MLB FORUM

Comments

  • sethayates

    I was originally going to start another thread about how high scoring NBA has been lately, but it looks like we have that covered already. From time to time, I see something interesting on Twitter but the 140 character limit on that site makes it impossible to discuss. A few days ago, I was discussing the correlation of advanced stats in MLB with @msonichdrhass (who is a great follow on Twitter). He was explaining that variation in hitter HR/FB% can be explained by three factors. They are: Hard hit , K, and WFA/C (runs created per 100 fastballs seen). He had something even more interesting to say about pitcher HR/FB%.

    This is copied from @msonichdrhass Twitter account:

    For pitchers HR/FB% is (almost) solely a function of Hard%. For every increase of 10% in Hard% HR/FB% goes up (on average) 3.85%. Round up to 4%, if you want. So, if a P allows 25% Hard%, his HR/FB% should be ~10%….this remains consistent over the years, with the ratio moving ~1% in either direction. This is both a simple and powerful way of looking at HR/FB%. The maxim is that the HR/FB% regresses to 10% for a P. But it appears that Hard% is actually driving the analysis. If a P’s Hard% is below 25% or so, his “low” HR/FB% isn’t “lucky”—-unless, of course, his Hard% is “lucky” itself. So, bottom line, if you want to help predict a P’s future HR/FB%, look at a P’s Hard% relative to his HR/FB%, not his HR/FB% relative to 10%.

    Ok, for all of you who didn’t get that. What he is saying is that you can possibly predict when a pitcher’s HR/FB% will regress to the mean simply by looking at his hard Hit%. For example, if a pitcher is allowing HR/FB of 17% but his Hard Hit% isn’t 45% or higher we can expect some regression coming. He should eventually get back to around league average of 10% unless we expect his Hard% to remain high. On the other hand, if a pitcher is allowing a Hard Hit% of 45% but somehow his HR/FB% is only 9% we can expect that the HRs will be coming.

    What are your thoughts on this? I’ll see if I can get the original author to join this thread. What other stats do you look at in MLB and what do you compare them to?

  • VChair23

    I hope @draftcheat pops in, he’s the one I’ve heard mention Hard% more than anyone else. Very interested to see where this goes.

  • rdogg63

    I definitely look at batted ball velocity. Ideally you find pitchers who are susceptible to hard hit contact and fly balls and/or line drives. This is why you should always feel good choosing guys batting against Ian Kennedy.

  • drhass

    thanks for moving this over, Seth! I was also planning on exploring this in greater depth. [I’m “drhass” here on RG, and “@msonichdrhass” on Twitter, btw]

    You basically down have what I found. Here’s a brief summary in different words: A certain percentage of hard-hit balls become HRs. League-wide, this percentage is remarkably consistent, varying by only a few percent each year. So, if you look at what percent of hard-hit balls a pitcher gives up, that should tell you what their expected HR/FB% should be.

    Let me use an example to illustrate:

    Marco Estrada. He used to be one of my favorite Ps to target for HRs, but something happened last year. Here are his HR/FB% numbers for each of his full seasons as a starter (2012-2015):

    2012: 10.5%
    2013: 11.9%
    2014: 13.2%
    2015: 8.7%

    Now, his Hard% each of those years:

    2012: 37.4%
    2013: 36.4%
    2014: 33.8%
    2015: 27.2%

    Without doing any math, the eye test says that his HR/FB% dropped a ton when he lowered the percent of hard-hit balls he was yielding. When I ran a linear regression on a whole bunch of stats to determine what effect they had on HR/FB%, I found that virtually ALL the variance (league-wide) in qualifying pitchers’ HR/FB% was explained by their Hard%.

    In short, the conventional wisdom is that HR/FB% will coalesce around 10%, given a large enough sample size. I found that HR/FB% is a function of how many hard-hit balls a pitcher gives up, so if HR/FB% is coalescing around 10%, it’s because Hard% coalesces around a certain value. It may be that a pitcher yielding a 17% HR/FB% is “unlucky,” but look to his Hard%. It may be that he’s yielding a ton of hard-hit balls, in which case his HR/FB% could be exactly what we expect.

    Of course, as baseball has pretty wild variance, the particular number explaining the relationship changes and shouldn’t necessarily be used as an exact calculation. It looks like this, mathematically:

    HR/FB% = X * Hard%.

    Last year, X = .384
    In 2014, X = .327
    In 2013, X = .326
    In 2012, X = .381

    You can see that X varies a little every year, but the relationship (at least when talking about starting Ps who meet the IP threshold to count as “qualifying”) is incredibly strong. I wouldn’t use this as a precise calculation of expected HR/FB%, but I think it can be used as a good framework to determine whether someone’s “(un)lucky HR/FB%” is really a matter of luck or whether it’s a function of them giving up more/less hard-hit balls.

    One last example:

    2015 – I. Kennedy, 17.2% HR/FB, 35.2 Hard%
    2015 – B. Anderson, 17.0% HR/FB, 24.1 Hard%

    Anderson’s HR/FB% is far more “unlucky” than Ian Kennedy’s.

  • drhass

    @rdogg63 said...

    I definitely look at batted ball velocity. Ideally you find pitchers who are susceptible to hard hit contact and fly balls and/or line drives. This is why you should always feel good choosing guys batting against Ian Kennedy.

    Yes—-I plan on looking at the determinants of Hard% next (Fastball velocity is one I expect to show a strong relationship). And good call on Ian Kennedy—-he’s the poster boy of “perhaps his HR/FB% won’t come down”

    I did look at a pitcher’s propensity to give up FBs, LDs, and GBs——those did not explain variations in HR/FB%. Of course, if a P gives up more FBs then, all things being equal, he will give up more HRs; but it didn’t explain the rate at which his FBs became HRs.

  • smallANDflaccid

    At what rate do you track or update HR/FB and/or Hard% – daily? weekly? monthly?
    Your stats here are yearly, but are retrospective – in terms of daily predictability what timespan is most relevant? I would assume some recent period is more valuable than last year’s?

  • drhass

    @smallANDflaccid said...

    At what rate do you track or update HR/FB and/or Hard% – daily? weekly? monthly?
    Your stats here are yearly, but are retrospective – in terms of daily predictability what timespan is most relevant? I would assume some recent period is more valuable than last year’s?

    I’m sure that the folks over at Fangraphs have a good statistical answer to this (there are folks who have calculated how many PAs one needs for certain stats to “level off”), but I think you can start looking at the present year after about a month or so of starts. And in the early going, I’d compare a pitcher’s Hard% with his historical performance there to see if what he’s doing is in line with what he’s done. Ditto with his HR/FB%.

    League-wide, I imagine the the value for X above (and for Hard% and HR/FB%) will settle down after a few weeks or a month. Until then, predicting what will happen will be a LOT more art than science.

  • MTro86

    RG Writer

    This would seem pretty reasonable, similar to considering K% regression in relation to SwStr%. I think I remember Eno Sarris having a long conversation on Twitter with someone saying that Hard% doesn’t correlate well with BABIP at all, but better with power numbers like wOBA and ISO.

    Makes sense, although I would guess that park factors play a role in this too (the equation might look different for a one pitching in Cincinnati than it would Seattle). Also, I wonder how batted ball tendencies might play into this. Might some pitchers allow hard ground balls more often, but softer fly balls (or vice versa)?

  • pokerrob1970

    DraftDay BLB Finalist

    Wow and actual thread here without people crying in it.

    This is really great stuff. Thank you for posting

  • Annihilus

    With quality of contact stats, I think it’s important to remember that these stats are based on the type of batted ball. There are hard, medium, and soft hit balls of each batted ball type. A weak line drive might be hit harder than a medium ground ball. A medium line drive might be hit harder and at a higher speed than a hard hit fly ball.

  • VChair23

    I use fangraphs for Hard%, does anyone have a site with a database for batted ball velocity over the last 1-3 seasons?

  • Undonersg

    • Blogger of the Month

    From Fan Graphs, here is their explanation of how it is measured.

    “Calculation:

    For the early years of quality of contact stats, the BIS video scouts had to make judgments, but since 2010, the video scouts recorded the amount of time the ball was in the air, the landing spot, and the type of batted ball (fly ball, ground ball, liner, etc) and the BIS algorithm determines if the ball was soft, medium, or hard hit.

    Unfortunately, the exact algorithm (the exact cut points/methodology) are proprietary to BIS and we can’t share exactly what constitutes hard contact, but the calculation is made based on hang time, location, and general trajectory. It’s not perfectly analogous to exit velocity, but until we have more complete StatCast data, it’s a step up from simply knowing line drive versus fly ball.

    Importantly, these stats are based on the type of batted ball. So there are hard line drives, medium line drives, soft line drives. A medium line drive might be hit at a higher speed than a hard hit fly ball.”

    I had a theory that hard%, medium%, and soft% was a great way to look at players that would regress towards the mean. If a player is contacting the ball at a hard % but has a .230 batting average, we would expect that the player would regress towards the mean because typically the harder you hit the ball, the better chance you have of getting on base. But if a player is hitting .470 over the past few games, but has a high soft % we would assume he would regress towards the mean because he is getting “luckier” than other players.

    I dont know if this theory holds any weight, and havent been able to look at any analysis of it, but from being a baseball coach for 15 years at the college and high school level it makes total sense. Would like to run some numbers on it, but dont have the time.

  • biggiesmails72

    @VChair23 said...

    I use fangraphs for Hard%, does anyone have a site with a database for batted ball velocity over the last 1-3 seasons?

    http://baseballsavant.com/

  • stevietpfl

    Morning Grind co-host, Lead NASCAR Analyst

    • 2015 FanDuel MLB Playboy Mansion Finalist

    • 2015 FAWBC Finalist

    I love this, great stuff!

    I’ve always thought a lower hard contact % would result in less homers, but never really did the math. From a pitchers perspective, hitting spots with junk, is always a good way to stay away from the long ball.

  • VChair23

    @biggiesmails72 said...

    http://baseballsavant.com/

    Question answered

  • biggiesmails72

    There have been many articles on FG about hard hit rate already

    http://www.fangraphs.com/community/the-importance-of-hard-hit-percentage/
    http://www.fangraphs.com/blogs/what-hard-hit-rate-means-for-batters/
    http://www.fangraphs.com/community/hard-contact-rate-and-identifying-breakout-candidates/
    http://www.fangraphs.com/community/hard-hit-percentage-outliers/

    I think there is more to looking at targeting under valued hitters based on their contact rates, than by looking at bad pitchers to stream against.

  • biggiesmails72

    @VChair23 said...

    I have baseballsavant but for example it only has data on about 60% of AB’s for Paul Goldschmidt. Is there a more complete publicly available database or is baseballsavant the best there is?

    Im pretty sure they have all pitches vs Goldy, maybe you are confusing his batted ball ABs with his total PA?

  • VChair23

    @biggiesmails72 said...

    Im pretty sure they have all pitches vs Goldy, maybe you are confusing his batted ball ABs with his total PA?

    Correct, moving too fast while I’m at work.

  • drhass

    @biggiesmails72 said...

    There have been many articles on FG about hard hit rate already

    http://www.fangraphs.com/community/the-importance-of-hard-hit-percentage/
    http://www.fangraphs.com/blogs/what-hard-hit-rate-means-for-batters/
    http://www.fangraphs.com/community/hard-contact-rate-and-identifying-breakout-candidates/
    http://www.fangraphs.com/community/hard-hit-percentage-outliers/

    I think there is more to looking at targeting under valued hitters based on their contact rates, than by looking at bad pitchers to stream against.

    All of this is certainly true. I was looking at the effect of hard hit rate for pitchers (well, at the outset, I was looking at what factors went into a pitcher’s HR/FB%)—-not hitters. So, those articles explore the other side of the coin from what I was looking at.

    You absolutely should NOT use only this one piece of analysis to identify which hitters have good value on any particular day. Completely agree.

  • Sjv

    Great stuff @drhass

    So looking over the last half of 2015 I noticed archer and gerrit cole sporting some of the highest hard% (mid-high 30’s) while the hr/fb% was very low. Archer was around 9% and cole 4.7%. Is this just a case of being lucky or is the talent enough to overcome certain correlations?

  • drhass

    @Sjv said...

    Great stuff @drhass

    So looking over the last half of 2015 I noticed archer and gerrit cole sporting some of the highest hard% (mid-high 30’s) while the hr/fb% was very low. Archer was around 9% and cole 4.7%. Is this just a case of being lucky or is the talent enough to overcome certain correlations?

    Cole is an interesting case. I was just comparing his first and second halves, and he was noticeably worse in the second half—-including giving up many more hard-hit balls AND fly balls. This should have translated to giving up more HR in the second half, but his HR% went down.

    On first blush, I’d say that this is just variance (or, more correctly, variance based on factors not being considered). It could just be “luck.” I don’t see a particularly rational explanation for it. I do note a couple things:

    1. His LD% is pretty high. Thus, more of his hard-hit balls are becoming LDs and not FBs. That doesn’t explain his low HR/FB%, but it’s an interesting data point.

    2. His fastball got tremendous results last year. When I first ran the regression, I found that the pitchers who had really good fastballs saw a very slightly lower HR/FB%. Overall, the effect of how good the pitcher’s fastball was had only a negligible impact on HR/FB%. But it might be entirely possible that for pitchers with elite fastballs the HR/FB% decline is noticeable.

    So, I’m going to go with “luck,” but I leave open the possibility that guys with elite FBs (as opposed to just merely “good” or “average” FBs) are much tougher to hit. I haven’t run those particular numbers to see.

  • MTro86

    RG Writer

    When writing pitching articles last season, I was often more than a little surprised that the Mariners and Mets would show up on top of hard hit% (even earlier in the season when the Mets were running a rag lineup out there). So park factors have to have some influence in results and may stop certain players/teams from regressing as much as expected. Also, shifts. A lot of times it doesn’t matter how hard you hit it if they know exactly where you’re going to hit it.

  • drhass

    @MTro86 said...

    Also, shifts. A lot of times it doesn’t matter how hard you hit it if they know exactly where you’re going to hit it.

    Would certainly explain why Hard% doesn’t always translate into BABIP (or some other similar stat), though it wouldn’t say much about fly balls turning into (or not turning into) HRs.

    I had no easy way of running park factors on this data. I agree that the park has to make some sort of difference. Certainly, fly balls go out more often at Coors or Yankee Stadium, but I don’t know by how much over another stadium (I couldn’t find a place where I could see how many fly balls were hit in a given stadium). My guess is that it makes a difference for the outlier places. If a pitcher pitches half his games in COL or NYY or HOU, I’d expect his HR/FB% to be a tad higher. Not sure if the effect is noticeable when you’re talking about a stadium that has only a small effect on HRs.

    If someone can find me the numbers, I’m happy to run them to see what I find.

  • Sjv

    @drhass said...

    Would certainly explain why Hard% doesn’t always translate into BABIP (or some other similar stat), though it wouldn’t say much about fly balls turning into (or not turning into) HRs.

    I had no easy way of running park factors on this data. I agree that the park has to make some sort of difference. Certainly, fly balls go out more often at Coors or Yankee Stadium, but I don’t know by how much over another stadium (I couldn’t find a place where I could see how many fly balls were hit in a given stadium). My guess is that it makes a difference for the outlier places. If a pitcher pitches half his games in COL or NYY or HOU, I’d expect his HR/FB% to be a tad higher. Not sure if the effect is noticeable when you’re talking about a stadium that has only a small effect on HRs.

    If someone can find me the numbers, I’m happy to run them to see what I find.

    @drhass does this site have what you were looking for?

    http://www.hittrackeronline.com/stadiums.php

    Really interesting to dive into

  • X Unread Thread
  • X Thread with New Replies*
  • *Jumps to your first unread reply

Subforum Index

RotoGrinders.com is the home of the daily fantasy sports community. Our content, rankings, member blogs, promotions and forum discussion all cater to the players that like to create a new fantasy team every day of the week.

If you or someone you know has a gambling problem, crisis counseling and referral services can be accessed by calling 1-800-GAMBLER (1-800-426-2537) (IL). Gambling problem? Call 1-800-GAMBLER (NJ/WV/PA/MI), 1-800-9-WITH-IT (IN), 1-800-522-4700 (CO), 1-800-BETS OFF (IA), 1-888-532-3500 (VA) or call/text TN REDLINE 1-800-889-9789 (TN).