What is the relative importance of each defensive position?

[Timofmars]

I know overall that offensive is more important than defense. But what about the difference between the positions?

If I'm selecting what positions to set for my players, I could try to maximize the rating totals overall, or I can give special importance to positions like SS,CF, even if favoring them reduces the combined total of defense points over the whole field.

I guess the ratings show how well they handle balls hit to their area, and certain areas get more balls coming their way. So then what are the relative rates of balls hit to each position?

Possibly another factor could be the expected value of a play not made in each position. Like a SS not making the play might give up a hit, but a CF not making the play can be the difference between an out and a double/triple.

[Brad K]

Nobody knows the relative value of defensive positions. Note the positional adjustments in WAR are estimates based on how managers use the players.

Reading the first sabermetrics book, on issue after issue it was determined baseball people didn't understand the game. But when it came to WAR, they based defensive value on the lowered level of offense... and this word was used... "tolerated" by managers to get an acceptable fielder at the position. I was stunned.

So the basis of positional adjustment in WAR is the aggregate judgement of individual managers based on their anecdotal observations, who sabermetricians already proved often don't understand their game.

It seems your question is something like this: If your SS is 5 points better at SS and 10 points better and 2B than your 2B, should your SS play 2B and your 2B play SS? I think that question cannot be answered with confidence due to lack of reliable data.

[Timofmars]

Yeah, the WAR position adjustments are what I found just before I posted. And it basically just tries to make a fair comparison between the offensive ability of a SS and a 1B/DH based on the fact that SS aren't usually as good offensively, so they get a bonus multiplier to make them more equal. If you don't agree with their arbitrary number, you'll need to compare players' WARs only against other players of the same position, or just take the positional adjustment out and use your own judgement.

But for the game, I think % of balls hit to each location could probably provide a decent rough estimate of the value of the position's rating.

The more you increase a player's infield abilities, the more it will raise his rating at a difficult position like SS, but it will have very little effect in raising his rating at 1B. I believe this scaling up of the value of skill at SS is meant to show that most balls hit to a position like 1B are easy for any player to field, and that there is very limited opportunity for balls to be hit in the very particular locations around 1B where the player's skill would make the difference.

But I don't think the rating accounts for the frequency of balls hit there, since that probably wouldn't work when the game tries use the rating to determine the result of a play based on where the ball was hit. So rating probably just affects chances of a potential hit being taken away (or giving away a hit on an error).

[Brad K]

We know the game simulates batting and pitching well so it follows that in the aggregate it must simulate fielding well. Your issue is with the details.

With a lot of work you could compile total chances by position at the game and compare that with what is at baseball reference. If that is a close match and it can be concluded the game simulates fielding in detail well.

But the knowledge that you are seeking, the SS 2b situation I described, has almost zero effect in the game.

[MBarrett]

Check out the Defensive Spectrum if you haven't already.

[NoOne]

if your stats and AI values and other settings aren't the same or at least similar... this stuff won't translate between RL and your League or any other ootp league that differes in relevant ways.

so, take real life research as a very weak guideline, unless very similar to the era you are comparing to your league or the era of your league etc.

http://seanlahman.com/baseball-archi...-james-primer/ links from the wiki... 2-3 need to be re-worded or thought out a bit more, but i didn't realize i agreed wtih him as much as i do. sometimes his perspective is silly - like war is silly... above replacement? replacement is sub-average... why-tf would you compare to that? it's not even intuitively known or understood exactly what it is... and it cahnges every year, lol.

trying to make a deicision easier for people that cannot unerstand the decisiosn itself is folly and useless. the fact that it is a pyrmaid like distribution, as mentioned, is further ereason why WAR is at the very least poorly callibrated with regard to what it is comparing players to to calculate a #WAR.

[Brad K]

If you developed WAA (Wins Above Average) you'd have a hard time defining what average was too. Is it the average everyday player? Average of every player at the position, everyday or bench? Do you weight the bench players as if they were playing full time?

An advantage of the WAR concept (if calculated accurately) is you know who doesn't belong on the major league roster. That's anyone with a negative WAR. With an average player based system, you wound't know that break even point. Unless you had WAR too. Which if you do why bother with WAA?

[NoOne]

you should look at war like any other aggregate statistic like that.. with a very open mind.

a negative war means they aren't jsut sub-mlb, but beyond even a replacement level player, who is below average. it's a terrible "zero" point, imo. i guess if you run a crap team it means something, but a replacement player is objectively junk. i.e. not an everyday player by any means.

i like ops+ more, but it has the same fundamental weaknesses. all-in-one, make it easy to decide stats are poor at best when used in a precise way... they are ballpark figures, pun intended.

first, it's not callibrated perfectly... if it were we'd be able to automate jsut about any decision based on the components of WAR being so accurate and not WAR itself. war is a one-size fits all... that doesn't jive with different roles on a team and what specific talents need to be maximized at which slot in the batting order.

you are better off comparing various resutls to baseline and distribution curve of each individual statistic and making a decision tailored to the role you intend to use that player.

i've seen a speedy ~10war guy with lesser #'s than a ~6-7war guy hitting 40-50hr. i guarnatee the 6 war guy was a better cleanup hitter than the 10war guy with 12hr.

war/ops+ are cute and fun, but they tell you very little useful info. same reason QBR can be junky for the dink&dunk qb's that never throw more than 5yards and have a high completion %. various 'lesser' stats that are exagerrated can cause a big shift in war and it's more about a flaw in the equation. (akin to high defense making a player's overall sky high even though they are crappy in results)

[Brad K]

Originally Posted by NoOne

a negative war means they aren't jsut sub-mlb, but beyond even a replacement level player, who is below average. it's a terrible "zero" point, imo.

A replacement player is ML quality the moment the first player gets injured. And when that happens, the next guy in line, negative .000000etc1 is ML quality.

I realize you'd never have such a player on the 1000 + run 130 win teams you talk about, but that's not real life baseball and probably not the experience of most people playing ootp.

[ThePretender]

Honestly in OOTP if you're a competent GM the floor of your team should be 2.5-3 WAR players (even for your #9 hitters), injury replacements should be 1-2 WAR. If you're ever using a replacement level player you're not running your team properly.

It's difficult in real life, but it's easy in OOTP whether you're playing online or solo to have every spot be a 3 WAR player. Most GMs struggle to balance short & long term sustained windows, as well as recognizing that spending all your assets on elite players and not developing/acquiring quality depth is a bad thing. This is why they fail to have 2.5-3 WAR players throughout their lineup and rotation. Maybe I just have higher standards but my fifth SP should be at least a 100 ERA+ and is usually better, which would be a #3 SP in real life.

Also to an earlier post WAR is based on defensive differences, not offence. At least if you use fWAR.

[NoOne]

Originally Posted by Brad K

A replacement player is ML quality the moment the first player gets injured. And when that happens, the next guy in line, negative .000000etc1 is ML quality.

I realize you'd never have such a player on the 1000 + run 130 win teams you talk about, but that's not real life baseball and probably not the experience of most people playing ootp.

yeah, i dispute the benefit of comparing to that... that's an option for an unprepared GM in RL.. in game, even easier to avoid a "replacement" player. it has nothign to do with me having a 130 win team at all.

the important decisions aren't about a 15 day injury replacement (longer injury ot key piece? season is basically over for most teams anyway or severely reduced probability of success). i'd rather compare to a WS caliber team and what is typically on one... that is the goal. comparing to a bottom feeder is sillyness.
-----------------

basically my bench is ~replacement or worse in some instances... they provide multi-position and defense in my system.. injury replacements are in AAA and prcured long before they are needed and > replacement in many cases. these guys are only used in the short-term, hopefully. not important relative to entire pie.

if you don't win 120-130 consistently in a video game like ootp, it's only due to limiting yourself by not doing somethign different than what happens in reality... a la avoiding improving our metrics for ~100years in an obstinate way in real life... refusal to progress or accept facts is an inherent problem in a % of humans, it seems. whether that behaviour is intended or unintended it is true. it's very realistic for ootp with default settings. not doing so is a choice or maybe they just aren't good enough to do it... i doubt the latter... anyone can succeed at this game, i hope. experience and a will to learn in an objective way is all you need.

settings could shift the win total down.. but teh same exact "ways" of doing things will result in the most wins... so really behaviour doesn't change at all with lower win totals due to settings. the same strategies will succeed the most.

[NoOne]

Originally Posted by ThePretender

Honestly in OOTP if you're a competent GM the floor of your team should be 2.5-3 WAR players (even for your #9 hitters), injury replacements should be 1-2 WAR. If you're ever using a replacement level player you're not running your team properly.

It's difficult in real life, but it's easy in OOTP whether you're playing online or solo to have every spot be a 3 WAR player. Most GMs struggle to balance short & long term sustained windows, as well as recognizing that spending all your assets on elite players and not developing/acquiring quality depth is a bad thing. This is why they fail to have 2.5-3 WAR players throughout their lineup and rotation. Maybe I just have higher standards but my fifth SP should be at least a 100 ERA+ and is usually better, which would be a #3 SP in real life.

Also to an earlier post WAR is based on defensive differences, not offence. At least if you use fWAR.

heh, i could hav avoided my last post...

i bet you liek to have a little disproportionate talent in first 1-5 or 1-6 spots in order though eh? if you ut a little more war (not my metric preference, just for ease) at 3-4-5 and skimp a bit 7-8-9 you'll put up better offense because of the greater number of AB && better talent likely in front of them.

i realize you're jsut talking about ~average war across a lineup, but you'd want to skew it for sure jsut due to # of AB to maximize results. in addition to role of leadoff vs batthing 4th etc.

rarely do i see war more accurately describe a player over ops+... i think smaller samples are better with WAR maybe? i've seen a crap player wiht sub-100 ops have a 2-3 war.. it was hilariously different about the same player with truly crappy stats. i wish i had screenshot it. i don't as often see it the opposite way in my leagues with a full season of data.. not some 80gp injury plagued nonsense skewing everything..

[Brad K]

Unanswered is where the line be drawn and how to calculate it and now to make it easily understandable. Criticism without a solution.

[NoOne]

i gave a solution..

compare to average WS teams... or average player on a Playoff team since 19XX.

comparing to a 15-day player isn't too helpful in any decisions that carries far more weight relative to total "pie".

even better solution was given -- don't use WAR or ops+ or waa or QBR or whatever.

do what the equation basically does but in a more tailored way to role of that player on your team.

a leadoff hitter should have a significantly different WAR-equation than someone who bats 4th.

also, if you understand baseline and distribution of each individual statistic of importance, that is far better than using WAR, et al. it is what WAR does, but you can more easily adjust to specific contexts of their batting role.

war would be great if all things were equal throughout the lineup.. and when a low power and high power guy calculate to the same WAR that also means they will have a ~similar impact at any spot in the lineup... none of that is true and exactly why it's a "pretty" stat instead of a useful one.

using war instead of what i described is the difference between feeding yourself or being fed by someone else that does it the same way whether it's drink or any type of food.

so, if one meal needs a spoon and a straw "they" only use a fork because on average that's what you need. (assumingn forks are most used, just a joke with made up metrics in that regard.. not neccesary to be factual, merely consistent in logical analogy if it were true.)

since 'they' are doing the work for you, you just get what they give you.. or what some aggregate equation amounts to without looking at the parts and relative to context of use.

[Brad K]

Setting zero as the average player on a World Series team makes as much sense as setting zero on a thermometer at the boiling point of water.