This is the primer I wrote to explain Merlin to TRABL 15 years ago. For NXPL, I have only ever run one pass, rather than all the fine-tuning.

Why Merlin

Merlin exists to address one simple truth – Pitcher grade in the APBA game engine only directly effects how many singles a pitcher allows. So Merlin assigns grades based more directly on hits allowed, not ERA and Wins, which are the basis for the official APBA grades. It also fine tunes the pitcher homer and control ratings.

Merlin is run only on the real-life rosters of MLB. Running it on a draft league’s rosters would be of no value since we are trying to fine-tune the disk to *real life* performance. 

Merlin process

Here’s the process, sprinkled with the logic:

1. I get the APBA disk and place each player on the roster on which he had the most IP or PA.

2. I run Merlin with the default settings. It changes only pitcher grade, HR and control ratings. No changes are made to batter cards or defense.

3. I run three full-season replays. Making key transactions during the season to make sure the players closely approximate the playing time they got for each team they played for.

4. I check out four key ratios: batting average, HR/AB, K/AB, and BB/PA. If those four ratios are within 3% of real-life total for both AL and NL, I call it good. If they’re not within 3%, I re-run Merlin with adjusted values and return to step 3. I also check R/G for the whole league, too. Runs are really what you’re trying to get right, using the other ratios to get there. 

How close the numbers get to real life 

Some years, I only have to run two or three sets of sims. One year, I had to run nine. Takes a long time because I take a lot of care to make sure players get their proper PA totals on the proper teams.

A sample of how close we try to get to real-life totals with the final set of sims:

          AL      NL

Average –.28%   +.55%

HR/AB   -.48%   -.42%

K/AB    +2.86%  +.73%

BB/PA   -1.03%  -.68%

               As you can see, we got under the 3% threshold on all the ratios. The only one that wasn’t within about 1% was that AL K rate. Runs across MLB were up 1.38% in the sims. That’s a raw 281 runs spread over 2400 games. Oddly, that happened with BA, HR, and BB a little low. I think it’s because errors usually run high on the sims.

               Anyway, the end result is, for the whole disk, the batting turns out right on, pretty much. 

Why offense may seem low early in a season 

As to why TRABL offense seems low, I think, yes, individual batters would have slightly lower numbers across the board. But not to a super-high degree with a 24-team league. Of course, the same is true for pitchers, since they are not getting to face the worst six teams’ worth of batters in MLB. Overall, since we use 930 batters, and give MBF 1000 to the worst pitchers, that all should even out in the league totals, and they should be close to real. I’ve never checked, though it wouldn’t be hard. However, the highs and lows of individual performances should be flattened out a bit.

My guess for why this seems apparent has to do with the fact that it’s early in the season. With our AIM set-up, you can use all your best SP innings right out of the gate. And some folks do that, strategically, or through inattention. So, your 14 Z starter with 150 innings can pitch every turn through August, then he’s done. Not so with your 400 PA guy who has a .950 OPS. He has to spread it out over the whole. Put another way, the weighed pitching average for pitchers used in April is higher than it will be in September. How much higher is anyone’s guess, though if you were bored, you could total it up. Sure, a few guys save their high-grade, low IP guys for late, but I think more just burn through the 130-inning guys than save them.

Merlined Wiz disk vs. Merlined APBA disk

                I used to wonder why the Merlined Wizard disk gave different grades (usually higher at the top end, flattened out at the bottom) than the Merlined APBA disk. The Merlin program’s author pointed out to me that Merlin creates the pitcher grades by analyzing the cards of the potential opposing hitters, weighted by the schedule.  To do this it analyzes the hit values of the numbers on the cards, precisely.  It does not use the hitter's stats to analyze. So the stats may be the same, but the cards are different.

               Wizard makes the cards on the original Seitz model for the cards (pre-1989/Mura).  The Wizard cards, then, have more hit numbers.  Need higher grades to tamp down the batting averages.