Authentic gameplay: Stealing

[SteveP]

There is a problem with stealing in OOTP, as far as authentic gameplay is concerned, but I’m not sure what can be done about it (or whether it’s worth caring about). That problem has to do with the frequency of double steals. Note: I am slowly going through a 1960 league (for the third time) and comparing what happens in OOTP with what actually happened in the same real life games. Nothing magic about 1960, just what I chose for the purpose.

IRL (in 1960), double steals happened very rarely. Less often than squeeze plays, to use one comparison. Even a team with several good stealers would do this only once or twice during a season. In fact, a manager was more likely to call a double steal with men on first and third (the man on first serving as a decoy – a play not available in OOTP) than with men on first and second. In OOTP, double steals happen much more often – perhaps 10-15 times as often, for a team with good stealers, to throw a number out there (I only have a count from a small sample, so that’s just a guesstimate).

This leads to several questions: 1) why so rare IRL; 2) Is this an era-specific problem only (that is, do teams today double steal at a rate comparable to what happens in OOTP? I doubt it, but can’t be sure); and 3) what can be done, if anything, to control or reduce double stealing, without unduly reducing other types of stealing?

Whether or not to care about this, is a hard question to answer. On the one hand, I can’t say for sure that it’s distorting statistical behavior. Markus has said in another thread that the game is accurate in the proportion of 3rd base SBAs to SBAs overall. Can this be true for a 1960 league where double steals are so out-of-line with reality? Don’t know. It’s a bookkeeping problem to figure that out. So, I may just take it on faith that the stats are OK.

On the other hand, when I see double steals in which the player on first is someone who is an unlikely base stealer (a somewhat pokey catcher, for example), I have to wonder about how the decision to double steal is made in the game engine. Also, now that I know this is absurd behavior for a 1960 league, it’s hard to ignore. (A little knowledge is a dangerous thing …)

So, this is just a post to see if anyone has some cogent info or insight to offer.

FWIW

[RchW]

Help me understand better. In the OOTP league, are you playing out or watching every game in order to see these double steals? I've played out well over 2000 games and can only remember 3 or 4 against while I've tried 10-20. I'm not saying your wrong but you need to quantify at least a seasons worth of data IMO

Beware of small sample sizes, not only can they mislead but they are often flat wrong.

Do you count a missed run and hit or hit and run with two on base as a double steal? What about a botched sacrifice? It would show up in the stats as 2 SB (or 1SB 1CS or an error) but there would be no way to tell the intent. Without knowing intent It cannot count as a double steal.

It seems to me that in the low scoring era of the 60's there would be a high possibility of putting runners in motion to find holes and generate offense but I didn't see enough games to know for sure.

IRL for the 60's how can you tell what play was on? Game logs? Certainly not stats. So where did you get the once or twice per season IRL?

I'm not trying to shoot this down but I feel that when we (players) make posts on authenticity or reality of specific play results in the game, we need to back it up with numbers that can be verified.

[RchW]

Oh, I just read your post again.

Note: I am slowly going through a 1960 league (for the third time) and comparing what happens in OOTP with what actually happened in the same real life games. Nothing magic about 1960, just what I chose for the purpose.

Are you using game logs (play by play for retrosheet) to see double steals? I'm not sure that is possible as neither document announces 'double steal".

See this from July 1,1960

Chuck Cottier moves to 2B
--- 98% H Smith Single to CF
O 1-- 98% L Jackson Bunt Groundout: C-1B/Sacrifice; Smith to 2B
-2- 98% J Javier Walk
OO 12- 97% J Cunningham Double Play: Strikeout, Smith Caught Stealing 3B (C-3B)

Is this a double steal? If the catcher does not get thrown out at third it would have been as Javier would not be standing still at 1st.

Since he was thrown out we must look at intent. I think we both agree that this was not a steal attempt, especially with a Catcher running. Hmm, looks like a play OOTP gets slammed for all the time.

[Garlon]

Stolen Base Attempts and Stolen Base Success rates are right on target for all seasons in history.

SteveP, I suggest that in your tests that you simulate a few years before doing analysis. League Totals Modifiers don't kick in until year #2 of your league. So if you want to look at the accuracy of 1960 stats, then you should start a league in 1957 and quick sim until 1960.

[SteveP]

Originally Posted by Garlon

Stolen Base Attempts and Stolen Base Success rates are right on target for all seasons in history.

I completely understand that you are satisfied if total season stats produce results that look like what happened IRL. Based on what I read on these boards, I know there are many OOTP players like you.

So, let me repeat myself : I am focused on what happens in individual games during the season. I don't regard season stats as meaningless, but I do think they only tell half the story of what happens in individual games.

The stats produced by OOTP don't tell you the proportion of 3rd base steal attempts to 2nd base steal attempts in a season. But even if you had that data, and could show it is consistent with the same data IRL, you still don't know how much of each was generated in double steals. It is for reasons like that I've been looking at dimensions of the game that season and career stats alone can't address.

Anyway, the total rate of SBAs and SB success I'm getting is consistent with history. And double steals are not. So, no, I'm quite sure that simming out multiple seasons will not cause double steals in OOTP to occur with anything close to the same frequency as they did IRL, at least for the era I'm looking at right now. I'm deferring judgment on whether this is era specific or not.

[SteveP]

Originally Posted by RchW

Are you using game logs (play by play for retrosheet) to see double steals? I'm not sure that is possible as neither document announces 'double steal".

If we can agree that this is a double steal,

--- 97% W Mays Walk
1-- 98% W McCovey Walk; Mays to 2B
12- 98% O Cepeda Mays Steals 3B; McCovey Steals 2B

then I think we are in business. This is what I am referring to. This happened very rarely in 1960 IRL. It happens much more often in OOTP, in my 1960 league. I am not counting the double steals involving men on first-and-third that happened IRL, because they can't happen in OOTP. I hope that clarifies.

[Garlon]

SteveP - If you can collect the data for steals of 2B and steals of 3B and data for double steals going for each individual season going all the way back to the 1950's from Retrosheet maybe we can model it. We don't know where we are until we A. have the historical league totals data and B. have the OOTP sim totals to track those events. Ultimately, since we already get SB Attempts and SB% correct, their impact on scoring will be basically the same.

Right now one of the main flaws in historicals is that league ERA and Runs Scored per 9 innings runs a bit high even when league OPS and all other categories are right on target. This suggests that OOTP tends to get a few more baserunners from first to third and from second to home on singles than perhaps it should.

[RchW]

Originally Posted by SteveP

If we can agree that this is a double steal,



then I think we are in business. This is what I am referring to. This happened very rarely in 1960 IRL. It happens much more often in OOTP, in my 1960 league. I am not counting the double steals involving men on first-and-third that happened IRL, because they can't happen in OOTP. I hope that clarifies.

Sure I agree. So what is the rate of double steals per game for OOTP vs Real life in 1960? If per game doesn't work for you choose whatever rate units you want. Right now all we have is anecdotal evidence.

[SteveP]

So what is the rate of double steals per game for OOTP vs Real life in 1960?

I provided an estimate of that in a previous post. After collecting some additional data, I think that estimate still stands.

If per game doesn't work for you choose whatever rate units you want.

The whole point of this thread is that the rate of double steals can't be adjusted in OOTP (unless you know something I don't )

Right now all we have is anecdotal evidence

Not sure what to say to that. I am reporting what I found from studying this on a game-by-game basis. That is the only way, since there are no stats IRL or in OOTP. I understand that since there are no stats on this, much of the OOTP world has no interest, but these threads on authentic gameplay are aimed at those OOTP players who do take an interest in what happens in individual games.

So, there is too much double stealing in OOTP for a 1960 league. Might be too much double stealing in OOTP period (but don't know yet). Apparently can't do anything about it anyway (but still trying different tactics). Am still curious about why it was so rare IRL in 1960. So that's what I'm putting on the table for anyone who has some thoughts about it.

[RchW]

Originally Posted by SteveP

I provided an estimate of that in a previous post. After collecting some additional data, I think that estimate still stands.

You estimated 1 or 2 double steals per year IRL in 1960. Let's choose 2. That is a rate of 2/162 games. You estimated OOTP to be 10-15 times that so the OOTP rate would be 20/162 to 30/162.

The reason I bring it up is that you might see 2 double steals in 3 games but not see another for 50 games depending on the opponent and who is on base.

Have you confirmed that OOTP produces that rate over a significant number of games? We also talked about other plays that could produce a double steal result.

The whole point of this thread is that the rate of double steals can't be adjusted in OOTP (unless you know something I don't)

I understand that. I was letting you indicate what rate to use (see above). Using 20/162 games is a little unweildy IMO, but it's all I can think of.

Not sure what to say to that. I am reporting what I found from studying this on a game-by-game basis. That is the only way, since there are no stats IRL or in OOTP. I understand that since there are no stats on this, much of the OOTP world has no interest, but these threads on authentic gameplay are aimed at those OOTP players who do take an interest in what happens in individual games.

So, there is too much double stealing in OOTP for a 1960 league. Might be too much double stealing in OOTP period (but don't know yet). Apparently can't do anything about it anyway (but still trying different tactics). Am still curious about why it was so rare IRL in 1960. So that's what I'm putting on the table for anyone who has some thoughts about it.

I thought I was taking an interest. This will be a difficult issue to document without a proper sample size.

[Garlon]

SteveP - If you are running year #1 tests of a simulaton league, the results don't mean much anyway. OOTP's modifiers won't kick in untl year #2. And things really tend to be very accurate in year #3 and beyond. If you want to analyze 1960, start your league in 1957 and sim up to 1960.

If you want to say that something is wrong with SB's then it needs to be quantified. How many steals of 3B were there vs steals of 2B? How often did double steals occur on a per opportunity basis? How often were both men successful? How often was the man out trying to steal 3rd, or 2nd on the double steal? Then do that for every season since 1957 and we'll have some data to look at.

You'd need to be able to cull all of this historical data first, then somehow cull all of the OOTP sim data. There's absolutely no way we can know where things stand without the data. There's no solution that can be presented to make things more realistic until we know what realistic is from a quantitative standpoint. We do know that SB Attempts and SB success is on target with history in OOTP.

The stolen base itself is not a very good run-producing play anyway. A 65% success rate is pretty much the break-even mark at which you'd score just as many runs as you would if you hadn't attempted any SB's at all. The double-steal is probably an even less rewarding play since the catcher's throw to 3B is only 90 feet vs the 127 feet it is to throw to 2B.

If you want to chat on AIM, my screenname is: wewantthedoors

[SteveP]

Originally Posted by Garlon

SteveP - If you are running year #1 tests of a simulaton league, the results don't mean much anyway. OOTP's modifiers won't kick in untl year #2. And things really tend to be very accurate in year #3 and beyond. If you want to analyze 1960, start your league in 1957 and sim up to 1960

This only makes sense to me if you tell me there are "hidden" modifiers built into the game that re-calibrate specifically 3rd base steals and double steals, at the end of a season, if they are out-of-line with 2nd base steals. AFAIK, the game primarily calibrates on total SBAs against expected SBAs as a % of runners on 1B. Then SBAs of 3rd base are handled during games as some hard-coded percentage or reduction in the probability of SBAs of 2nd base. What triggers double steals specifically is a mystery to me.

But to come back to your point: if in 1960 actual SBs were 920 (or the relevant # for SBAs if you prefer), and in my league I am getting 920 SBs for that season, my expectation is that the derivative stats (e.g. steals of 2nd base, steals of 3rd base, steals of home, etc.) will line up reasonably well with history. I don't expect precision, just reasonably well.

If you are saying that this expectation is wrong -- that, for example, it is not only possible but even likely that SBAs of 3rd base (let us say) will be way off in a first season, even if total SBAs are correct, that would be good to know. Moreover, if you are saying that such discrepancies are tracked by the game engine (in a way hidden from us) and will be corrected (again, in a way hidden from us) in future seasons through built in "modifiers" (also, hidden from us), then this would be good to know too. As much as I like OOTP, I figured out pretty quickly that there is a more going on "under the hood" in the game engine than meets the eye, so I am certainly open to the possibility that this is the case here.

Otherwise, my expectation is that I could be getting exactly the correct number of SBs or SBAs each season for ten seasons, but the frequency of double steals would still be higher than they occurred IRL because there is nothing that specifically re-calibrates double steals. That is not a complaint, by the way. I was just trying to pursue some interesting questions about this, both in terms of what happened IRL, as well as what happens, and what can be done about it, if anything, by OOTP players.

I appreciate the offer to pursue this off-line, but I only started these threads as a way to pass along what I learned from my game-by-game analysis to others in the OOTP community who care about how closely gameplay (and by that I mean what happened in individual games) in OOTP looks like real baseball. I assume some of these people are reading these threads and trying to glean anything useful from them. As long as there is useful information or insight that emerges, then let's share it. I'm also happy to be shown the error of my ways in public, if that helps to better inform those who are most interested in these topics. In any event, I have no interest in beating any of these topics to death. Not my style.

[SteveP]

Originally Posted by Garlon

The double-steal is probably an even less rewarding play since the catcher's throw to 3B is only 90 feet vs the 127 feet it is to throw to 2B.

I'm not sure how much I want to jump all over this, because I think you just meant to convey some good baseball horse-sense. But, I guess the temptation is too great.

I've said that IRL, in 1960 at least, double steals were very rare. In OOTP, they are much more frequent (maybe 10-15 times as frequent). So, based on what you've said here, OOTP is not only unhistorical, it's even bad baseball.

I've also noticed that in OOTP the catcher throws to second about as often as he throws to third (if he throws at all). That could also be bad baseball, except that I think the AI is just going after the slower runner -- who probably shouldn't have been stealing in the first place (because maybe he is a pokey catcher ), which is the sort of thing I see as questionable behavior in OOTP.

Having said that, I'm fully conscious that I've only done this analysis for one year, so it's certainly possible that the frequency of double stealing in OOTP is noticeably closer to authentic in later years. I plan to look at a later year at some point.

And, let me say again -- because I probably can't say it enough -- this is not about making complaints or raising issues. It's about talking with others in the OOTP community who are interested in getting more authentic gameplay in individual games. If people who are involved with the development of OOTP take an interest, that's frosting on the cake.

[SteveP]

Originally Posted by RchW

You estimated 1 or 2 double steals per year IRL in 1960. Let's choose 2. That is a rate of 2/162 games. You estimated OOTP to be 10-15 times that so the OOTP rate would be 20/162 to 30/162.

The reason I bring it up is that you might see 2 double steals in 3 games but not see another for 50 games depending on the opponent and who is on base.

Have you confirmed that OOTP produces that rate over a significant number of games? We also talked about other plays that could produce a double steal result.
I understand that. I was letting you indicate what rate to use (see above). Using 20/162 games is a little unweildy IMO, but it's all I can think of.

I thought I was taking an interest. This will be a difficult issue to document without a proper sample size.

I appreciate the interest, and sorry if my response seemed curt.

I knew someone would put me on the spot about stats, so I started extending my count on DS as I've been looking at other things happening in my games (like bunts). So, here's what I can give you now: in 38 games, IRL there was 1 (that's a one) double steal. In OOTP there have been 9 (that's the total for both teams in all games). However, the 1 that occurred IRL was the only one for that team all season (the Giants, which is the team I am primarily keying on). In other words, in those 38 games in real life, none of the teams playing against the Giants had a DS, while the Giants had the only one they did that season. In OOTP, the Giants and their opponents combined for 9 DS in those same 38 games.

Now I've been through this season a couple times, but I haven't specifically focused on DS frequency so much before. Still, I'm pretty sure there were only 1, maybe 2, other DS that occurred in the Giant's schedule that year. I fully expect to see at least 30 in their 154 games in my OOTP league.

If 38 games is not a large enough sample for you, that's fine. I'll probably have a bigger count at some point. But, considering how often one sees men on 1st & 2nd, less than 2 outs, 38 games ought to be enough to tell you what the AI is likely to do the next time you see it.

More to the point -- and this is the essence of the problem for someone like myself trying to get authentic gameplay -- I will see DS roughly every 5-6 games as I watch the Giants play their OOTP league. As a fan with season-tickets to Candlestick in 1960, I would have been astonished to see such madness on the base paths (in fact, I might even be saying "no wonder Bill Rigney got fired"). I hope people reading these threads understand that this is what motivates my interest in the whole authentic gameplay question.

Anyway, if you have more questions along these lines, go ahead and ask them and I'll try to answer them.

[SteveP]

At this risk of making this thread look like a SteveP blog, I'm going to throw out one more post, because I may have solved a couple of puzzles in my own mind about DS behavior.

1. I suspect that the total # of actual DS attempts IRL in 1960 was actually zero. I think that what I see in the game logs in that year, that look like DS attempts, weren't. Just guys on 1B taking advantage on their own of opportunities to steal 2B, while a play was being made at 3B. It's the only plausible explanation why a team like the Giants would DS one time early in the season, and never do it again. They never called a DS that time either. So, now it looks even more certain that part of the problem is era-specific.

2. But I also think it's probably not entirely era-specific. A true DS play (as opposed to an accidental, opportunistic DS) is a difficult, risky proposition. It requires coordination, a good call on what pitch to run on, avoiding one of the runners getting picked off first, doubling your chances of losing a baserunner, etc. Teams that develop experience with the play can reduce that risk, but it's still a daring thing to do.

Given that, my intuition is that DS plays in OOTP happen too frequently and too routinely, and teams get away with it too often, to realistically model the risks and difficulties of that play. As an analogy, I think that if gameplayers saw successful squeeze plays every 5-6 games during a season, they would think "whoa, what's this about!" In 1960, squeeze plays occurred quite often (about once in 20-25 games), but no DS. Food for thought here.

Again, I don't have data from a later season IRL, so not yet a complaint -- just a placeholder for a possible future complaint.