5+ hits in a game

[thbroman]

I haven't seen this being discussed in a while, and I wanted to ask whether the Lords of the Game (Markus and Andreas and the betas) consider this to be a problem in the current version of the game and whether it will be addressed in a patch. Or is OOTP2007 now finished?

To my mind, there are WAAAY too many 5+ hit games. There's some odd little thingey in the engine that cranks them out too often, although I'm not sure what the overall impact is on the game. Certainly it doesn't make much difference for season stats. Thus the question: is this a problem and if so, are there any plans afoot to deal with it?

[injury log]

It was discussed recently in Tech Support:

http://www.ootpdevelopments.com/boar...d.php?t=149377

No one seemed to have hard data about how often 5-hit games occur in real baseball, which makes it difficult to judge whether there are too many in OOTP. Does anyone know how common 5-hit games are in real life?

[Burgandy]

I can see them being as infrequent as a no-hitter IRL, because there are so many games where a guy won't get 5 AB's, or even 5 PA's. I did a quick google and found a Pawtucket Red Sox site that indicated a guy got 5 hits in May 07, and it was the first time since '04 that they had someone do it. Obviously, that's a small sample, I'd like to see some numbers too.

I agree they're maybe a little too frequent in the game, largely because of the factor of "this guy's hitting everything tonight, I'm going to get him to chase something" And he'll either succeed or end up walking him. That's what may happen in real life that doesn't in the game.

Another thing that makes it hard to get 5 hits IRL... If you have the plate discipline needed to be a high avg hitter, you're usually going to work a walk or two on days when you're seeing the ball well enough to get on at will.

[Burgandy]

Okay.. the highest end of the spectrum...

Ichiro, 2004, sets record for most hits one season... had 4 5-hit games. I wouldn't expect anyone could repeat that unless they make a run for the single-season hit record. They better bat high in the lineup, not take a lot of walks, and have great speed down the line. I'll bet a couple of Ichiro's 4 games had some infield singles in there.

[injury log]

Originally Posted by Burgandy

I can see them being as infrequent as a no-hitter IRL, because there are so many games where a guy won't get 5 AB's, or even 5 PA's. ....

If you follow the link above to the Tech Support thread, you'll see some concrete numbers. I'd still like to know what the MLB rate is per season. 5-hit games are certainly more common that no-hitters, in any case. I'll quote my post from that thread -->

I think that's the point; 5-hit games aren't that rare. Ty Cobb did it 14 times, Roberto Clemente did it two days in a row, Dave Winfield had three 5-hit games in one month, even Simon Pond has done it 5 times in his minor league career. I wasn't able to find league-wide numbers in a quick web search, but I did find that 10 Marlins have had 5-hit games in franchise history, for one every 1.4 seasons. If that's a representative modern day pace, then in a universe with complete minors (roughly 210 teams, 60 of which play half-seasons) you'd expect it to happen about 130 times in a year. I'm sure that number is off (I'm using the Marlins as a baseline after all), but it's not a rare event.

[Eugene Church]

A Beta Team member said his research showed in 2006 21 players got five hits in a game.

Another Beta Team member ran a 50-year sim and found this:

He sampled three seasons and found that a player got 5 hits 38 times, 43 times and 59 times.

I know it is a sample of only three seasons, but OOTPB averaged 46, while in real MLB last year it happened only 21 times.

In my 33-year test league, I checked one year and found that there were 17 players who got five hits in a game.

This was just a 16-team league.

MLB would have had around 30 at the rate that it occurs in OOTPB.

For my league to be accurate, it should have had about 8 players get five hits in a game.

[Zeyes]

I posted my 15-year sim stats over in the above-mentioned tech support thread, and I've now looked through the Retrosheet files for the last 10 years (minus 1999, which isn't currently available):

AL:
Year   5+   AVG
1997   12  .271
1998   13  .271
2000   14  .276
2001   10  .267
2002   14  .264
2003   13  .267
2004   26  .270
2005   14  .268
2006   14  .275
---------------
      130
 
weighted AVG: .2699
 
5+ hits per 1000 games: 12.7
 
 
NL:
Year   5+   AVG
1997   11  .263
1998   12  .262
2000   24  .266
2001   15  .261
2002   11  .259
2003   15  .262
2004   15  .263
2005   21  .262
2006   10  .265
-----------
      134 
 
weighted AVG: .2625
 
5+ hits per 1000 games: 11.7

The difference between the leagues should be explained mainly by:

- the difference in batting averages
- the DH (which may be partially reflected in higher batting averages already, though)
- double-switching in the NL (resulting in fewer players getting 5+ plate appearances in NL games, even discounting the DH/pitcher spot effect).

I suspect that batting average is the primary driver...as a crude estimate, let's use the probability to get 5 hits in 5 at-bats, which increases by about 20% with each .010 in batting average.

Anyway, my test universe's batting average was about 7 points above the AL stats above, roughly the same as the AL/NL difference coincidentally, so we might expect that my test universe should have seen about 14 to 15 five-hit performances per 1000 games if it adheres to recent MLB level. In fact it had almost 25, which does seem rather high in comparison.

[pstrickert]

Just thinking out loud . . .

In real life, would a manager be more likely to pitch around a player with four hits in the game? If so, that would make it harder for someone to get that fifth hit. I doubt that type of strategy is reflected in OOTP.

[Sweed]

Originally Posted by pstrickert

Just thinking out loud . . .

In real life, would a manager be more likely to pitch around a player with four hits in the game? If so, that would make it harder for someone to get that fifth hit. I doubt that type of strategy is reflected in OOTP.

I'd tend to think this is part of it, and also an ootp player doesn't know the significance of coming up already having four hits. While a real player may be affected knowing he has a chance for 5.

The game uses ratings to determine outcome, so the question becomes how would you go about fixing this? Or do you even want to?

One solution is to have the AI consider pitching around the player. I wouldn't want this to be automatic though, could be you have a poor hitter that is just having a lucky day. And for all we know it already does this but the players are still coming up with hits.

Or do you have the computer "know" when a batter comes up with 4 hits and reduce his chance of getting another (clipping his stats for reality's sake), in effect reducing his ratings for that AB? Not something I'd want to see. If that is how it would have to be fixed I'd just as soon leave it as is.

[thbroman]

I wonder whether it would be possible to match 5-hit performances against those times when a player is designated as hot. Or is the "hot" tag simply a statistical report, with no in-game effect?

[Zeyes]

I've decided to break down the 5+ hit performances by at-bat count, thinking that perhaps some subset is over-represented. Here are the breakdowns of the MLB numbers listed above, and my OOTP sim. The last number in each line is the incidence per 1000 games:

AL:
5-5       78     7.6
5-6       39     3.8
5-7        5     0.5
5-8        1    <0.1
6-6        5     0.5
6-7        2     0.2
 
NL:
5-5       77     6.7
5-6       46     4.0
5-7        5     0.4
5-8        3     0.3
6-6        3     0.3
 
Combined:
5-5      155     7.1 
5-6       85     3.9
5-7       10     0.5
5-8        4     0.2
6-6        8     0.4
6-7        2    <0.1
(Long*    14     0.7)
 
OOTP:
5-5     1495    13.6
5-6      866     7.9
5-7      167     1.5
5-8       48     0.4
5-9        5    <0.1
5-10       5    <0.1
5-11       2    <0.1
6-6       86     0.8
6-7       33     0.3
6-8        3    <0.1
6-9        4    <0.1
6-10       1    <0.1
7-7        3    <0.1
7-8        1    <0.1
(Long*   235     2.1)

* "Long" is a summation of all counts in which there are at least two more at-bats than hits.

My initial suspicion was that 5-for-5 might be the main culprit, but after running the data, it looks like OOTP is overshooting the MLB numbers on all combinations, and especially so on very high at-bats counts (though that may be a sampling artefact).

Maybe OOTP is having more players reach 5 at-bats in a game overall, and that's what's driving the numbers. I'll see if I can extract that data from my simmed game and compare it to the MLB stats.

[Zeyes]

dola,

unfortunately it turns out I have a limited sample to work with...just the last season, and part of that is already gone as well as some leagues have already reached the pre-season. Still, what's left is 6 leagues with a 30-man roster limit and 8 leagues with a 22-man limit, all using the DH. On a per-game basis, here's how many players had at least 5 at-bats:

30-man leagues: 3.43*
22-man leagues: 3.98
AL 1997-2006: 3.01

* This number is probably a bit low...the csv export for these leagues seems to be missing all batting lines for retired players. Overall, about 7% of at-bats and 8.5% of game appearances are unaccounted for, so the per-game count may be closer to 3.60 or so. (The other set of leagues isn't affected as they're feeder leagues and nobody has retired there.)

Anyway, that's a pretty significant difference assuming it's not limited just to that one season, and I don't think it's simply because my leagues had slightly higher offensive levels. This probably goes some way toward explaining the increased prevalence of 5-hit performances in OOTP.

From what I can tell, OOTP's AI seems to be using fewer players per game than happens in MLB, increasing the probability that a player will stay in for 5+ at-bats, though it's hard to quantify that since none of my leagues used a 25-man roster limit.

[injury log]

Originally Posted by Zeyes

...

From what I can tell, OOTP's AI seems to be using fewer players per game than happens in MLB, increasing the probability that a player will stay in for 5+ at-bats, though it's hard to quantify that since none of my leagues used a 25-man roster limit.

Interesting research. This would certainly explain why there are more 5 AB games. In real life, though, how often does it happen that a player who already has 4 hits in a game is replaced? I have no idea, but if it happens only rarely, then real life substitutions shouldn't have a significant effect on the number of 5 hit games.

[Zeyes]

Originally Posted by injury log

Interesting research. This would certainly explain why there are more 5 AB games. In real life, though, how often does it happen that a player who already has 4 hits in a game is replaced? I have no idea, but if it happens only rarely, then real life substitutions shouldn't have a significant effect on the number of 5 hit games.

Hmm, good point. Even if it's a factor, it can explain only part of the difference anyway... I don't know how much I should trust my sim's one-year numbers, but taking them as correct for now, about 3700 players per 1000 games reached five at-bats, and around 25 of them had five+ hits, or ~1 in 150. For the AL it's 3000 players, 13 with five+ hits, so ~1 in 230.

I have the nagging feeling that there are half a dozen or more reasons for the inflated 5+ hit totals, with each contributing maybe just +10 or +20% on its own.

[injury log]

Following up on thbroman's suggestion above, I did a quick 'back of the envelope' calculation using Zeyes's data. Assuming streaks don't really exist in real life (a player's at bats are independent events), but streaks do exist in OOTP (the fire thing), then if 25% of OOTP players getting to 5 at bats are streaking, they'd only need a 60 point boost in their batting average to explain the discrepancy in Zeyes's numbers. That's assuming my math is ok; I had to make quite a few assumptions. In any event, I've read elsewhere that the fire/streak thing in OOTP does affect performance, and I'd guess it explains at least some of the 5-hit frequency. It would be interesting to do a correlation study of streaks in OOTP and real life to see if there's more volatility in OOTP performance.

[Eugene Church]

Thanks for sharing your research.

You need to send it to Markus.

Good research catches his eye and he often responds to it by tweaking the game.

RonCo and Matter2003 did a lot of research on this subject and thought 5 hits happened too often in the game.

I remember one of them saying in a Beta Thread that OOTPB has too many players getting 4 hits in a game and thus causing too many 5 hits games.

I am not sure I have that exactly as they stated it.

[mh2365]

Dustin Pedroia got 5 hits last night