Welcome to MABLMadness 2020!

---

***MABLMadness is the work of an independent, three-man MABLmetrics team. Our content is not vetted by the TCMABL, nor has the TCMABL explicitly or tacitly endorsed our work. Any views or opinions expressed by our team do not necessarily reflect the official views, policies, or positions of the TCMABL.***

***Sport has the ability to pull us together. In a world where a virus is pushing us apart, the MABLmetrics team thinks we could use sport more than ever. Our modest contribution comes in the form of a sandbox simulation. In the grand scheme of things, the outcome of our simulation couldn't be less significant. But if it provides some of our many friends in the TCMABL an occasional escape from the distressing spectacle of the coronavirus, we'll count our simulation a success.***

MABLMadness is a March Madness style, single elimination tournament simulation involving thirty-nine TCMABL all-Franchise teams.

Games will be simulated using the Out of the Park (OOTP) simulation engine and live streamed through the MABLMetrics twitter account.

We invite everyone to fill out and submit a bracket. The winner of the tournament will receive a $50 gift certificate to Dick's.

Additional information about the simulation may be found below in a Q and A between the MABLMetrics team and Bill Lumbergh. Lumbergh is the Vice President of the Texas-based software company Initech. Lumbergh was prominantly featured in the 1999 documentary film Office Space.

MABLMadness Q and A

---

Bill Lumbergh: Uhhhh, yeahhhh, this weekend I bought 70 rolls of toilet paper for me and the missus. If you could ask your simulation if that's going to be enough, that'd be great.

MABLmetrics: Actually, it doesn't work that way. We didn't create an artificial intelligence. Out of the Park is a very sophisticated baseball simulator that allows you to import players into their simulation engine. The MABLmetrics team converted 13 years of TCMABL player data into OOTP-formatted player files. We then imported those player files into the simulator.

Bill Lumbergh: You turned 13 years of TCMABL player data into Out of the Park players? How did you do that?

MABLmetrics:

Bill Lumbergh:

MABLmetrics: Well, you've probably played a sports video game before. Players are assigned ratings across various skill categories. OOTP define players the same way. We mapped TCMABL player stats to those OOTP categories. The conversion process itself looked like this:

TCMABL player stat=>standard deviation=>20-80 MLB scouting scale=>OOTP 1-250 scale

Bill Lumbergh: Oooo...yeahhhh, ummm...If I'm being honest you may have lost me there.

MABLmetrics: Well, that conversion is important, as the verisimilitude of the simulation will depend on how well we translated real world players into their electronic counterparts. So let's use an example to unpack this.

Power is an OOTP player rating category. We calculated Emery Hull's OOTP power rating as follows:

1. We identified Emery's slugging percentage during his peak TCMABL season.

2. We determined the number of standard deviation between Emery's "peak season" slugging percentage and the average slugging percentage across all TCMABL player seasons. Here is a quick primer on standard deviation. Though for our purposes, it's enough to know that we are using this standard deviation method to position Emery's slugging percentage prowess relative to his TCMABL peers.

3. Here's a fun fact: The 20-80 major league scouting scale is a standard deviation approach to scouting. A fifty score marks an average ability, and each 10 point increment represents a standard deviation better or worse than average. In a normal distribution, three standard deviations in either direction should include 99.7% of your sample. That’s why the scale is 20 to 80 rather than 0 to 100. Fangraphs has an interesting article on the subject.

Anyway, that primer on standard deviation indicates that roughly 70% of a population will fall with one standard deviation of the average and 95% will fall within two standard deviations of the average. This "clustering" around the average is why you see so few 80 power scores: Only the extreme outliers in a population will sit on either "tail" (i.e. 20 or 80) of the scouting scale.

Of course when it comes to the population of TCMABL players, Emery is that extreme outlier as a power hitter. Thus it was no surprise that when we factored his slugging percentage in terms of TCMABL standard deviations and mapped that to the 20-80 scouting scale, Emery scored an 80 in the power department.

4. Finally, with this power score in hand, we used a formula provided by OOTP to convert that power score to the 1-250 scale used by OOTP for its rating categories.

We repeated that process for every peak player season in TCMABL history and all of the mappings between their TCMABL stats and their OOTP rating equivalents.

Bill Lumbergh: MMMMMKay. If I could just get those mappings in a TPS report format, that'd be terrific.

MABLmetrics: That's a negative on the TPS format. But try this.

Bill Lumbergh: So you took all of the stats mentioned in that report and converted them to OOTP player ratings. Can I see what those converted player records look like?

MABLmetrics: Yes. In the Season Stats tab of this web page we allow you to view these converted player records. For each player we included the peak TCMABL season used for conversion purposes, and then the OOTP skills scores that resulted from that conversion. We'll wait here while you go take a look.

Bill Lumbergh: So, yeah. I took a look. The ratings are all over the place.

MABLmetrics: As they should be. Every TCMABL player has a unique set of skills. If we did the stat translations properly, those unique skill sets should render as unique collections of skill scores for their OOTP counterparts.

Bill Lumbergh: Uh, bear with me, I'm sorta playing a little catch-up here. Tell me again how you determined peak TCMABL seasons?

MABLmetrics: For pitchers we based it on the Fielding Independent Pitching (FIP) metric. For hitters we based that on the runs created metric. Specifically, the 'technical' version of that formula, which looks like this:

((H + BB – CS + HBP – GDP) * ((1B + (2*2B) + (3*3B) + (4*HR) + (.26 * (BB – IBB + HBP)) + (.52 * (SH + SF + SB)))))/ (AB + BB + HBP + SF + SH)

Bill Lumbergh: So if I played ten seasons in your league, my peak season would be the one whose statistics, when fed into this formula, produced the biggest number?

MABLmetrics: Yes. Had we used OPS or wOBA for that formula, your peak season might have come out differently. But we're partial to runs created.

Bill Lumbergh: Uh, yeah. So what I'm getting from this is that you turned a bunch of TCMABL players into OOTP players. How do you get from that to teams and a tournament bracket?

MABLmetrics: Well just a second there, professor. As for the team piece, that was simple enough. For every team lined up to play in the upcoming TCMABL season, we created an All-Franchise team consisting of the top fifteen peak offensive seasons from that franchise's recorded stat history and the top ten peak pitching seasons from that franchise's recorded stat history.

Bill Lumbergh: Yeah. But this Humery El guy. If he's as terrific a player as he sounds, wouldn't he have more than one of a team's top fifteen seasons?

MABLmetrics: Emery Hull. And yes, nice catch Bill Lumbergh. No player clones on these rosters. Rosters consist of the top twenty-five player seasons, with each player representing one season, his peak season. Though a player who registered a top hitting and pitching season for a franchise will have an OOTP hitting record and an OOTP pitching record for that team. And a player who played on multiple teams may appear on multiple rosters.

Bill Lumbergh: So not everyone who has played in your league will have a counterpart in this simulation?

MABLmetrics: Not by a longshot. And to be clear, the TCMABL's dead internet era in conjunction with teams' different approaches to record-archiving means there are certainly many elite players from TCMABL history who are not involved in this simulation. Even if this simulation were perfect in every other respect, given the gaping holes in our player database, the MABLmetric's team would never claim that the results of this tournament should reveal the most talent-rich franchise of all time. We're just having fun with the incomplete stat history we have in hand. As with everything else that MABLmetrics puts out there, this tournament should be viewed within that context.

Bill Lumbergh: mkay. If we could get to the tournament now, that'd be great.

MABLmetrics: Almost there. With twenty-five man, all-franchise rosters in place, we were ready to set up the tournament bracket. But how to seed the teams? We might have used historical records. But then we realized that we could use the OOTP simulation engine itself to power rank our teams. What we did was to simulate a 150 game season worth of games between our all-franchise teams. We then seeded our bracket based on the final standings.

Bill Lumbergh: Yeah, I'm gonna need you to get me those final standings.

MABLmetrics: Can do.

Bill Lumbergh: I'm also gonna need you to get me the individual player stats.

MABLmetrics: Yup. The Season Standings tab on this web page displays the end of season standings. The Season Stats tab displays player stats for both hitters and pitchers.

Bill Lumbergh: So if I wanted to handicap the teams for this tournament, I could use those pages for research?

MABLmetrics: Yes. But with a couple of important caveats:

First, many teams with shorter histories do not have ten peak pitching seasons meeting our minimum inning requirement. Even some teams that have been around for a while missed that mark. The Skeeters, for instance, have relied on just a half dozen or so pitchers over the years to cover their innings. For teams like the Skeeters, we created additional "replacement level" pitchers with low staminas, two pitch types instead of four pitch types, and other stats that were 85% of the "real" staff averages. For instance, if the six "real" Skeeter pitchers averaged a Stuff rating of 100, we filled out the rest of the Skeeter pitching staff with pitchers assigned a Stuff rating of 85.

The idea there was to fill out the staff with "Skeeter like" pitchers, but to hopefully diminish them enough in terms of their stamina and abilities so that the simulation engine would default to one of the six Skeeters we brought over whenever one of them was available to pitch.

Obviously that had an impact on the season-long simulation. Teams with "shorter staffs" (i.e. fewer "real" TCMABL pitchers and more of these derived and diminished replacement level pitchers) were at a disadvantage relative to teams with longer staffs. But the takeaway here is that this effect may be mitigated somewhat in a single elimination tournament format. We (or, we hope, managers and/or players belonging to teams in this simulation) will be setting the starting lineups and selecting a team's best pitcher as the starter. The simulation may not rely on much more than that starting pitcher to cover the innings for that tournament game. So if you're looking to pick some upsets, you might want to look at lower seeds with a very top heavy rotation.

Second thing to note: Defensive ratings play a role in the game and we lacked defensive stats to derive them from. So we assigned every player on a team's roster the same rating, and we used this formula to derive the rating:

Runs allowed(by team since 2007)/Earned runs allowed(by team since 2007) => standard deviations => 1-80 scale => OOTP 1-250 number

Essentially, we relied on the team's "unearned run surrendered" rate as an indication of a team's defensive acumen. This produced some outliers -- there were a handful of teams whose ratings surprised us -- but our supposition was that losing teams that may not have scorebooked a lot of errors (and thus who gave up fewer unearned runs, which yielded stronger defensive ratings) would have their history of losing baseball "pass through" into the pitchers stats. Meaning that if our defensive rating didn't catch bad defensive baseball, the pitching rating would have in terms of elevated FIPs.

Finally, we know that these rosters aren't perfect. For instance, our stat processing interwove unrelated franchises with the same name. Thus the Brewers have a couple of players on their roster from the now defunct Brewers franchise of an earlier era. The good news, at least for the Brewers, is that this glitch only helps them, as those past Brewers posted better peak seasons than any current Brewers they pushed off. Similarly, our processes merged the Padres history and the Cubs history into the Angels history. There again that helps the Angels -- that roster is composed of peak seasons from two TCMABL franchises --but its likely that some former Cubs and Padres would have preferred to see their teams represented separately.

There are almost certainly other glitches here and there. And it appears we need to tweak some of our stat conversion formulas. The strikeout rate for both pitchers and hitters, for instance, is a little high. This is our first time doing this. We'll do better next time

Bill Lumbergh: Uhhhhh, yeahhhhh. This is terrific. But isn't it all a little ... much?

MABLmetrics: Good heavens, yes. In italics. MABLmetrics jumped the shark in our very first episode, when in our Palm Spring Profiles we compared Hall of Fame Major League Baseball careers with the baseball talent in a men's adult recreational baseball league. There is an aspect of the fantastical, even the absurd, in our effort to elevate MABLmetrics into something like the FanGraphs of the TCMABL. We're aware of that. We don't take ourselves too seriously. And neither should you.

We've had a lot of fun with this simulation already. We're hoping you fill out a bracket and join us.