The 18th Out, Part 3: Head to Head, Face to Face, Butt to Butt

A howdy, a hey and a hi-yo Silver to all the hot-rod rowdy Razzball readers in deep anticipation of this; The third installment of my series on the oft spat upon Quality Start (Part 1 and Part 2).  I’m so glad to be back at the grindstone so soon; Things around here are getting better every day. Recuperation from a neck surgery is progressing daily, and little John is sleeping a little longer every night. In fact, I’ve sat down to start this article three different times already; Each time ending up down a different rabbit hole of QS stats which set me on a productive, yet different path than intended.  So for the delight of the crowd (and the detriment of the nerve endings in my fingers) the H2H part of our QS exploration will become a series within a series.  Just as a note going forward; Always keep in mind that point league formats can vary greatly. I will be using the format from my own CBS Home League which is only slightly varied from the standard: +0.5 per out, +1.0 per K, -0.5 per runner, -1.0 per run, +7.0 per W, +5.0 per QS, -5.0 per L. We wanted to make sure that in the event of taking a Quality Start + Loss, (which we call ”eating the cock-meat sandwich”) that the QS negates the Loss. All leagues are different so make sure to adjust for your own format as we progress.

When we left off last, we wrapped up our probe, failing to successfully argue for the relevance of the Quality Start in the Rotisserie baseball oven of life. We concluded that, (as they would say down south) ”That dog don’t hunt”. We found that neither the run average accomplishment (4.50ERA) nor the relative frequency of its occurrence (41+%) merited the disproportionate reward generally granted in H2H leagues.  We also found that aside from elite Starting Pitchers getting Quality Starts as a function of their skill, there is no worthwhile correlation to be made between achieving Quality Starts and Roto finish rank by position (I will actually quantify this and compare it to its counterpart later in this piece).

The distance element of the Quality Start stat failed as well, very much resembling our prior comparison. Elite pitchers go further into games than the average bum, and any real correlation ends there; The stress on Roto being firmly placed on the quality, and not quantity of innings pitched. For example, a 5.2 IP outing with 2 ER is significantly better in Roto than a 6 IP outing with 3 ER. In points league, the complete opposite is true; That single out to get into the 6th inning is worth surrendering the run. In this case it yields a 4-point net difference, representing a 20-30% boost in that single outing’s score, and adds approximately 3% to the entire weekly score. THAT IS  MASSIVE AND CAN NOT BE UNDERSTATED! Multiply that fact across the six to nine starts every week and you have the difference between winning and losing not only weeks but that juicy bonus cash for mo$t point$, and of course the trophy.  At the end of the day fantasy baseball is about more than analytics or statistics or adrenaline or psychologically abusing your closest friends; It’s about closing the deal and bringing home the paper. Diapers are expensive, and so are sports cars.

All of that being said, I will approach our exercise today a little differently dealing with weekly QS Point Leagues.  For Roto formats, the undertaking sought to discover if we could find value in chasing Quality Starts.  We could not. For Points we know we can, so our primary objective will be to find out just how much. This train of thought was the first rabbit hole (Oooooh I hate that rabbit). I thought to myself (SELF!) that if I lined up the Quality Start leaderboard with the 2018 SP overall point leaderboard, I could verify the assertion by actually quantifying and calculating this correlation. Then we could do the same for the top of Roto leaderboard for a definitive comparison.

We usually use correlation coefficients (a value between -1 and 1) to display how strongly two variables are related to each other. A correlation coefficient of +1 indicates a perfect positive correlation, which means that as variable X increases, variable Y increases and while variable X decreases, variable Y decreases.  On the other hand, a correlation coefficient of -1 indicates a perfect negative correlation. As variable X increases, variable Z decreases and as variable X decreases, variable Z increases. It’s important to remember here we are looking for negative correlation; As a Pitcher’s number of Quality Starts decrease, his overall rank by point total will increase.

In statistics, the correlation coefficient r measures the strength and direction of a linear relationship between two variables. The value of r is always between +1 and –1. To interpret its value, see which of the following values your correlation r is closest to:

• Exactly –1. A perfect downhill (negative) linear relationship

• –0.70. A strong downhill (negative) linear relationship

• –0.50. A moderate downhill (negative) relationship

• –0.30. A weak downhill (negative) linear relationship

• 0. No linear relationship

Let’s begin with the first real benchmark as far as total Quality Starts, 20. Only 15 Starters were able to achieve this feat in 2018.  We’re going to run our correlation formula and just from a glance I believe the stark contrast between our two favorite formats will present itself immediately.  Here’s a look at the leaderboard of the 2018 columns juxtaposed to that Pitcher’s finish last year by position to give a visual perspective to some of this over-the-top nerdiness (Think back to Lincoln Hawk : ”What I do is I just try to take my hat and I turn it around, and it’s like a switch that goes on. And when the switch goes on, I feel like another person, I feel, I don’t know, I feel like a… like a truck. Oh Sly, how I miss you…)

• Rotisserie  –0.46.  This is between a weak and moderate linear relationship, and it’s within reason to argue that in combination with our previous work that we can safely bury the use of QS in Roto in the backyard with Fluffy and those 3 hamsters I replaced in secret.

• Points –0.73.  This is between a strong and perfect linear relationship, and it’s already beginning to validate our initial assertion. We will pursue this and see if the correlation continues, or even extends.

Let’s see if we can broaden our sample before we go to back our regularly scheduled programming, using Starting Pitchers in 2018 that had at least 6 Quality Starts (132 qualifying players) and compare that to Point finish by position rank.  WOWZAH! Not only has the correlation been maintained, but we find that it actually strengthens! There is a -0.830 negative correlation between these two variables (+13.6% increase) as we expanded. Even having played QS H2H for years I find this staggering and well beyond what I thought we would discover when embarking on this research.  You can already see why this foray has to be split into sections. The next article will jump right into how to interpret this correlative data, on our way to developing our strategic advantage & draft strategies.  Having dissected the very top of the leaderboard was extremely productive, but eventually we’re going to have to expand what we’ve learned. In coming articles I’ll provide a detailed look into home/road splits to identify possible player pairs, or pitchers that have value in lopsided splits whose mediocre cumulative totals scare off lesser informed fantasy players.

That should do it for Part 3 of The Eighteenth Out. I hope everyone is enjoying reading this as much as I enjoy putting it out there. I love the reader comments! Keep them coming! And for any other questions or free access to galleries of my custom charts & graphs please follow me on Twitter @MLBMovingAvg.