bumping an old post (searched for threads related to Contact)... https://www.whatifsports.com/forums/Posts.aspx?topicID=495195&threadID=10994021#l_10994021

It is very important to realize that, conceptually, "Contact" means different things in HBD terms than it does in human terms

The steps of the decision tree hierarchy are:
1- Random events / strategy events? If no-> Plate appearance
2a= Plate Appearance-> "HBP or no HBP?"-- some combination of batter Temper vs pitcher Temper (log5). (Uncertain whether pitcher Control has anything to do with it or not ...)
2b= When no HBP-> "Walk or no Walk?"-- batter Eye vs pitcher Control (log5). If no-> AB
3a= At Bat-> "Hit or Out?"-- relevant batter Split vs relevant pitcher Split (log5)
3b= when Hit-> HR/hit = batter Power vs some factor of pitcher Pitch (log5); 3B/hit and 2B/hit are determined by a combination of batter Power & Speed & BR vs pitcher pitch (log5 ... XBH are overwhelmingly driven by the batter according to the historical data that the log5 is built upon)
3c= when Out-> Strikeout vs Ball-in-play Out (...with the possibility for bad defense to equal a minus play or error, therefore runner reaches base)-- batter Contact vs pitcher Velocity (log5)
4etc.. (it goes on but I'm stopping here at Contact)

Let's take a moment to appreciate just how ridiculous this concept is.

The outcome of Hit or No Hit gets decided before the question of [Has the batter actually contacted the ball at all?] has even been asked.

In the human decision tree, Strikeout vs Ball-in-play should occur at the same point in the decision tree as Walk vs No Walk, because both outcomes are a function of balls-and-strikes. However, in HBD-land the log5 for strikeouts comes AFTER the log5 for Batting Average. Realistically, they needed to design the game so that Splits referred to Batting-average-on-balls-in-play as opposed to just plain batting average.

So for example just tossing out some common, easily multiplied numbers here... Let's compare two guys who get 600 AB and 50 BB (and 0 HBP or whatever, we're just sticking with AB and BB). For arguments' sake let's say 50 BB correlates with 50 Eye rating. Ok, Now let's say player A gets 150 strikeouts (0 contact?) and let's say player B only gets 100 strikeouts (50 contact?).

In human being terms, "splits" would be based on the leftovers of balls-in-play (after walks and strikeouts), so player B would have the opportunity to get a hit on 500 balls in play but player A would only have the opportunity to get a hit on 450 balls in play. So if both players had the same splits values (aka same "ability" vsR and vsL), they would both get hits at equal rate, but would end up with different hit volume-- player B would get MORE hits because he strikes out less and puts the ball in play more often. Let's say each player's equivalent split value equates to the same .300 BABIP... Batting average on those two numbers ends up being Player A .225 and player B .250. Both players have equal "hitting ability" aka split, but player B has better contact and therefore gets more hits because of that ability.

Whereas in HBD terms, player A and player B's splits both get judged on the same overall 600 AB and end up with the exact same batting average. Contact is a very distant afterthought.... The ability to put more balls in play gets a player greater opportunity for base hits IF AND ONLY IF THE OPPONENT PLAYS BAD DEFENSE. When you're playing a team with ridiculously good defense such as Franchise Profile: Golden Gates with .985 (not really even that good) and 129+11- ratio (outstanding), then that means maximum contact will reach base an extra 15 times out of 1000 on errors (still gets scored as 0/1 on the boxscore) aka 0.015 aka 1.5%, and an extra 11/4425 (minus-plays divided by total-putouts-plus-minus-plays) aka 0.0025 aka 0.25% lol. Thus, contact under those circumstances is virtually irrelevant. On a positive note, contact does generate additional RBI on sacrifice flies and run-scoring-groundouts, but on the downside it does lead to more double-plays.

So if the question is Would you rather have Eye (the very beginning of the decision tree and a direct correlation to OBP via isolated-discipline) or Contact (the ability to hit .271 or .272 instead of .270, and maybe an extra 10 circumstancial RBI per year on sac-flies and groundouts), the answer seems obvious. I personally rank Contact as a very-distant 6th-most-important blue attribute (7th if you include Speed... and arguably 8th if you include Temper)

This is great, but do you feel confident in the underlying assumption that the developers are being 100% truthful and complete in their answer? I'm not a contact guy, feel it's easily the 5th most important attribute but I'm not sure I'm convinced it's as irrelevant as you say here..

What is your source for the decision tree that you are referring to?

@pjf-- as you know, I agree with the basics of the hierarchy as presented; it fits my observations and studies as well as being presented in that slide that someone showed a couple of weeks ago from a presentation sitestaff did. But as someone whose offensive models (outcome-- OPS) rate contact as considerably more important than that, I have a couple of questions.

1) Are we sure contact is only a "K vs. no K" rating? It appears in my models that contact affects AVG way more than that, which it could do if it also were a factor in step 3a along with splits (note that, since sitestaff has said that catcher PC has its effects at the OAV level, I expect catcher PC fits in at step 3a as well).
2) Similarly, I think I remember that pitches have been said to factor into OAV as well... do you have strong reason to believe that they're not part of 3a?
3) I'm not comfortable that splits, both batter and pitcher, don't have a role in converting singles to 2B/3B (although I agree with you that SPD seems to be the primary factor in 3B). In any case there are tons of very high power guys who hit 40 HR and 15 2B a season, so I'm pretty sure power isn't a factor here. Why do you think what you think about the 2B/3B vs 1B algorithm?
4) (small issue) A quick look at HBP tells me that pitcher temper isn't a factor here-- batter temper is, and pitcher control a little bit (although I didn't look hard enough to know if the effects of pitcher control are related to simply facing more batters).

In my experience it seems like contact heavily impacts power. What I mean by that is a 70 contact 70 power guy will hit more homers than a 20 contact 70 power guy. I have also seen high contact/high power/high eye/low splits guys perform very well compared to a low contact low splits. Your decision tree doesn't support this which makes me wonder am I suffering from observational bias or if the tree is wrong.

Posted by oriolemagic on 7/6/2017 7:53:00 PM (view original):
In my experience it seems like contact heavily impacts power. What I mean by that is a 70 contact 70 power guy will hit more homers than a 20 contact 70 power guy. I have also seen high contact/high power/high eye/low splits guys perform very well compared to a low contact low splits. Your decision tree doesn't support this which makes me wonder am I suffering from observational bias or if the tree is wrong.

I fully agree with this. I'm seeing guys with power in the low 60s with a higher contact rate hit 20+HRs.

Yeah, I don't buy this. I don't value contact nearly as high as I do batter eye, but I still think it's worth more than K vs. groundball/flyball out.

pj, as much as I like to read ideas of how the game engine works, this current thread seems so obviously counterintuitive as to be unrealistic, and you trying to handwave away clear counterexamples seems really forced, especially when you talk about "ballpark proof power" and other strongly non-algorithmic explanations. If you see results that don't match your model, the proper response is to examine the model, not insist that the model is fine but there are unknown or unknowable external factors that magically produce the divergent results.