Spring Training is here, and Mr. Moving Averages is cranking out content like some kind of crazy content making machine.  Countless hours of research, critical thinking, data mining, analysis & writing all come to a head as we approach Opening Day. Draft season is upon us, even if there may be snow on the ground in Brooklyn as I write this. Our immersive work, now a laser guided missile, began without an initial definitive direction by simply opening a door. That lead to another door. And another, and another until we wound up down the rabbit hole, too far from home to turn back now.  I have a sneaking suspicion the trend of ignoring point leagues continues, and this next leg of our statistical expedition will end up with us frustrated at the lack of detailed stats for our purposes. So what do we do? Soldier on.  Follow the white rabbit. Reminds me of the old adage, the call to unification; ”Where we go one, we go all”.

Related image

Towards the end of Part 6 we identified starting pitchers who must be considered for win progression in the 2019 season. However, seeing those win percentages associated with Quality Starts really got me thinking to myself (SELF!). The numbers on the surface are accurate, but somehow unfulfilling.  It’s not enough. It’s lacking depth and context.  The broad definition of a QS is so flawed that if it were a house, I’d knock it down and rebuild from scratch. If any Quality Start, defined as a start of least 6 IP allowing no more than 3 ER garners a win roughly 55% of the time, what about other distance/earned run outcome combinations? Does a pitcher increase his chance at winning by going an extra inning? It would seem safe to assume so. Does a pitcher increase his chance at winning giving up less earned runs? That would also stand to reason. BUT BUT BUT does a pitcher gain a better probability at winning by going an extra inning while surrendering another run? What about two runs? As my readership, I consider you all friends. You’re all getting to know me by now. So I don’t have to tell you that once I see the numbers start to fly, I have every intention of bringing in order and analysis to calm the chaos.

Image result for math gif

  • Keep in mind this sample set below is the entirety of the 2018 game set, opposed to our previous work which only calculated starters with a minimum of 5 QS. Our sample of 5 QS min yielded a QSWC% of 55.19%, and this broader total drops the overall QSWC% to 54.76%; A very slight difference, and something to keep in the back of our mind as potentially useful in the future.  My gut tells me this drop, by including more exceptions to our bullpen + team win % model is going to further validate our use of this stat for regression & progression candidates in the future.

Let’s start tackling the data. There’s a lot to take from this information and personally, I’ve never seen it presented anywhere else in this fashion. There’s really no telling what this could set the table for next. Speaking of tables (who ordered the segue?), I’m going to provide four values for each Distance/Runs Outcome (We will now refer to this new metric as a pitcher’s D/RO). Keep this in mind, it’ll save us (mostly me) a ton of time. To be completely clear; For example a 6+/3 D/RO is 6+ IP 3 ER, the worst distance run outcome allowed before disqualification.

I’ll present the same tables a couple of different ways to help us codify this mountain of information for differing perspectives. I’ll add a quick table analysis for standouts in each set and tackle any useful subsequent relationships I identify at a glance. We may be back here for another taste but I don’t want to get too distracted before we move on to custom Home/Away Split QSWC% leaderboards, & the resulting specific late round player pairs/streamers.  There’s way too much here to start feasting, but a little nibble never hurt anyone. Remember this draws from 2018 results only.

  • T — Total number of games with that specific D/RO
  • T% — Number of games with that specific D/RO as a % of the entire 2018 game set
  • W — Total number of games won with that specific D/RO
  • W% — Number of games won as a % of T at that specific D/RO

 

6+ IP QS Outcomes

6+ IP 0 ER T 6+ IP 0 ER T% 6+ IP 0 ER W 6+IP 0 ER W%
514 10.57% 387 75.29%
6+ IP 1 ER T 6+ IP 1 ER T% 6+ IP 1 ER W 6+IP 1 ER W%
1186 24.39% 753 63.49%
6+ IP 2 ER T 6+ IP 2 ER T% 6+ IP 2 ER W 6+ IP 2 ER W%
1670 34.35% 980 58.68%
6+ IP 3 ER T 6+ IP 3 ER T% 6+ IP 3 ER W 6+ IP 3 ER W%
1996 41.05% 1093 54.76%
  • The most baseline QS, a D/RO of 6+/3 has a Quality Start Win Conversion % rate that sits just under 55%; The one thing we knew about QSWC% coming into this exploration.
  • A  D/RO of 6+/0 is not only rare, occurring only +10% of the time, but it yields an extremely high conversion rate of +75%. At the very least, this will have to be considered for a promotion to the highest tier when I dive into separating Quality Starts using some new custom stats I’ve been tinkering with involving weighted D/ROs (Part 9 teaser, with some exclusive custom content that will redefine our point pitching analysis).
  • There is a notable 11.80% drop in QSWC% after the first run is allowed through 6+ IP. No other single run had a greater effect on QSWC% in this entire exercise, as you may have already assumed; Just for reference, the next highest is 7.49%.
  • In the seventh inning (remember 6+ = 7th inning), QSWC% drops an average of 6.84% for each of the 3 surrendered runs in this DR/O set.

 

7+ IP QS Outcomes

7+ IP 0 ER T 7+ IP 0 ER T% 7+ IP 0 ER W 7+ IP 0 ER W%
274 5.64% 211 77.01%
7+ IP 1 ER T 7+ IP 1 ER T% 7+ IP 1 ER W 7+ IP 1 ER W%
561 11.54% 390 69.52%
7+ IP 2 ER T 7+ IP 2 ER T% 7+ IP 2 ER W 7+ IP 2 ER W%
736 15.14% 478 64.95%
7+ IP 3 ER T 7+ IP 3 ER T% 7+ IP 3 ER W 7+ IP 3 ER W%
828 17.03% 515 62.20%
  • Similar to the 7th inning (and in agreement with reasonable suspicion), the first surrendered run has the greatest impact on QSWC% (7.49%).
  • A starting pitcher even having surrendered 3 ER in the 8th only lowers QSWC% to 62.60%; +7% above the average; Distance matters. In coming segments we will revisit our custom distance metric from previous articles and factor in OPO (Outs Per Outing) with these varying DR/O to look for anything that catches our eye.
  • I was surprised by the infrequency of a +7/0 outcome to be honest. We now know it happened roughly once in every twenty starts in 2018. For a quick perspective, if we estimate a full season’s workload for a starter at 30 GS, then it only happens 1.5x per season per pitcher. I say this as the basis for my development of a more specified Quality Start stat count.  These rare starts with extremely high relative win conversion percentages should not be grouped and counted equally with the 41% of all starts that end up scored as a QS. Cultivating this is going to be extremely time consuming, but I really do feel in my heart of hearts that it will be our crown jewel; And possibly redefine these outcome classifications for an entire industry in the future.
  • In the 8th inning, QSWC% drops an average of 3.70% for each of the 3 surrendered runs in our DR/O.

8+ IP QS Outcomes

8+ IP 0 ER T 8+ IP 0 ER T 8+ IP 0 ER W 8+ IP 0 ER W%
71 1.46% 54 76.06%
8+ IP 1 ER T 8+ IP 1 ER T% 8+ IP 1 ER W 8+ IP 1 ER W%
120 2.47% 88 73.33%
8+ IP 2 ER T 8+ IP 2 ER T% 8+ IP 2 ER W 8+ IP 2 ER W%
151 3.11% 107 70.86%
8+ IP 3 ER T 8+ IP 3 ER T% 8+ IP 3 ER W 8+ IP 3 ER W%
166 3.41% 114 68.67%
  • The table mostly speaks for itself. Getting to any QS qualifying DR/O of 8+ is a rarity, happening no more than once in every thirty attempts. That’s an easy translation into once per season per pitcher; Another very strong argument here for a stratification of the scoring. If a single pitcher were to have fewer total QS, but more of these rare accomplishment QS, that’s something I would want to know.
  • Any QS qualifying DR/O of 8+ also greatly increases the rate of conversion. The worst case scenario of 3 ER still yields a win on average almost 20% more than the average QS! How can this start be compared to a 6+/3 start? My answer? It won’t be for long. Dividing these varying QS into calculated categories could change the way we rank pitchers (especially high end SPs) during the offseason.
  • In the 9th inning, QSWC% drops an average of 2.46% for each of the 3 surrendered runs in our DR/O.

 

3 Run QS Outcomes

6+ IP 3 ER T 6+ IP 3 ER T% 6+ IP 3 ER W 6+ IP 3 ER W%
1996 41.05% 1093 54.76%
7+ IP 3 ER T 7+ IP 3 ER T% 7+ IP 3 ER W 7+ IP 3 ER W%
828 17.03% 515 62.20%
8+ IP 3 ER T 8+ IP 3 ER T% 8+ IP 3 ER W 8+ IP 3 ER W%
166 3.41% 114 68.67%
  • A QS qualifying D/RO ending in 3 aided the cause for wins by adding 7.44% & 6.48% respectively for each extra frame completed (6.96% average).

 

2 Run QS Outcomes

6+ IP 2 ER T 6+ IP 2 ER T% 6+ IP 2 ER W 6+ IP 2 ER W%
1670 34.35% 980 58.68%
7+ IP 2 ER T 7+ IP 2 ER T% 7+ IP 2 ER W 7+ IP 2 ER W%
736 15.14% 478 64.95%
8+ IP 2 ER T 8+ IP 2 ER T% 8+ IP 2 ER W 8+ IP 2 ER W%
151 3.11% 107 70.86%
  • A QS qualifying D/RO ending in 2 aided the cause for wins by adding 6.26% & 5.92% respectively for each extra frame completed (6.96% average).
  • Even with only 2 runs allowed, a starter really has to get through the 7th inning to have a notable effect on their QSWC%

 

1 Run QS Outcomes

6+ IP  1ER T 6+ IP 1 ER T% 6+ IP 1 ER W 6+ IP 1 ER W%
1186 24.39% 753 63.49%
7+ IP 1 ER T 7+ IP 1 ER T% 7+ IP 1 ER W 7+ IP 1 ER W%
561 11.54% 390 69.52%
8+ IP 1 ER T 8+ IP 1 ER T% 8+ IP 1 ER W 8+ IP 1 ER W%
120 2.47% 88 73.33%
  • A QS qualifying D/RO ending in 1 aided the cause for wins by adding 6.03% & 3.81% respectively for each extra frame completed (4.92% average).
  • The roughly 6% increase in win conversion with each additional frame pops up again here; I think we need to keep that average win% expectation in mind; It may prove useful in the future.

 

0 Run QS Outcomes

6+ IP  0 ER T 6+ IP 0 ER T% 6+ IP 0 ER W 6+ IP 0 ER W%
514 10.57% 387 75.29%
7+ IP 0 ER T 7+ IP 0 ER T% 7+ IP 0 ER W 7+ IP 0 ER W%
274 5.64% 211 77.01%
8+ IP 0 ER T 8+ IP 0 ER T 8+ IP 0 ER W 8+ IP 0 ER W%
71 1.46% 54 76.06%
  • A QS qualifying D/RO ending in 0 has a very high QSWC%. As intuitive as this may sound we’re talking about an incredible average boost to conversion percentage of at least 20%, and quantifying these changes is important.
  • Our first really interesting anomaly is found here. A D/RO of 8+/0 technically had a lower QSWC% than a +7/0, however negligible. I’m not going to draw anything from this other than the sample of 8+/0 is really small, and the QSWC% remained consistently high, and in line with the surrounding D/ROs.

 

 

WHEW, THAT WAS A TON. I’m cutting our immersive statistical exploration here for now. There’s still a lot of table analysis to be done here but I’m satisfied with the groundwork for now.  We just added a brand new perspective towards dissecting the Quality Start, and we should be proud; I know I am. We’ll pick up next time with some Home/Road split custom QS stats and hopefully create some late round player pairs & streamers being overlooked by the less thorough players. I appreciate your time and focus as we continue down the path less trodden. Follow us @Razzball and @MLBMovingAvg for all the latest in fantasy baseball analysis. Remember, life is a fantasy draft.

Good game, good game, good game, ding dong! See everyone soon!

Image result for baseball yankees handshake gif

 
  1. BJFOHOHL says:
    (link)

    I’d be more into QS if it was a sliding scale or based on ERA for that game. I did the stat for years and it grew tiring to keep justifying it. I hope they change itone of these days

  2. John L

    John L says:
    (link)

    I can’t really make the argument for using QS as merit based, as seen by work done earlier in the series. What we can do, is optimize our opportunities while working within the given rules.

    So it may not be perfect, but if there are rewards for a QS either in Roto or points we can find a way to take advantage

  3. jb says:
    (link)

    The better QS pitchers may have to be targeted in the 1st round or even the 1st pick, so it may not be the primary strategy to use. If we also knew where the pitcher ranked in each of say five previous seasons (or even all of his seasons) to see how he has been trending, it would be something nice to see as shown in your spreadsheet analysis. I may use it to compare similar pitchers, but it doesn’t show whether they’ll win or not because there are other factors that determine that. I don’t think QS stats are supposed to answer that question even though one would think it was related.

  4. John L

    John L says:
    (link)

    We showed that a QS in general does not lead to wins, but when you consider the bullpen + team win% model from prior articles, in accordance with a new D/RO metric we will not only begin to more effectively tier top starters, but using % stats (pt 9 tease), well also identify pitchers being drafted late with elite stats in these new categories. Hang tight, it will come together

Comments are closed.