enhancement requests for Bernie

[keyzick]

[quote:8864294f3c="Geoff Brookes"]Here is a brain teaser question -

IF the clutch hitting is a "sham" - i.e. given sufficient actual at bats, any player's RBI production would tend to be predicted and replicated PURELY based on their hits and normal player characteristics - then why does a player like DeJesus exist (there are others like him) who are CONSISTENTLY rated as clutch hitters by strat?

In effect, doesn't this prove that some clutch element really DOES exist in the data, that is not PURELY reflected by hits and normal player card elements???

Geoff Brookes[/quote:8864294f3c]

Just as it can regulate RBI's down, it can also regulate RBI's up...so perhaps SOM figures DeJesus's straight card would not be enough to generate his RBI total without a little help.

[qksilver69]

GB - what keyzick said is right on. The SOM formula relies on RBI per PA, a real hackjob approach to clutch. But that is the only measure that allows them to adjust RBI totals accurately for their replay purposes. So non-power hitters like DeJesus who get a lot of RBI - which has been proven to mostly be a product of on-base of the guys in front of you and NOT a skill as you suggest - will always have good clutch in SOM, regardless of whether they are truly clutch hitters (meaning they hit better with RISP/2 out).

The issue I have is really this: does SOM *need* to accurately recreate RBI totals? This is an old-school vs. new-school argument.

SOM believes that RBI totals are an important part of their recreation agenda, and that they need clutch to make them accurate. I (and sabermetrics in general I think) would argue that if you build the other stats correctly, the RBI variances would be incidental and clutch is not needed.

GB, I hear you in that some guys may be better at situational hitting than others, but I would also argue that real situation hitting happens with zero or 1 out, not 2 outs! So the way SOM uses clutch is a complete misnomer.

Think about it - trying to advance a runner with a well placed groundball or a sac fly only happens with less than 2 out, but SOM's clutch just impacts RBI totals by lowering or inflating a hitter's BA with RISP & 2 outs!

For Coyote - I am not advocating that we make a confusing set of options for use in basic leagues. I think the options I mentioned should only be applied when the user selects "advanced" for league creation, so would not impact usability the way you describe.

[gbrookes]

qksilver - you are absolutely right that moving runners means nothing with 2 outs! So, if I was right, then the game mechanism would be wrong anyway.

What I think I am saying is that, before we throw the baby out with the bath water, I think it is worth investigating whether there is something real and true going on with some players - not just a product of favourable situations that have nothing to do with their inherent abilities. (Again, if DeJesus is above average clutch just because he has runners on base in front of him, wouldn't A Rod be killer clutch by the same logic, with great hitters getting on base in front of him? But A Rod (and many others on great teams) are not great clutch hitters in strat, so I don't really believe that explanation).

If there is something inherently good at producing RBIs in some players, let's study the reality of what's going on, and then improve the game design. If the study indicates that those player's get more RBIs through no characteristics of their own, then the rule should be dropped. Obviously, I am skeptical that this is the case. But I think it is worth studying.

Geoff

[SteadyEddie33]

I don't know, I find myself in the middle when it comes to clutch. I mean qksilver, marcuswilby, et al are absolutely right in terms of the real statistical meaning of the RBI. It is itself dependent upon many variables. On the other hand, I think --without much basis, I admit-- that although the clutch factor reduces to a straight RBI/PA, there must be other terms in the equation related to slugging percentage of the batter's card. Also, we've got to remember all the mediocre pitchers that never take part in our 12 team 'all-star' leagues.

Conversations like these always take me back to the '85 Cards and Tommy Herr's 100-plus RBI behind Vince Coleman and Willie McGee in old Busch. I don't have the stats handy, but it's pretty easy for anyone to see that their OB, SB and baserunning played a major role. He didn't hit 40 doubles and didn't get 200 hits, despite having a good avg. For his card to produce the right RBI in a StL. replay, I think he'd need additional singles to ensure he got enough of them with men on 2nd.

That said, you put Adrian Gonzalez in that lineup, or a Michael Young '09 for that matter, and both will produce far more RBI than their true performance. This is despite the major clutch reductions. So, I guess I'm saying: since this is likely true, the clutch system works. Why turn it off?

Great hitters with solid slugging without clutch reduction will go bananas instead of simply overproducing, lol. That is, vs the full set of pitchers.

[qksilver69]

GB - the only way I can see to search for clutch per what you are writing is to do 2 things:

1) normalize for runners on base (meaning adjust so that a weak OB lineup in front of a player does not penalize them in terms of RBI)
2) compare hitters with similar BA & SLG to each other

By nature hitters with high slugging should drive in more runs per PA than hitters with lower slugging if the baserunners in front of them are equal. So the only way to really test for clutch is to look at players with similar BA/SLG and see if one player can consistently can produce more RBI than another.

From your theory on Dejesus, he should be able to show a higher RBI per PA rate over his BA/SLG peers because of his skill set.

It does not take away from the fact that SOM is mixing apples & oranges in a near-criminal manner when a) measuring RBI per PA, then b) applying a "fix" that moves the BA up or down with RISP & 2 out. My point is there IS no "baby" to throw out with the bathwater, it's ALL BATHWATER. :D

[quote:40404e6abc]That said, you put Adrian Gonzalez in that lineup, or a Michael Young '09 for that matter, and both will produce far more RBI than their true performance. [b:40404e6abc]This is despite the major clutch reductions.[/b:40404e6abc][/quote:40404e6abc]

I disagree with the last part & its the essence of why in these leagues I want to be able to turn off clutch. In every league I have Adrian G, his RBI ends up close to real life, but his BA is artificially low because my lineups generate a lot more RBI potential than what he had in real life. So in order to keep him close to 99 RBI or whatever he had in reality, SOM is now depressing his BA 70 pts. Which is more important? That he gets to 120 RBI or that his BA is within shouting distance of real life?

[JKolak]

You still like arguing, don't you qk? As you have throughout the years qksilver, you made a lot of valid points. They are good ideas to incorporate into the on-line game. The on-line game got to this point thanks to pioneers like you who first made suggestions to improve this game starting back when it premiered in 2002.

[gbrookes]

Here's a different idea for an enhancement to the strat-o-matic game (I know this is beyond TSN's involvement - this would be strat itself changing its game):

How about something like N or W power ratings and card results, but on the BATTER's card!

If it is a good idea for the pitcher's card (with respect to batter's power), it has to be at least 1/2 way as good an idea to have the same concept on the batter's card, but this time with respect to a pitcher's home run control!!!

Consider this - what are the odds, realistically speaking, the you can duplicate Ubaldo Jimenez's 2009 performance, playing out a Colorado season? Some may argue, and I'm sure some may claim that a sample of 50 games results in a mean or median number of homeruns allowed that is the same as his 2009 statistics. But really, playing 1/2 his games in Coors, facing hitters usually with N power and ballpark homerun readings on the batter's card time after time after time, how often will U. Jimenez duplicate his 2009 home run control?

So, why not have readings on homeruns and ballpark homeruns on the batter cards as follows:

-some readings the same as before - no N or W rating;
-some with readings that are * N homerun (N for normal), but if the pitcher has a C rating (for control), the result is a single instead.

The way I see it, most pitchers would have an N rating. Only a few pitchers that have excellent homerun control (basically those who would tend to have difficulty duplicating their real life homerun control performance without this system) would get the C rating. Even then, a significant portion of the batter homerun results would be homeruns regardless of the pitcher homerun control rating.

I think this would really enhance online strat gaming, and the realism of the game. In my experience (with my teams or watching my fellow gamer's teams), the super-pitcher suffers all too often because the opposing batters keep rolling homeruns on their own card, for which the super-pitcher has no answer or influence on the result (in my opinion, unrealistically).

What do you think?

By the way, I intend to suggest this enhancement to strat-o-matic directly.
Geoff Brookes

[gbrookes]

By the way. with respect to clutch hitting, I agree that the current system is nowhere near as good as it could be, either in terms of the methodology (if it really is that simplistic) or in terms of the overall structure and execution of the idea. A much better system would do a much more sophisticated analysis. I would strongly argue for using multi-year data for each player, measuring the difference between batting average with Runners in Scoring Position (RISP) (regardless of the number of outs) versus the batting average in all other situations. You could possibly layer on top of this analysis a second idea, which is advancing the runners (essentially my DeJesus argument), but you would also have to take into account the extent to which this would be likely to occur through the normal rules operating (e.g. groundball results). It could very well be that this second clutch element is not worth the hassle, in terms of a small overall effect. Then the rule would operate in a way similar to its current form, except that it would operate regardless of the number of outs.

With respect to multi-year data, why not? The non-clutch hits would have to be adjusted to produce the same probability of the overall batter performance, but with the benefit of getting a clutch hitting effect that is statistically significant, and not wildly variant from card to card, it seems like it would be a fun element to the game and its strategy, and more true to life and meaningful than the current system.

Geoff Brookes

[gbrookes]

As another post-script, I agree that the adjustment based solely on real life RBI's is "all bathwater". It is far to crude a measurement, and it makes an unrealistic assumption that there would never be any variation in real life between hitting performance and production of RBI's - EVEN WITH ALL OTHER FACTORS BEING EQUAL. In other words, whose to say that a particular batter isn't just getting lucky with the timing of his hits, or unlucky? Or, is it asking too much to try to analyze ALL of the factors that result in RBI production and then normalize them!!!!!???? I think so!!!!

So, let's discard that idea, and just measure what a batter actually does, in terms of raw data, and let the RBI's follow what happens in real baseball when you get more than the average number of base hits, or do a good job hitting the ball and putting it in play!!!!!!!!!
:)
Geoff

[pcbaseballsims]

[quote:0fa8e339f9="Geoff Brookes"]

If there is something inherently good at producing RBIs in some players, let's study the reality of what's going on, and then improve the game design. If the study indicates that those player's get more RBIs through no characteristics of their own, then the rule should be dropped. Obviously, I am skeptical that this is the case. But I think it is worth studying.

Geoff[/quote:0fa8e339f9]

Makes sense Geoff. I concur...keep "clutch" hitting such as it is.

How about enhancements for P's ratings? (Like a certain other company).

Such as...Ps ability to[b:0fa8e339f9] NOT [/b:0fa8e339f9]walk the leadoff batter...or when they allow HRs they often come w/ [b:0fa8e339f9]NO [/b:0fa8e339f9]runners on base.

Can you say "Doc: Halladay? Cliff Lee. Tim Lincecum. Go Giants! Beat da Phillies!

:P