SSI is an Official Sponsor of the David Pinto Baseball Musings website. We love da bum.
For those who just joined us, David created Probabalistic Model of Range, one of the five top D-metrics out there, so is a card-carrying Top Sabermetrician. We also love his well-rounded approach to analyzing baseball and his easy-going, cool-as-a-cucumber, just-the-baseball, JFro-type style.
David does a lot of things on his site. One of the things he does, is project AL* offenses in a very objective way. He collates for us these two things against all ML teams:
Marcel the Monkey, if I understand it at a glance, is a basic projection system that mainly takes a player's last 3 years and averages them, weighting the most recent years more. (I think this system was used in the first roto mag ever published, circa 1871.) The system is about as good as anything else. It is unbiased, is the main point.
................
Geoff Baker pointed out that Pinto's system expects the M's to add +150 runs of offense in 2010 -- coincidentally, just what Bill James said the Mariners needed to add to win the division.
That can't be right, so SSI clicked over to Musings to see what the problem was. Check Musings' article on the M's, and the Lineup Analysis Tool numbers used, and see if you can spot the problemo in ten seconds or less ...
.
.
.
.
.
.
.
.
.
.
.
.
.
Hm. Does Musings assume that the Opening Day nine play 162 games apiece? Heh! :- )
That's not explained, whether the LAT adjusts for the play of reserves by subtracting .3 runs a game or somesuch. So, to figure out the system a bit, we did something rare: a little work. We tabulated Musings' expected R/game against each team's 2009 runs per game:
Gain per game ... Team ... 2009 ... 2010 projected
Looks like the above projections show an average league gain of +0.25 runs per team, if you edit out the Mariners.
Of course, you don't assume offense to gain 0.25 runs leaguewide. You assume the same level of offense, for these CPA profit-and-loss tables.
So, that's interesting: edit out the benchies and it is 0.25 runs that you gain? Apparently the 4-5 benchies are dragging each team down by about 1/4 run per game?
.
Maybe this intersects a bit with the fact that you can't find a way to calculate fewer than 88-90 wins for the M's.
The 70-run RC gap from last year, plus the 'stealth' eliminations of the terrible 09 hitters, the reinforcing of the bench hitters ... maybe the re-shaping of the M's has, a 2nd winter in a row, simply been beyond the perception of the pundits?
Yeah -- 1/4 run sounds about right to me - as an average number. Obviously, however, everyone doesn't have the same bench, of course. Think about Seattle in 2009. Which non-April 1st regulars got the most ABs for the season? Just the guys that got 100 PAs.
Johson -- (.615 OPS) - 290 PAs
Sweeney - (.777 OPS) - 266 PAs
Cedeno - (.504 OPS) - 206 PAs
Wlad -- (.625 OPS) - 170 PAs
Hannahan - (.656 OPS) - 167 PAs
Josh Wilson - (.693 OPS) - 138 PAs
Bill Hall - (.578 OPS) - 131 PAs
Saunders - (.537 OPS) - 129 PAs
Langerhans - (.697 OPS) - 122 PAs
Jack Wilson - (.562 OPS) - 116 PAs
One guy on the entire list with an OPS above .700. Four guys under .600. That list of 10 players adds up to 1,735 plate appearances. That's 2.5 player SEASONS from the 10 benchies. Now, 2009 was a particularly harsh and volatile year for the Ms with excessive roster turnover. But, typical starter PAs for a season run 600-700 (if healthy all season and playing 150ish games). Seattle had only 3 guys break 600 PAs in 2009.
Benches are ignored mostly because they are stocked with below average players for every team - so the default assumption is that every team has "roughly" equal benches. The assumption is, of course, wrong. Go over to Anaheim in 2009 and they had a grand total of FOUR (4) non-starter players with 100 PAs or more: Macier (.794), Matthews (.697), Mathis (.596) and Quinlan (.614). That quartet managed about 1200 PAs between them with Maicer and Matthews getting 800 of those. Imagine the difference if Seattle had gotten 800 PAs between Sweeney and Josh Wilson instead of the 400 they provided -- and a bunch of those 5-something OPS numbers vanished.
Honestly, given the landscape - subtract 40 runs from ALL the listed run projections and you're probably going to be as accurate as any prognosticator out there.
There are three things I've observed in my most recent article.
ONE: Our bench is ridiculously good by AL standards this year
TWO: Our bench was ridiculously bad last year
THREE: I find it very unlikely that the Mariners are going to struggle to get full seasons out of a number of their key players this year. The guys Z added are more reliable than the guys they replace...so the bench won't be called upon quite as much this year, aside from strategic uses (Sweeney platoon, Tui super-sub, Byrnes helping Bradley).
I think we really are 100+ runs better than last year...I literally cannot find a way to see this as less than a 90 win team right now unless we have catastrophic injuries, get a bunch of major down-spike seasons all at once, or I'm way wrong on the several players for whom I projected a stylistic change or for whom there was not much data to go on when making my projection. (Gutierrez, Lopez, League, Kelley, Texeira)
Is consistently 'beyond the perception' of us pundits. Pretty much all of us.
IIRC, Matty ran an earlier WAR calculation and didn't at first glance spot why the 09 M's didn't win more games... and we found that the Ronny Cedenos, Yuniesky Betancourts, Garrett Olsons and Bill Halls were like -6 below replacement level, once you added them all up.
Teams can and do gain or lose 5, 6, 8 games because of stoploss performances at the end of the roster. This may be why teams like the Yankees win so often - their org depth means they're getting these invisible 4-8 wins all the time.
Probably has been true of the Angels, right?
Very optimistic that the M's will now benefit from this.
My first WAR Stack projection back in late January came to 94 wins...a figure I thought was too high and couldn't understand why I was so high. The reason was largely linked to the bench players being under-represented in my playing time estimates. Even with the above average bench we're carrying this year, it's still a step down from Milton Bradley to the next outfielder or from Chone Figgins to Matt Tui...and the second set of projections does a better job getting the playing time distributed to the weaker players more realistically.
The same tihng happened to me last year...my projection had the club winning 92-94 games...they won 85...and the gap was largely the tremendous negative-win performances of bench players who didn't even get into my projections or got way more playing time than I expected.