Q: How is the new engine different?
A: Unlike simbasev2 and strat, which use half-pitcher-card, half-hitter-card types of outcomes, simbasev3 is based on log5: the probability of an outcome is proportional to the product of the pitcher's probability and the hitter's, not their sum. This allows for more extreme distributions of outcomes: you can literally have pitchers who never walk anyone, for instance. Also, it results in interesting interactions. For instance, ball-in-play hitters will likely do better against pitchers who don't walk anyone, and groundball hitters will do better against flyball pitchers (and vice versa).
Q: What are some practical results of this?
A: One practical result is the existence of cool degenerate stat lines. Another thing worked in here is Voros McCracken's DIPS and its children: pitchers and hitters have equal control over batted ball types, but hitters have more control over the outcome of, say, a ground ball. DIPS is only in half-way, as in real life pitches don't have much control over line-drive rates. But I wanted more skills.
Q: Fatigue was totally broken in simbasev2. Have you fixed this?
A: Sort of. The horrible "fatigue loops" that one could get into in simbasev2 have been fixed by a sort of exponential unfatiguing, so the stupid mop-up man who always had 10 fatigue doesn't exist. Also, pitchers' fatigue points have diminishing marginal returns, so there is a reasonable upper bound of crappiness. Incidentally, hitters also have fatigue in simbasev3.
Q: Are there injuries in simbasev3?
A: No. However, fatigue is random enough that players can get fatigue spikes, which can be thought of as injuries. Also, you need to play your bench players because your hitters get fatigued.
Q: What season length will happen in simbasev3? How fast will games be played?
A: Subject to change, we will aim for a season length of around 45 games. In general, games will be played two-a-day, meaning that the regular season will take about 22 real days. Playoffs will probably add on another five days or so, and we'll have another week or so in the offseason, so maybe just over a month per season.
Q: Are there minor leagues in simbasev3? How will transactions work?
A: Yes. There will be some reasonable rules in place (e.g. similar to MLB's). At press time, no minor league stats are generated, but this might be changed.
Q: Retirement was totally broken in simbasev2. Have you fixed this?
A: Yes. Unlike simbasev2, players will never retire if they are tendered a contract. Players placed on waivers who are not on a major league roster at the end of the year will retire. So, players will play as long as someone wants them.
Q: How will aging work?
A: Players will start at age 15, when they will all be almost-identically crappy. All players in draft pools and on initial roster have been aged up since age 15. Aging is designed so that a player reaches his peak in every attribute at age 25. The aging curve is quite steep outside of the 20-to-30 range, and somewhat less steep inside it. It would be extremely surprising for a player to be of much use after age 35. You can find an empirical study on player OPS by age (average and standard deviation) in this directory. Players will be aged once at the end of every season.
Q: How will the draft work? How will rosters work?
A: Unlike simbasev2, there will be a much larger draft pool than the number of spots available on rosters -- five players per team with random ages between 17 and 21, and random positions, will be generated each draft. The draft will be four rounds -- roster limits will be 21 on the major league roster and 5 on the minor league roster. Since players are "designed" to be useful from 20 through 30, and since players are never forced to retire, many late picks may simply hit waivers immediately (after a spring).
Q: Tell me about the year-end scouting reports.
A: You will get reports from your scouts estimating the OPS magnitude of the offseason changes to your players. You will also get a scouting report on the draftpool: estimated AVG/OBP/SLG along with secondary stats (speed/arm/defense). All of these are based on simulation. The OPS changes are based on 2000 simulated PA before aging, and 2000 after aging. The draftpool estimates are based on 300 simulated PA, i.e. three springs' worth.
Q: What is spring training?
A: At the beginning of each year, you will get to see some spring training stats: AVG/OBP/SLG, as well as the secondary stats (speed, arm, defense). These are from 100 PA of simulated data (200 for extended spring training the first year.) You can form your roster manually, or the auto-manager can do it off of the spring training files. Be warned: it goes only off the spring training stats for the past three seasons, and does not consider regular season stats at all, so if your #1 starter has an awful spring, he may be cut (that said, this is unlikely).
Q: How reliable is this simulated data?
A: Unclear. It is run using the same game engine, but against a vanilla defense and pitcher/hitter which is probably substantially worse than the major league average will end up being. In the test league, it seems somewhat more unreliable than actual data, but not hugely. I don't think we know the answer to this yet.
Q: What do the PR/FR/AV/VG/EX ratings mean?
For speed and arm, the cutoffs between ratings are 35, 45, 55, 65. For defense (at the primary position), the cutoffs are 45.5, 49, 52.5, and 56. Note, however, that the ratings are based on a certain number of random "checks" against the skill in question, so those are not hard cutoffs.
Q: What control do I have over the manager?
A: You have no direct tactical control. What you can tell it is, at present: lineup, which player to use as a backup at each position, starting pitching priority chart, relief priority chart. You can also tell it which backups you want used as pinch-hitters or defensive replacements. Even if you don't, because of hitter fatigue, you can expect your backups to get quite a bit of playing time.
Q: Tell me about the pitching.
A: You can see the format in the .mgr files. For starters, it will go down the list adding the handicap to the pitcher's fatigue and will start the pitcher with the minimum. Order is used to break ties. So, for instance:
15,0 16,1will start pitcher 15 on 0 or 1 fatigue, then pitcher 16 on 0 fatigue, then pitcher 15 on 2 fatigue, then pitcher 16 on 1 fatigue, etc. If you keep those lines but reverse the order (basically, switching the positions in the list on the manager webtool), pitcher 16 on 0 fatigue will be preferred to pitcher 15 on 1 fatigue.
Relievers are indicated in the same way. In high-leverage situations, the auto-manager will pick the best available pitcher. In low-leverage situations, the auto-manager will pick the worst available untired pitcher. There is no way to indicate a "closer", but the best pitcher only comes in in severe leverage situations, which are more likely to be late-game (but are more numerous than the name makes it sound). Short relievers are designed to pitch 1 inning, middle relievers 2-3 innings, and starters 6 innings, but pitchers will be able to pitch more if economical with their pitches. Pitch counts exist and are relevant; pitches per batter are quite different between pitchers (each pitch is simulated, not just each at-bat as in simbasev2).
Q: How many starters is the game designed for? What should my roster look like?
A: The recommended roster is 13 position players, 3 starters, and 5 relievers, with 1 of the relievers a middle reliever potentially capable of making spot starts. It depends on your personnel, though. Even 1 starter would be steady-state doable, although he would always have so much fatigue as to be terrible. Fewer starters (or relievers) just means a higher equilibrium level of fatigue for each start. If you have 4 starters, you will probably have every start go with 0 fatigue, but with 3, you may be carrying 1 per start or something. I have not investigated a lot. Play with it. (EDIT: In simbasev3 so far, most people seem to be carrying 9 pitchers and 12 position players.)
Q: Can I use a pitcher both as a starter and in relief?
A: Yes, but the right way to do this is to have a full-time reliever who makes spot starts. If you designate a starter as the back end of the pen, they won't only be used in desperation. The auto-manager will think that this is simply your crappiest reliever, so they might get innings in a blowout, thus preventing them from making their next day's start. So, a reliever making spot starts will work fine. A starter making spot relief appearances will sort of work, but the auto-manager will consider the pitcher as a primary reliever and in particular will not consider when the pitcher might be needed to start tomorrow. (This may or may not be what you want.)
Q: Is there a DH?
Q: What stats are being kept?
A: Most things you would want are being kept. Check out a stats page. I am dictatorially imposing some scoring changes: a starter doesn't need to pitch 5 innings to get a win (in extreme blowouts, a pitcher may be pulled earlier if the crappy members of the pen are rested), and there is no 3-inning save. Also, sacrifice flies count as at-bats. Other than that, I believe that all scoring is as in MLB.
Q: Can I change my players' positions?
A: Yes, and this is important. A player's listed position only matters during the aging process (except for how the automanager uses it as a backup). Every player is rated at every position defensively, so you can play any player anywhere. During aging, each defensive skill is adjusted based on the player's assigned position -- he will, relatively speaking, get better at his assigned position, tread water at related positions, and get worse at unrelated positions. All of that is infused with a healthy dose of randomness and age. Players' initial positions do matter, because they are aged up from 15 playing that position, so they have a head start, but if you switch them young, they will become playable at a new position (though again, over time, not immediately). Between the end of the season and aging, you can reassign positions to as many players as you want.
Q: What positions are related? How important is defense at various positions?
A: All outfield positions are related, as are all infield positions other than catcher. Catcher is unrelated to anything. Defense roughly parallels what it is in MLB. SS, C, and CF are the most important defensive positions. 1B is extremely unimportant (in every way, it's 1/3 as important as SS.) So, if you stick a poorly rated defender at SS, it will suck; if you stick one at 1B, you will basically be okay.
Q: Do speed and arm matter on defense, or just defense rating?
A: It depends on the position. Arm is very important for a catcher. Speed and arm are both relevant for outfielders, in addition to defense rating. However, for infielders, only defense rating is relevant.
Q: Can I also convert my reliever to a starter or vice versa? What is the point of that?
A: Yes. The tradeoff is that relievers become better (since they can air it out for fewer pitches) at the expense of stamina. This change happens immediately, but any switched player will have a penalty of -2 to start with, which will go away as he becomes familiar with his new role.
Q: What offense level will there be in simbasev3?
A: I'm not exactly sure. I tried to tune it to about 5 R/G. In the test league, things seemed to average about 5.5 R/G. However, I think that offensive levels with human GMs will be somewhat lower: the auto-manager does not take defense into account, for one, so human GMs taking defense into account will presumably field better defenses and worse hitters, reducing scoring. Year 1 of simbasev3 had an average RA of 4.50.
Q: Is there handedness?
Q: Box scores and game logs and stuff?
A: Yes. We have plenty of stats. We even have a replayer which replays games!
Q: Does my scouting report include conversion?
A: The scouting report includes decay of the conversion penalty, but no effect from conversion that offseason. So if I convert a reliever to a starter, his bonus is changed from +3 to -2. The scouting report OPS differential does not include this 5 fatigue. But the -2 decays (25% on average), and that decay is included in the OPS differential (i.e. the before sample is before decay but after conversion, and the after sample is after decay).