uncovering the following results was getting a usable dataset which is easier said than done. Many of the injuries I’m using for the analysis aren’t well documented, if at all. Real men play baseball and they play hurt because that is what real men do and most importantly, they don’t complain about. Besides the machismo, a player has every right to keep his medical data to himself so vagueness thrives. Simply, there is no good available data. Even with the hurdles, I dug into each of the hitters who were reported to have played through an injury the past three seasons (2017, 2018, 2019).

Read the rest of this entry »

On Monday, I laid out my thought process for selecting Mike Trout first overall in 2020. To summarize, we have several legitimate choices. Absent secret information, they’re all a shrug away from the same, albeit in slightly different shapes. Trout stands out because he’s been so consistently good for so long. The alternatives all have shorter track records or a chance to fall out of the top 30 players – even without an injury. By comparison, we know a healthy Trout is a monster fish.

Today we’ll take a peek at all the players you could reasonably select over Trout on draft day. Let’s toss them into a few buckets. ADPs are from NFBC.

Read the rest of this entry »

I’ve had a semi-fixation on hitters playing through injuries and how the diminished production could hamper the next season’s projection. At first, I found some correlation. Then, I didn’t. One possible answer to there being no bounceback is that the injury becomes chronic and the hitter never improves. Or the dataset could be too small.

I want to dive further into the subject, but the information around injuries is sketchy at best. Most of the time, there are no usable details. The lack of an answer means that I should stop coming back to the subject but I’m stubborn.

Very.

I’m going to go through this past season’s hitters. The dive has a couple of goals. One is to create a better dataset for future reference. The second is to understand why some hitters may have struggled when creating a profile. And just maybe, I’ll find out if I can put to rest the notion that hitters who played through injuries are under projected.
Read the rest of this entry »

There’s a debate going around Twitter. If you want some cheap likes and retweets, try dropping this poll:

Who would you pick first overall?

Mike Trout

Christian Yelich

Ronald Acuna

Other

Read the rest of this entry »

Last year, I took a stab at developing what might be loosely defined as park factors using Statcast data. (I called them park “impacts” because they lacked the requisite rigor to be true factors, although it’s all semantics, truly.) I sought to use Statcast’s expected wOBA (xwOBA) metric, specifically on batted ball events (BBEs), such that we would have a measure of xwOBA on contact (or xwOBAcon). This metric accounts for exit velocity (EV), launch angle (LA), and little else — which makes it perfect for this purpose.

The difference between actual and expected wOBA on contact indicates the amount of luck, whether good or bad, a hitter might have incurred on a particular batted ball event. In other words, given ‘X’ exit velocity and ‘Y’ launch angle, what is the most common wOBA outcome, and how much did the actual wOBA outcome differ from it?

The beautiful part about xwOBAcon is it strips away all other context. It removes elements that confound other park factor calculations, such as hitter and pitcher quality or even sequencing (vis-à-vis run-scoring). Except for fielding. Can’t control for fielding, unfortunately.

With this approach, we have the exit velocity. We have the launch angle. We have historical results for that particular combination of EV and LA to use as a benchmark. And then we compare.

Read the rest of this entry »

Over the years, I’ve been working on how to fine-tune my player evaluation process. The following are six datasets that I’ve found useful I’ll not go into detail on any of them since I provide a link to the original article. The following is basically a referenceable data dump.

Note: I know there is a lot of content and when questions arise, make sure the area in question is obvious in the comment. Also, I’ll only answer questions here and not in the original articles.

Voit/Muncy All-Stars (link)

These are older AAA hitters who have shown signs of a breakout.

Read the rest of this entry »

Gred Gigerenzer (yes that’s his real name) has been a leading advocate on how to correctly measure and articulate risk. I’d highly recommend his book Risk Savvy: How to Make Good Decisions, but today, I’m going to focus on some passages from another book of his, Calculated Risk, which focuses on risks in the medical profession. Some of the passages seem to resonate with me about the fantasy expert community, especially this question: what should be the intent and expectations of touts?

One point Gigerenzer hopes to get across is the difference between Risk and Uncertainty. For him, Risk is measurable such as pitcher X as a 40% chance of going on the IL based on his age and past injury history. Uncertainty involves values that can’t be (or aren’t) measured like player Y is going through a divorce so his production is down.
Read the rest of this entry »

Edit: Further investigation has brought to my attention that the results presented below are slightly askew, although not incorrect. All discussion below regarding hit frequency (BABIP) and contact quality (expected wOBA on contact, or xwOBAcon) should have been framed specifically in the context of non-home run batted ball events. This is significant, because home runs are a big deal, but it’s also insignificant. Allow me to explain.

When we re-include home runs, the relationship between launch angle tightness (stdev[LA]) and contact quality weakens dramatically. I think it comes down to the graph shown in the middle of the post below. Removing home runs narrows the range of productive launch angles, thus making a tighter range of launch angles (confined primarily to line drives) more appealing. When you include home runs, it expands the range of productive launch angles to include productive fly balls in addition to productive line drives. There’s literally more margin for error when we reconsider home runs, making a tighter range of launch angles was valuable.

That doesn’t mean launch angle tightness isn’t important! If anything, removing home runs was a nifty way to demonstrate this fact.

Anyway, I have updated this post with red text to clarify that references to contact quality exclude home runs — and that the findings from this post are technically correct, just through a certain lens.

* * *

Last week, I published some work regarding launch angle “tightness,” aka a hitter’s ability to replicate his average angle as closely as possible as often as possible. Effectively a measure of consistency, I found launch angle tightness (consistency, variance, whatever you want to call it) bore a moderately strong relationship with batting average on balls in play (BABIP).

Truth be told, I began to question my finding almost immediately for reasons I’ll discuss shortly. After inquiries from The Athletic’s Eno Sarris, FantasyPros/PitcherList’s Nick Gerli, and even Cody Asche (this is the mildest of brags) that echoed my internal self-doubting dialogue, I dove into the question further.

Ultimately, the best explanation for the importance of launch angle consistency is to simply elaborate upon launch angle generally. So, consider this a de facto primer on launch angle. It’s probably not the first and certainly won’t (or shouldn’t) be the last. But in the context of my post from last week, it simply makes sense to bring the conversation full circle and wrap it up nicely with a bow. And the final result is gratifying, I hope.

Enjoy (or not, I’m not your dad):

Read the rest of this entry »

Historically, I’ve “corrected” hitter projections to my own liking and every time I’ve backtested them to the actual results, my adjustments have failed miserably. So why create more work when the end results make my final product worse? Am I a glutton for punishment? In all fairness, I’m sure a heavy dose of Dunning-Kruger is going on but I also believe there may be a sweet spot where personal scouting can come into play. Today, I’m going back to the well one more time to see if some injured hitters should have more encouraging projections because they may have played hurt.

First, I’ve always thought playing through an injury meant that the team and the player were accepting suboptimal production. Then the player could come back healthy and full productive the next season.
Read the rest of this entry »

A few weeks back, I posted that the 2020 edition of The Process was available in e-book form for downloading. All the loops have been jumped and now all it is available in paperback form at Amazon.

Here are some of the additions:

• A comparison to see if it’s more efficient to buy closers versus starters in the draft or wait for free agency for each one.

Read the rest of this entry »