At Mets360 we hold these truths to be self-evident:
• That proper statistical evaluation is essential to understanding how baseball games are won and lost
• That the stats that are available today are superior to the ones available in the 1950s for this purpose
• That there is no one perfect stat to evaluate players
The closest we have to a perfect stat is WAR, which attempts to put all of a player’s contributions into one number to indicate how much value he created towards team success. Just to be clear, WAR is not perfect. But it is extremely reliable, which makes it extremely useful.
As you might recall from your math classes, correlation is the measure of the relation between two or more variables. Correlation ranges from (-1) to +1, where positive 1 means perfect correlation, which we rarely find. Generally, a correlation of 0 means that the two variables have no correlation, a positive number less than 0.5 implies a weak correlation and a value of 0.8 or greater means there is a strong correlation.
Jordan Tuwiner over at OriolesNation did the heavy lifting for the 2011 season, comparing fWAR to actual team Wins. He writes, “I compiled a ton of WAR data in an excel sheet and determined that the correlation between WAR and actual wins for the 2011 season to be 0.88.”
Tuwiner had done a similar study in 2009 and found that the correlation was 0.83 that season. He attributes the higher correlation last year due to the inclusion of a baserunning metric into the fWAR calculation last year that previously was not there.
There are still improvements to be made to the calculation, most notably in catchers’ defense. But even as fWAR stands now, we can use it to understand the vast majority of what goes into a winning player and a winning team.
WAR measures Wins Above Replacement, so you may be curious as to how many Wins a team comprised totally of replacement-level players would achieve. Tuwiner ran the numbers for MLB last season and concluded that in 2011 such a team would produce 42.2 Wins.
With that as our backdrop, I copied the Fans projections over at FanGraphs for the eight starting position players and the expected five starters in the rotation. Here is the wisdom of the crowd in fWAR form:
2.5 – Josh Thole
4.0 – Ike Davis
3.3 – Daniel Murphy
4.7 – David Wright
2.5 – Ruben Tejada
2.0 – Jason Bay
3.2 – Andres Torres
2.4 – Lucas Duda
24.6 – Total from starting position players
3.2 – Jonathon Niese
3.1 – R.A. Dickey
2.1 – Johan Santana
1.8 – Mike Pelfrey
1.5 – Dillon Gee
11.7 – Total from starting rotation
If we add the two totals from above to our 42.2 replacement number, we get 78.5 Wins. Now, it’s important to remember that these are projections, not actual numbers. Anyone can go over to FanGraphs and make a projection and I have no doubt that some of these are optimistic. My completely non-scientific way to handle this was not to include anyone else except for the team’s main players. Not included was Justin Turner’s 1.1 fWAR prediction or Frank Francisco’s 0.7 projection and anyone else who would play a game for the 2012 squad.
While recognizing the flaws inherent in this approach, I submit this as another reason why it’s foolish to project 100 losses for the 2012 Mets.
Finally, let’s look back and see how fWAR did in projecting the actual record of the Mets in 2011.
When you added up the fWAR from the players on the 2011 Mets, you came up with 23.5 in hitting and 8.8 in pitching for 32.3 total fWAR. Add in the 42.2 replacement level and you get 74.5 Wins. On the field, the Mets won 77 games. So we have 2.5 Wins that are unaccounted for by fWAR. Perhaps this is managerial impact or clubhouse chemistry or any other intangible. But this is the upper limit of how much those things mattered to the Mets last year.
Now, 2.5 Wins is hardly an insignificant thing, especially for a club that finishes a game or two out of the playoffs. But 97 percent of the Mets’ wins in 2011 can be accounted for in the statistical record of the players. And that’s why we acknowledge that intangibles exist but place very little importance on them.
Interesting read. Right now I look at the Mets being a .500 team, but they are in the one of, if not the, toughest division in baseball. But I’m not certain that will be a bad thing for the Mets. They could improve their performance because of the added sense of competition. I honestly think the Mets can make a run at the Wild Card if everything goes right. Then again, we know as Mets fan, nothing goes right.
Adding up hitting and pitching WAR to project a teams win total can be a bit of an imperfect exercise, because pitching WAR doesn’t include trivial things like pitcher’s offense and defense. That may sound pointless, but R.A. Dickey contributed nearly a whole win on defense.
I don’t think that these projections are too offline. Projecting Bay and Duda to be around 2.5 WAR a pop might be a bit pessimistic, but I don’t think Thole will get quite that high either. I’d say this is a nice baseline for the team. You could have someone like Willie Harris come in and cost the team wins in so many ways, but sometimes bench players and pen arms can be very valuable. Sometimes spot starters and random players somehow become very valuable. Time will tell.
While it’s true that if you go to a pitcher’s page on FanGraphs, the WAR number is just for his pitching — it appears that the hitting contributions for pitchers was used in the calculation of offensive WAR that Jordan Tuwiner used in his study and that which I quoted here.
Pitcher fielding is trickier. FanGraphs uses UZR in its defensive numbers and shows no UZR for pitchers. If we substitute DRS and add up those numbers for the 23 pitchers the Mets used in 2011 we get a total of 5. Since 10 runs equals a win, we can say the fielding by Mets’ pitchers contributed 0.5 WAR.
So, with stats we can account for 75 of the 77 wins of the 2011 Mets.
Ah, yes – Life, Liberty and the Pursuit of Comprehension Through Quantification!
Personally, I have no qualms about the value of analyzing statistical data as tool for predicting and understanding results, and I readily accept your assertion that today’s metrics are superior to those used in the past. I can also appreciate your dismissive attitude toward the so-called intangibles, although I don’t agree with the reason you give for it.
For the sake of clarity, let’s define what comes under the heading of “intangibles”. In my view of it, that term refers to those aspects of the game that cannot be quantified, thereby becoming annoying flies in in the ointment of statistical analysis. You mentioned managerial impact and clubhouse chemistry, but I would add to that such things as: players who seemingly always rise to big occasions (coming through in the clutch) and those who don’t (chokers); players who have hidden injuries or are dealing with personal problems; slumps; dimensions and other variable physical properties of ballparks; the effect of hostile or supportive fans and/or media; the distracting behavior of owners and other administrative personnel; the effects of contracts and contract negotiations; the weather. There are others, of course, but those are just some off the top of my head.
It’s easy to trivialize a lot of that stuff, and players will routinely deny that much of it affects them (“I’m not thinking about my contract right now, I’m just focused on doing what I can on the field to help the team…”) but it all has effect. How much effect? I can’t say. It’s hard to measure such things! I’m fine with acknowledging that, and in fact I appreciate that there are unknown and unknowable factors, intangibles, at play. But then, I’m not a statistician.
I appreciate your ernest dedication to crunching the numbers, Brian, and I think it gives you insight to the game that a lot of people, me included, don’t have. That’s one reason why I read your blog. It wouldn’t surprise me one bit if the predictions you’ve made for the Mets performance this year, based on your analysis of the available statistical information, is real close to being on the money. On the other hand, being a long-time Mets fan, it wouldn’t surprise me one bit either if the Mets did much worse than that. It would be a surprise, a happy surprise, if they did better than anyone predicts and were still in contention in September.
Hey, it’s possible – you never know. Intangibles happen.
Love the humor in your first graph!
I think the majority of the things you mention in the third graph are things that we can measure, certainly in regards to how they affected performance in the actual year they happened. The trickier part is filtering that information and making the proper judgments when making predictions for following seasons. Those are some of the many factors for why making accurate predictions is so hard.
I think a perfect example of that is Justin Turner. He had such a great month of May for the Mets last year, seemingly getting an RBI every time he came to the plate with a man on base. He fell off from that pace, which was unsustainable anyway, but he was nursing an injury. So anybody making a prediction for him for 2012 has to factor in playing time, how his thumb is doing and how he will do in those “clutch” situations – along with what his “true” talent level is at this point in time.
Regardless of how difficult it is to make a prediction on what he will do in 2012, I think we do have the tools necessary to make accurate judgments to how valuable he was to the team’s success in 2011.
Also, thanks for the feedback on the logo. Does the black background instead of the crowd shot work better?
Yeah, I think the banner graphic looks better the way it is now. Here are a couple follow-up thoughts, now that I’m seeing it instead of imagining it:
Black might not be the best color. I don’t know if you have the ability to play with it, but you might try some other solid colors to see how they look. A dark blue or plum color… yellow, maybe? But something solid for certain.
The “Mets 360” logo could stand to be bumped down a little so it didn’t ride so high up in the space it occupies, stopping short of having it come too close to the card on the lower left of the montage. The words “Past Present Future” could come down with it, or stay where they are up at the top… again, it’s easier to look at it and see than to imagine what difference it might make.
Anyway, it looked good to begin with and I think the change improves it. I liked the old one (the one you used before the baseball card version), but Jose being featured so prominently in it was a bit awkward, so I think you did the right thing with your decision to go with a new look.
Thanks for being open to suggestions.
I, too, appreciate your use of statistics to enlighten an old-timer like myself. But (there’s always a “but”), all teams face injuries during the season. The problem is that one can never predict how hard those injuries impact a team (ex: I. Davis, J. Santana). Will the bench, the farm system or a trade help the team mitigate these injuries?
Or, how about a mid-season personnel change that may not have been foreseen, whether injury motivated, or not.
Predictions on how many wins a team will reach, whether they are stat. produced or heart produced, are dicey because of the unpredictable.