The Turn DP rating—does it do anything?

[Garlon]

This is something that needs some testing as Matt mentioned.

In the file I have I took the 10 worst seasons by 2B at turning double plays and they averaged -25 double plays turned for the season with a relative rate compared to the league of 0.76, so they turned 76% as many double plays as expected.

For the top 10 seasons, they are averaging +28 and are at 1.32 relative to the league, so they are 32% better than expected.

Keep in mind that these players are part of a double play combination, so the SS on these teams are about the same too. You could realistically credit about each partner with being responsible for half the turns on the double play. So if I take these 10 worst 2B and pair them with an average SS then that makes about -12 double plays turned by the team for the season. If we pair the top 10 2B with an average SS we should gain about +14 double plays for the team.

In fact I just looked at the same information for SS and the results are the same for the +/- double plays turned and the percentage relative to the league for the best and worst seasons.

I think that a few versions ago of OOTP I posted about how the middle infielders were performing far too strong and turning way too many double plays and perhaps this was adjusted in the game. We have this information that suggest we are basically between 0.76 and 1.32 relative to the league for any actual 2B or SS.

From there this needs to be calibrated within the game so that we are getting these results approximately.

So if you play a 3Bmen out of position at 2B how poorly should they perform? My suggestion would be no worse than what the worst 2B can do simply for balancing purposes. A 3Bmen out of position is going to perform very poorly in terms of other fielding plays though too.

You should conduct this test using a historical season so that the league totals modifiers are being applied to the results.

[Garlon]

I am very interested in the topic because I want fielding to be implemented correctly in the game. There are many fielding issues that I have posted about regarding historical games.

[Furious]

Originally Posted by Garlon

I am very interested in the topic because I want fielding to be implemented correctly in the game.

We are 100% in agreement there.

[Mat]

If you cannot turn a double play, you most likely wash out of the system very early on.

Can anyone name a player who has been moved from 2B due to their poor DP performance?

[Isryion]

Originally Posted by Mat

If you cannot turn a double play, you most likely wash out of the system very early on.

Can anyone name a player who has been moved from 2B due to their poor DP performance?

In OOTP, you can put anyone you want at 2B, including RF. So, while turning a double play may be important to be a second baseman, not being able to turn a double play doesn't wash you out of the system entirely. Double plays are not positional ratings in OOTP.

I believe a couple years ago, Brian Dozier was getting a lot of flak from Dodgers fans (and also Minnesota fans before that) because he was unable to complete what should have been routine double plays. I thought I read that was one of the reasons he started to play less after being traded to LA.

[Isryion]

Just to add some data here (and my testing is nowhere near Furious's level) because there was some question about the fact that people believed that since it was 2B involved in Furious's data, they weren't convinced there was a problem.

I put together a fictional modern league. For 4 teams I moved a 1B/OF who had no 2B experience and low DP rating to 2B. 4 four other teams I did the same but switched out BOTH their SS and 2B. Rating Scale was 1-100. Below are the team DPs and individual for one season. (Injuries and fatigue were off)
Image 1, Image 2

While there are some clear differences individually, as a team, all teams were within the range of ML double plays and players with ratings below 5 (on 100 pt scale) and teams. Even with OF/1B at both middle IF positions were not far from the mean as a team.

Jack Castro (DP 4/100) managed to turn 100+ double plays with his fellow LF as his DP partner. Jorge Paz managed 100+ double plays with a DP of 1/100. Basically, anyone in OOTP is, at the least, competent at performing a double play.

[Garlon]

Well the top of the list you have teams apparently with over 140. We really need double plays and double play opportunities to evaluate this better. For example, consider all those errors by those players out of position. They just gave themselves a bunch more opportunities to turn a double play by allowing more players to reach base.

The 1906 Cubs turned 100 double plays and the Boston Beaneaters turned 102 double plays. The Cubs had the best Defensive Efficiency in baseball with .735 and the Beaneaters had the worst with 0.669. The Cubs had an estimated groundout rate of 54% while the Beaneaters had about 58%. The Beaneaters also struck out 140 fewer batters than the Cubs. When you account for these things the Beaneaters turned about 4 more double plays than expected, while the Cubs turned 23 more double plays than expected for the season. That puts the Cubs at 30% more double plays turned than expected and the Beaneaters at 4% more turned than expected for the season despite the fact that the Cubs turned 2 fewer double plays than the Beaneaters.

[Isryion]

Originally Posted by Garlon

Well the top of the list you have teams apparently with over 140. We really need double plays and double play opportunities to evaluate this better. For example, consider all those errors by those players out of position. They just gave themselves a bunch more opportunities to turn a double play by allowing more players to reach base.

The 1906 Cubs turned 100 double plays and the Boston Beaneaters turned 102 double plays. The Cubs had the best Defensive Efficiency in baseball with .735 and the Beaneaters had the worst with 0.669. The Beaneaters also struck out 140 fewer batters than the Cubs. When you account for these things the Beaneaters turned about 4 more double plays than expected, while the Cubs turned 23 more double plays than expected for the season. That puts the Cubs at 30% more double plays turned than expected and the Beaneaters at 4% more turned than expected for the season despite the fact that the Cubs turned 2 fewer double plays than the Beaneaters.

I get your point about opportunities and am aware that opportunities ae an issue. You'll also have to explain to me how you're getting expected double plays from your numbers because it's not clear to me. I could reverse engineer that here, since those numbers are available in OOTP.

But for me, I don't actually need more data at this point. I mean, how many more opportunities would you want to assume that Paz had, in the above scenarios to make Paz turning over 100 DPs acceptable? For me, the number would be something in the 500 range of double play opportunities (with him converting then about 20%). Besides, Furious's study already looked at that by leveling almost the entire playing field as best he could, which would include several seasons and even out opportunities to a large extent with the teams being exactly the same except for DP ability of their first basemen.

I will give you some numbers, just comparing Paz's team, Chicago, to Memphis, which is where Mata, the player with the most double plays played and had the most in the league

Paz's team gave up 444 more hits than Memphis but Memphis actually walked 52 more players. If we assume all 444 hits were singles (I don't think anyone thinks that would be true, but let's do it) and take away the 52 extra walks we're at 392 extra runners at 1B. We should probably cut that into 1/3 though, because we know that you can't turn a double play with 2 outs. That cuts it down to 261 runners on first base more than Memphis had with less than 2 outs. Paz had 21 more errors than Mata. That's bad, but it doesn't result in a lot of extra actual double play opportunities, so I will bump the 261 back up to 282 (and that's assuming they all happened with less than two outs).

We know not all of these situations then involved a possible double play result but I'm not super comfortable going beyond that because it involves a ton of estimating (look at the number of hits +walks and then the resulting double plays - you'll see that for normal teams double plays are consistently between 5-10% of that number and lower when you toss in errors). Either way, we're looking at something like 40-60 double play opportunities extra (conservatively high, imo since it's assuming that every extra hit = a single). Even 100 DP player wouldn't convert all those, since lots of other factors apply. Before anyone asks, Chicago pitchers (DP 1 1B at 2B), had a worse slugging than Memphis (DP 93 2B), so it's likely the difference of runners sitting at 1B is even smaller.

Maybe that's too much estimation, but we're nowhere near the number of expected opportunities where'd I'd expect a 1/100 double play rating to get 100+ double plays. Also, I don't yet see anyone giving any hard data that shows that double play rating is having a significant impact or where I would expect a double play rating of 1 to be. I just keep seeing questions that some of us are trying to answer. I'd love to see a sim that proves it wrong and that the double play rating means something significant, the way other ratings do.

[Furious]

Originally Posted by Isryion

Jorge Paz managed 100+ double plays with a DP of 1/100.

Originally Posted by Garlon

We really need double plays and double play opportunities to evaluate this better.

Garlon... with all due respect... a player with a Turn DP rating of ONE on a scale of ONE TO ONE HUNDRED turned over 100 double plays in a season and your reaction is "well, we really need to see more data"?

1 out of 100. How many more DP opportunities do you think he could have had? A thousand?

[Furious]

I’m now going to take a swing at this from a different angle. I’m going to try to think like a programmer, even though I’m not one.

Since tests indicate that even players with the lowest possible Turn DP rating can turn double plays in OOTP, it would appear that the programmers have essentially built in some “automatic” double plays. The rating of the fielders is irrelevant on these plays—anyone with a pulse can turn them.

This is not inappropriate in my opinion. The game assumes that every professional player—even one with no middle infield experience—is capable of fielding an easy ground ball and making a short throw to the other middle infielder, not every time but certainly some of the time. That seems correct to me. The “zero point” in double plays should not be zero, but something higher than that, something that includes some percentage of the plays that all players are capable of making.

So the issue, I think, is simply that the zero point has been set way too high. In Isryion’s test a guy with the lowest rating possible turned over a hundred double plays in a season. That result would suggest that the game is saying that about 100 double plays a season are automatic. 100, or thereabouts, is the zero point.

Well, in some major league seasons, there are teams that don’t turn 100 double pays in a season, and of course those teams aren’t starting the real-life equivalent of a one-on-a-scale-of-one-to-a-hundred double play turner in the middle infield. If something in the neighborhood of 100 double plays a season were “gimmies”, all teams would turn far more than 100 double plays every year.

Yes, I know that double play opportunities vary from team to team and season to season. Yes, I know that pairing up Player X with a good double play partner means Player X is going to turn more double plays, and pairing up Player X with a lousy double play partner means Player X is going to turn fewer double plays. Yes, I know the League Totals Modifiers will affect the number of double plays turned. I’m not saying any of that is irrelevant. I’m saying that all those variables aren’t nearly sufficient to explain how a player with a rating of one-out-of-a-hundred was able to turn 100+ double plays in a season. It’s not a fluke. It’s a matter of the game “giving” a certain number of double plays to anyone.

I don’t know what the right amount of “gimmies” or “automatic double plays” should be. I know it shouldn’t be a hundred. Are there stats that tell us how many double plays are “routine”? If there are, that might be a good starting point. The lowest-rated DP-turners should be able to turn some (but definitely not all) routine double plays. Whatever that number or percentage is should be the zero point.

[Rain King]

I don't think that is what was tested though. The original test was with a player that had a bunch of "good" 2nd base skills with the exception of the Turn DP rating. I think if you are looking for that zero point you need to look at a player who is not fit at all to play the position by OOTP standards.

I think it is likely a false assumption that "only" the Turn DP rating contributes to how well a player converts double-plays.

[Craig Scarborough]

Originally Posted by Furious

For this test I set up a six-team league, 140-game schedule, injuries off, player development off, morale and chemistry off, all parks identical, all teams exactly identical except the second baseman. The second baseman is the same (DP rating of 4; 140 “under the hood”) on four of the teams, but on one team he has a Turn DP rating of 5 (200 under the hood), and on one team he has a Turn DP rating of 1 (30 under the hood).

I don't really test OOTP that much, but this sounds interesting. Is this easy to set up? Can this be shared as a template? I'm just thinking that this can be used for a number of parameters (not just DPs) to see their true effect.

[Isryion]

Originally Posted by Rain King

I don't think that is what was tested though. The original test was with a player that had a bunch of "good" 2nd base skills with the exception of the Turn DP rating. I think if you are looking for that zero point you need to look at a player who is not fit at all to play the position by OOTP standards.

I think it is likely a false assumption that "only" the Turn DP rating contributes to how well a player converts double-plays.

With all due respect, did you look a few posts up where I ran a season where 8 teams had LF/RF/1B who had no 2B experience and low DP ratings, among many other low IF ratings? Furious and I have run three tests that we've posted results for.

-The first was my initial experiment with one player.
-The 2nd was his experiment that started this thread where everything was the same except double play rating.
-The third was five posts above yours and answers your question pretty completely in my opinion, despite a limited study compared to Furious (see the two linked images). The player we've been discussing a few posts now, Paz, was a 1B in that game and probably should be a DH. His ratings full fielding ratings are Range: 21, Error 39, Arm 25, Turn DP 1. The guy can barely play catch at 90ft, but he turned 100+ double plays. He wasn't the only one, but he should put the idea that DP rating is keying off of anything else to rest.

[Furious]

Originally Posted by Craig Scarborough

I don't really test OOTP that much, but this sounds interesting. Is this easy to set up? Can this be shared as a template? I'm just thinking that this can be used for a number of parameters (not just DPs) to see their true effect.

Yes, it’s a very useful way to isolate certain functions of the game to see how they effect things. I used a similar file to test park factors some years back.


I’d be happy to share the file. Compressed it’s about 12 MB. Anyone know the best way to make it available?

[Craig Scarborough]

Just got a hold of the file. Thanks, Furious.

First run was to just test 3B range. Yeah, it makes a HUGE impact.

[kriscolic]

It makes perfect sense that the DP rating should have an insignificant impact on a player's ability to turn a DP because we don't really have any idea who is good or bad at it in real life.

It's just like chemistry. Some people are undoubtedly good/bad for team chemistry. But, outside of anecdotal evidence, we have no way of identifying who they are. And we cannot in any way measure the effects of chemistry on team wins and losses. So, in OOTP, chemistry is a thing but it has a relatively small effect. Seems reasonable to me.

We can say that Bill Mazeroski was great at turning two, but we don't really know for sure and any attempt to place a number on it is guesswork. DP totals depend on so, so many factors including, but not limited to, number of opposing baserunners, opposition SB attempts, opposition hit'n'run frequency, pitcher groundball percentage, quality of fellow fielders, defensive positioning, and precise batted ball location. With present day Statcast data we can begin to measure some of these things. But for historical data, forget about it.

I think OOTP is correct to downplay the significance of this particular skill.

[kriscolic]

I wonder what would happen if one were to play a left-handed thrower with the same ratings at 2B?

[Furious]

Originally Posted by kriscolic

It makes perfect sense that the DP rating should have an insignificant impact on a player's ability to turn a DP because we don't really have any idea who is good or bad at it in real life.

We can say that Bill Mazeroski was great at turning two, but we don't really know for sure and any attempt to place a number on it is guesswork

Well, I respectfully disagree. Every statistical measurement depends on “so, so many factors”. Using your argument one could state that “we really have no idea” if one pitcher is better than another, since so many pitching statistics are heavily dependent on team defense.

The fact that fielding statistics are more difficult to measure and less exact than hitting statistics doesn’t mean you can just throw them out and say they mean nothing. We may not have a perfect understanding of how much better (or worse) Bill Mazeroski was at turning double plays than some other player, but it’s not true that we have no idea.

If I were in a historical league and Bill Mazeroski was in it I would want the game to simulate his skills as accurately as possible, but frankly that is completely irrelevant to what we’ve been talking about. Trying to determine how best to simulate historical players’ abilities is a separate problem from the one I addressed in the original post.

Obviously I haven’t looked at every rating OOTP has bestowed on every historical player but I’m going to go out on a limb here, and say that there are probably ZERO historical second basemen (except those who only played a few games at second) with Turn DP ratings as low as the ones we’ve been talking about. For real second baseman, the ratings are probably sandwiched into a much narrower range. If you’re using a 20-80 scale, I’m going to guess that there are no 20s or 25s or 30s or 35s among players who played a full season’s worth of games at second base, anytime in history. I’ll bet 40s are very rare too. Probably almost all of the “legitimate” second basemen that have been rated by OOTP have Turn DP ratings in the 45-80 range, or maybe 50-80. I’m sure someone will correct me if I’m wrong. If I’m right, though, it means that the effective range of Turn DP ratings for a real second baseman uses only about half of the spectrum of possible ratings. I don’t have a huge problem with saying that the difference between Bill Mazeroski and an average second baseman is not enormous, and may be somewhat difficult to measure. I DO have a huge problem with saying that the difference between Bill Mazeroski and an inept first baseman who’s never played second base trying to play it isn’t enormous or easy to measure.

I’m in a fictional league. In a fictional league we have to assess players by their ratings (and also by their statistics, but we don’t have statistics until the player has played). An extremely low Turn DP rating should tell us “don’t put this guy at second base, he’s not a second baseman”. But that rating, as it stands right now, doesn’t tell us anything of the sort. Right now the Turn DP rating says “ignore me, I have almost nothing to say.”

[kriscolic]

Originally Posted by Furious

Using your argument one could state that “we really have no idea” if one pitcher is better than another, since so many pitching statistics are heavily dependent on team defense.

Not SO, BB, HBP, HR. Defense independent pitching is one of the foundational concepts of modern baseball analysis.

Originally Posted by Furious

The fact that fielding statistics are more difficult to measure and less exact than hitting statistics doesnÂ’t mean you can just throw them out and say they mean nothing.

I never said that fielding statistics should be thrown out. I was talking about one specific defensive statistic: DP for second basemen.

...

We don't know, in real life, who is good at turning double plays. We don't know, in real life, the difference between the best and the worst at this particular skill in terms of DP turned or run value or any other metric you care to choose. So there is no basis to say that OOTP is getting it wrong (or right, for that matter). In that case, I think it's fine to err on the side of caution, as it were, and decide that the difference between the best and the worst is insignificant.

Your mileage may vary. But I would pose the following questions: What should the difference between the best and the worst DP turners be, and how did you arrive at that figure?

[Craig Scarborough]

Originally Posted by kriscolic

Your mileage may vary. But I would pose the following questions: What should the difference between the best and the worst DP turners be, and how did you arrive at that figure?

Well, I'd think you'd look at the empirical data and determine the best GDP/GDP opportunities vs. the worst and use that as your range that you're hoping for OOTP to generally replicate. However, I'd think a spread of 30 (probably below you're typical player that would earn any kind playing time as a major leaguer) to 200 is probably a larger swath and therefore the results to be outside "normal" range as seen by real data.