Ted Williams: Part II

I received a lot of feedback about my Ted Williams article in which I opined that Williams was the greatest hitter who ever lived and presented my case by combining Williams’ real stats with five years’ worth of simulated stats that were generated by Micro League Baseball, a computer baseball game that came out in the early 1980s. I’ve responded to most of the feedback, but I wanted to use this forum to go into a little more detail and I thought it would be fun to compare Micro League’s output to the projections of others, including Richard Johnson and Glenn Stout, Steve Treder at The Hardball Times, and Bill James’ Brock2 career projection method, which James introduced in his 1985 Baseball Abstract.

Here’s a comment from Mark Parzych, who saw a link to my article at Baseball Think Factory:

“It’s fine if you want to project his stats for the time he missed, but be realistic, especially in HR’s. His career high is 43, which was the only time he hit over 40. You have him hitting 43 again, 50, then 59. It just wouldn’t happen. You can’t just assume he has his best years in his career when he actually didn’t play. I know your point is to say he is the best ever, but the argument gets lost being so optimistic. If you put in conservative data, (which is still great) you will present a stronger case since people just won’t disregard what your saying.”

And here’s my response:

“Thanks for the feedback. I didn’t assign arbitrary home run numbers to Williams, nor did I project them, I ran a computer simulation that split them out. Granted I had to enter his statistics into the computer that were used during the 1943-1945 seasons and I had to make a guess as to what they would have been. I ran this simulation about eight years ago and I no longer have the program that I used, but I’m pretty sure I assigned about 35 homers to him in 1943 and went from there. I based his numbers on his averages from the seasons that sandwiched 1943-1945.

Secondly, and I should have explained this better, Williams was the only player I “brought back” from WWII, so he was facing depleted pitching staffs, which surely helped his numbers. I could have pretended there was no war and entered stats for all the players, but frankly, I didn’t want to go to that length.

Lastly, authors and historians Dick Johnson and Glenn Stout projected his stats filling in the numbers from the 727 games he missed and the numbers my computer simulation spit out are very similar to their projections. Runs differed by only 55, hits by only 20, and doubles, triples, walks, and strikeouts were almost identical. Homers differed by 49 (Johnson and Stout projected those to 686) and RBIs by 214, so your point about his 50-homer seasons is well taken. I may try running the simulation again with Out Of The Park Baseball, which has a career feature that will adjust Williams’ ratings based on his career to the point he went into the service. That may be more accurate and less optimistic, but I’m pretty happy with Micro League Baseball’s results.”

To be honest, I don’t recall exactly how I determined what stats to use for Williams from 1943-1945 and again from 1952-1953; this simulation was something I did eight years ago. If I remember correctly I studied the years prior to Williams’ departure and then the years after his return and used averages based on those numbers. For example, Williams hit 37, 36, 38, and 32 homers, respectively in the two years prior to 1943 and the two years after 1945, which comes out to an average of 35.75 homers a year. Since he was more or less the same hitter upon his return from the service that he’d been before he left, and his career to that point was marked by a remarkable consistency, I naturally assumed that had he played from 1943 to 1945, he would have put up numbers similar to those he produced in the years sandwiching the three years he missed.

Once I averaged out all of his stats, I input the numbers into Micro League Baseball, the left it up to the computer to decide how to interpret those numbers.

One reader at Baseball Think Factory brought up a valid point, insisting that having Williams face depleted pitching staffs “spoiled the exercise.” I can’t say that I disagree. I wanted to restore all the rosters to their pre-war status, but that would have meant estimating stats for more than 300 players and I wasn’t prepared to do that at the time. According to Treder there were approximately 149 pitchers missing from big league rosters in 1945. Had I restored those hurlers to their respective rosters, the results surely would have been different. I am planning on restoring all the rosters and simulating those years in the future, however.

Another reader at BTF found it hard to believe that Williams would have hit 15 homers against a 1944 St. Louis Browns pitching staff that surrendered only 58 round-trippers all season.

My response to that was:

“I wish I had kept better records or had the disks with the stats on them, but I don’t. Bob Muncrief and Sig Jakucki surrendered 48% of the Browns’ total home runs in ‘44; maybe Williams teed off on them in the simulation. Unfortunately I can’t tell you whom he hit the homers against. I’d be interested to know who hit the most homers against that staff, though.”

Out of curiosity I did a little digging and came up with six instances in which Williams hit 10 or more home runs against an opponent in a season.

The closest he came to his 15-homers-against-a-staff-that-only-allowed-58 simulated feat was when he belted 11 against an Indians staff that surrendered 84 in 1946. I checked my SABR Baseball List & Record book to see if there was a category for most home runs by a player against a team in a single season, but I didn’t find one. I’m assuming that information can be found at baseball-reference.com, but there database only goes back to 1957 and I’m more curious about the years prior to that.

Anyway, the method I used for determining what stats to enter into Micro League Baseball wasn’t the most scientific and it pales beside Steve Treder’s method, which I’ll get to in a minute, but the output was very similar to others’ projections, with the exception of Williams’ inflated home run totals.

Here’s Treder’s methodology as explained in his article “War Begone,” which he wrote in 2005:

The methodology for projecting stats for missing seasons is essentially to weight the closest actual season as twice that of the next closest, and the second as twice the third closest. In practice this means closest seasons comprise 57% of the projected season, seasons two years away comprise 29%, and seasons three years away comprise 14%…

…The formula for a player who missed only the 1943, 1944, and 1945 seasons (such as Ted Williams):

For 1943 stats: 1942 x 0.57144, plus 1941 x 0.2857, plus 1940 and 1946 x 0.0714285
For 1944 stats: 1942 and 1946 x 0.2857, plus 1941 and 1947 x 0.14285, plus 1940 and 1948 x 0.0714285
For 1945 stats: 1946 x 0.57144, plus 1947 x 0.2857, plus 1942 and 1948 x 0.0714285

The actual 1942 stats of all players were adjusted by the following amounts, to bring 1942 to the combined average offensive levels of 1940 + 1941 + 1946 + 1947: Runs and RBI, 1.072; Hits, 1.019; Doubles, 1.072; Triples, 1.115; Home Runs, 1.406; Walks, 1.06; Strikeouts, 1.102. At-bats were increased equal to the number of additional hits.

Instead of using straightforward averages the way I did, he took the extra step of weighting the seasons, which certainly makes sense. Treder’s results are more conservative than mine—Williams hit “only” 36 home runs in 1945 in Treder’s study whereas he blasted 59 home runs in the Micro League simulation—but Williams’ career totals in Treder’s article are extremely similar to what I came up with. So are Johnson and Stout’s projections and the output of James’ Brock2 system.

Here are the four projections for comparison’s sake.

No matter which set of projections is most accurate, the fact is no one truly knows how Williams would have done had he played; we can only venture a guess. But it’s fun to speculate and it would have been interesting to see how close he would have come to 700 home runs and whether or not he could have passed Babe Ruth on the all-time list. Treder projected that Williams would have finished with 709 home runs and asked “…think he would have played one more year?” I can only imagine.