Vanishing Left-Handed Pitchers

[grmagne]

I’ve had numerous OOTP seasons where my right-handed platoon hitters played very little because my team didn’t face many left-handed pitchers. It doesn’t happen at the beginning of a league but tends to occur years later after the original pitchers are replaced by a new generation of fictional pitchers. It’s happened often enough that I decided to run a 45-year simulation from 2015-2060 to see whether LHP were vanishing or if this was just my imagination (or just a SSS fluke).

I downloaded 46 years of my simulation data and imported it into Microsoft Access. After I compiled the data the answer became clear: Yes, LHPs do indeed slowly disappear over time. I think this can easily be fixed and I’ll propose a few simple solutions to the developers at the end of my posts.


First I checked the real MLB data from 1980-2015 using the Lahman database. In a typical MLB season about 29-30% of starts and 30-31% of relief appearances are by left-handed pitchers, although this can vary a fair bit year to year. In my OOTP simulation, Rookie league southpaws accounted for roughly 29% of starts and 29% of relief appearances. So far so good; the numbers are very close. However, at each and every level above Rookie ball there are fewer and fewer left-handed pitchers. In my simulation, the major leagues only had 23% of starts and relief appearances by left-handed pitchers. So over the course of a full MLB season, a right-handed platoon hitter will be in the starting lineup about 37 times per year versus 125 for a left-handed platoon hitter. In MLB, it’s about 48 times in the starting lineup for a right-handed platoon and 114 for the left-lander. I think that’s a problem because right-handed platoon players are already getting the short end of the stick for playing time, and this makes it worse.

(I'll post again in this thread to summarize the results of the simulation)

[grmagne]

I ran a simulation of a 16-team major league with five minor leagues: R, A-, A, AA, AAA. It went through dynamic expansion early, so there are actually 20 major league teams and 100 minor league teams through most of the simulation. The league ran from 2005-2060 but I only counted 2015-2060 because I wanted the minor leagues to have drafted players rather than inaugural draft players.

The league database has an enormous sample size: 207,674 player-seasons of hitting stats across 6 levels of play and 46 seasons, plus 119,092 player-seasons of pitching stats. Nobody’s going to convince me I don’t have a large enough sample to reach these conclusions.

I’ve attached a chart below that shows the number of left-handed & right-handed pitchers that appear at every level in the simulation as well as their composite ERA. The interesting thing is that lefty pitchers always have noticeably worse ERAs than righties at every level except the Major Leagues. I think what’s happening is that a typical minor league team has more right-handed batters than left-handed batters and minor-league managers rarely use platoons. So RHPs get an advantage by facing more RHB while LHPs are facing basically the same lineups, except at a platoon disadvantage. That must be causing the ERA differences, because over a 46-year simulation LHP had a 0.14 to 0.17 higher ERA in the minor leagues, and this difference was consistent across every minor-league level.

[grmagne]

I’m not sure if teams are cutting their LHP pitchers more often because of poor statistics or if the lefties get lower morale from their higher ERA (maybe low morale prevents the pitchers from reaching full potential?). Whatever the cause is, LHP always have worse statistics that RHP and they slowly vanish from the minors. The interesting thing is that at the major-league level LHPs suddenly have slightly better ERAs than RHPs. That must be because major-league managers often use platoon hitters and for the first time right-handed pitchers are facing LHB-majority lineups on a regular basis. Also, there’s probably a “survival of the fittest” mechanism for LHP where fewer of them make it to the majors but the survivors had to have higher talent levels.

There’s a similar trend for hitters but the end result works out better. The OOTP drafts have a high percentage of right-handed batters but they slowly disappear from the minor leagues, probably for the same reason the LHP disappear. But once you get to the major-league level the Left/Right/Switch percentages are very close to real life. From the Lahman database, in 2010 MLB was 31% left, 54% right and 15% switch hitters. In my simulation it was 33% left, 57% right and 10% switch hitters. Interestingly, even though Left/Right players go through different levels of attrition I found that every single league from rookie to majors consistently had 10% switch-hitters. I think that demonstrates that platoon disadvantages cause high attrition because if were adifferent factor it should have affected switch-hitters.

I’ll toss out 4 simple suggestions, anyone single one of which should at least partially fix this problem:

1. Increase the percentage of left-handed pitchers generated by about 5%. After the attrition there should still be at least 28% LHP in the majors.
2. Give the user the option to choose the percentage of Left/Right pitchers that are randomly generated.
3. Increase the amount of platooning in the minor leagues (not historically accurate though, since minor league teams are trying to teach their hitters to hit both L&R pitching).
4. Increase the number of randomly generated switch hitters by about 4% and decrease right-handed hitting by about 4%. This would bump the number of switch-hitters close to the modern MLB amount and take away some of the advantage that RHP have in the low minors.

[fcabanski]

I'd like to see the year by year %'s from the sim.

It always seemed to me, just by playing, that left handed pitchers become more and more rare.

[swampdragon]

Or an AI that values being lefthanded, which would have the advantage of mirroring real-life GM preferences. Then more of them would move up the ladder despite their minor league platoon disadvantages.

[injury log]

The best solution may be to determine the reason for the higher attrition rate among LH pitchers, and to fix their development paths, if, as I imagine is true, the real life attrition rate is roughly equal for LH and RH pitchers.

Great research, by the way.

[grmagne]

Originally Posted by fcabanski

I'd like to see the year by year %'s from the sim.

It always seemed to me, just by playing, that left handed pitchers become more and more rare.

Sure, I’ll attach the year-by-year data below.


In real life there’s quite a bit of fluctuation year to year, as shown in the first chart in my original post. Only 24% starters were left-handed in 1995 versus 34% in 2005 (real MLB). Similarly, my league varied from 21%-36% in rookie ball and 17%-30% in the majors. Over the full 46 year simulation it averaged 22.4% in the majors, which I find is about 7 percentage points too low. That causes right-hand platoon hitters to get about 11 fewer starts per year, meaning only 36-37 starts during a 162 game season.


Here are the year-by-year charts for both Rookie and Major Leagues. The Rookie leagues averaged 28.1% left-handed starters during the simulation but only 22.4% in the Majors. The number of LHP in the Majors plummeted very quickly but then increased again around 2044. There’s a lot of year to year variation, so I think the down then up trend was just a fluke in this particular simulation. If you noticed a much lower LHP percentage in the later years of your league then it was probably just random chance. But the observation that there are too few LHP is definitely true.

[NoOne]

it looks like the percentages are saying something, so don't take this the wrong way... (sticking to SP for simplicity in anything i say below unless noted.)

mlb ranges 24%-34% in your small sample of real life data... not much we can do about that sample. you'd probably have to stick with data since 5-man rotations or even after league expansion possible... if the data stays relatively the same throughout history regardlesss of other factors, then it can be lumped together and used as one solid piece of data.

fictional is steadily in the 20%'s. could be a 3-5% difference in averages? * which is really a 12-25% drop in LHPs (SP data for simplicity... if relievers have a different split, then things will be a bit more complicated than just amping up # of created L pitchers).

you were looking at the majors, so you shouldn't look at the minors for comparison - if looking for a "why" it could be useful. unless somethign looks VERY odd, it's probably in proportion with the rest of your game world.

about the Sample:
did you use a real player league to start? if so, the first 35 years there are "real" players still in the league. so, the first 20some years are strongly skewed toward the real-life player. with such a small and tainted sample (as fictional players filter in) you won't be able tell a trend from randomness very well. the data of fictionally created players in a real player league starts in 2035+. be safe, zoom further out - start at 2051. there's just not much you can do with "real" players in the video game as far as data mining - forever a small sample of statistical results and not consistent from year-to-year to due infiltration of fictional players as time goes by. since it's 'like' the mlb, i'd say those 'real' are in the same proportions, which would explain why some see a trend of LHP starts dropping... but it's actually just progressing from real players average to ficitonal players average, which seems slightly lower, then it fluctuates around that new fictional player average LHP starting rate.

even if you started with a fictional league, you must zoom out 30 years before collecting data to use. i know ootp'17 "improved" the initial players that fill the teams per ootp info, but they are still not made the same way as the league will create them in the future, from what i know. that's what i interpretted when i read "improved". (less years for fictional because the amatuer drafts are not filled with pre-made fictional players for X years).

L/R go through the same creation process - even if slightly different splits are created due to handedness. all they'd have to do is increase the % of L created per year. the game modelling does the rest of the work weeding out those that likely cannot play in the mlb.

i doubt it's the gm's influence causing less to be promoted... if they have the ratings, they will play in the mlb barring acts of god. X% of prospective LHP will make it. it's all about the # of created per year, then progressing through the development process and all that entails, just like right handed players. if they are happy with the creation process, it's all about quantity created at that point.

[grmagne]

Thanks for all the info. I took a deeper look at my data and I’m planning to run more years of simulation in a few new scenarios. My plan is to run (1) another league for 66 years and then cut off the first 20 and then (2) a league where minor league managers use frequent platoons and pinch hitters (I hope this is possible) and (3) a different league where I set the pitcher development rating much higher, like 1.5 for example. League number (1) would confirm whether I can duplicate the original results in a new league before I proceed with (2) which would test my theory that an absence of platooning is causing the higher LHP ERAs à lower morale à ruined development. League number (3) would test whether fewer years in the minors minimizes the attrition rate on lefties. They’ll all be fictional leagues with 100% fictional players.

I agree that minor league promotions aren’t the direct cause, I think it’s more likely that they’re not being promoted because low morale (from poor performance/statistics) is ruining their development.


In this WBL simulation, there were very few starting pitchers from the inaugural draft left in my simulation. I began the league in 2005 but my analysis is from 2015-2060 (first 10 years ignored). As I show in these charts, the number of inaugural draft pitchers disappears quickly and I found they affected the final results by less than 0.1%. My next simulations will chop off the first 20 years, but I’m still confident in this first WBL simulation.

[grmagne]

I found another interesting statistic. In real MLB the number of left-handed starters from 1970-2015 was 29.8%. In my fictional WBL from 2015-2060, that percentage was 22.5%. But what I found really interesting was MLB has a fairly steady LHP% across all ages (not a big surprise) whereas in my league the percentage of left-handers increases substantially for older ages. I’ll run more simulations to see if that holds up with a larger pool of data.
So to summarize, and I hope to replicate this in more simulations (and maybe other users can check their long-term leagues):

1. About 29% of amateur draft & IFA pitchers in my simulation were left-handed.
2. In each and every level of minor league ball LHP have a worse ERA than RHP. **
3. At every level of minor league ball the LHP are less likely to be promoted to the next level than RHP. **
4. The relatively smaller group of LHP that reached the majors had a slightly better aggregate ERA than RHP and also have greater career longevity (see chart below on LHP% by age: real MLB vs. fictional WBL).

So something is causing LHP to disappear at a faster rate in the minors, but also to survive longer in the majors.



** Note 2 & 3 above are also true of right-handed batter's hitting stats & promotions, which supports theory that a lack of platoons is hurting the development of both LHP and RHH.

[injury log]

I'm not sure if the AI Eval settings are used when the AI decides when to promote a minor leaguer, but I suspect they probably are. If you're observing that LH pitchers are producing worse stats than RH pitchers in the minors, those poor stats might be inhibiting LH pitchers' rate of promotion. It might be interesting to try a league where AI Eval is set to 100% ratings, and one where it's set to 100% stats, to see if that makes any difference.

[NoOne]

so after real players leave, the player creation and development portion of the game amounts to ~23% LHP starts in the mlb. a 25% increase to get to normal is fairly significant. just a matter of quantity created, most likely. ugh if they change that i am going to have to redo my LTMs.... %$#@#@ everythign is fine here... nothing to see.. move along.. # of lefties is just fine!

i can't say this for all pitch combinations... but...

for the most part, RHP do perform better than LHP in ootp -- all other things remaining equal. no idea if that is true to life or not, but definitely the case in ootp for most pitching combinations. you can even see it in the prediction of ERA in the player profile.

i'd assume it's true to life. the batter 'tends' to have the upperhand if they are the opposite handed to the pitcher. lhp v R is the mirror of a rhp v L. it hapens a heck of a lot more to a lefty pitcher than a righty.

i would guess the greater weight to stats in ai decisions will cause less lefties to be used.