Owing to the width of most of the many data tables on this site, it is best viewed from a desktop computer. If you are on a mobile device (phone or tablet), you will obtain a better viewing experience by rotating your device a quarter-turn (to get the so-called “panorama” screen view).
The Owlcroft Baseball-Analysis Site
Baseball team and player performance examined realistically and accurately.
You can get a site directory by clicking on the “hamburger” icon () in the upper right of this page.
Or you can search this site with Google (standard Google-search rules apply).
This page explains in detail the significances of the various tables, their headings, and their data lines, as appear on the site page Major-League Team
Performance Tabulations.
The Manifested-Ability Table
This table—perhaps the most important on the site—shows the 30 major-league baseball teams listed in order of their expected win percentage as calculated from their TOP and TPP numbers, with that percentage and the number of wins it would mean over a full 162-game season shown. In this Table, you can thus see how each team ought to finish if it were to continue playing at the same quality level it has exhibited so far. It looks like this (except the sample below just shows a few team lines, while the full table has all 30 teams listed):
You will note—especially in the earlier part of the season—that some teams’ actual results are quite a ways off the projected win percentage. As you will by now understand, those error sizes will shrink as the season wears on, but…they provide a good leading indicator of what will often happen later in the season. That is, a team whose present win percentage is wildly off at the moment is fairly likely to close up the gap with time; so, a team that “should” be playing .550 ball but is currently under .500 can be tagged as likely to go on a big winning surge at some point, while teams far ahead of their projected win rate will frequently have long losing streaks in store.
That kind of prediction is by no means sure. While teams well behind their projection can “catch up” with a win streak—bringing their real percentage up to their projection—they can also “catch up” by starting to play more poorly than they have—which will bring their projection down to their actual accomplishments. But extreme differences do normally even out, so any such big difference thus usually signifies a team whose fortunes are, in one way or another, due for a significant change as the season wears on.
(And remember that expected deviations only go down as the square root of the sample size goes up: that is, it takes four times as much data to merely double our confidence level. Putting it another way, the error rate a quarter of the way through the season will be roughly twice what it will be by seasons end.)
Note that each team name in that table is a click-on link to that team’s full performance-data page on this site.
(We don’t reproduce those tables here because each is utterly self-explanatory.)
The “Projected (not actual) Standings” Table
Because the projected win percentages table is arranged in simple descending order, we provide this little interpretation of what those projections might signify toward actual eventual finish positions. As the text above it plainly says, It is not a “prediction” of how the races will end up: it is an assessment of how things would stand now if not for luck (as opposed to performance). It is a “prediction” only to the extent that if all teams continued to play from now till the end of the season at exactly the same level of performance they have exhibited so far—which is unlikely—this is how things would most probably turn out.
For each team, the overall “Runs” heading (light grey) comprises two broad subheadings: Runs Scored (salmon color) by the team on offense, and Runs Yielded (bright green) by the team on defense (those include all runs, earned or not). Each of those subheadings is in turn further subdivided into the categories of Real runs (light red or light green)—meaning what you will find printed in your daily newspapers—and expected runs (dark red or dark green) calculated from the appropriate set of team statistics (batting or pitching); the calculated runs are labelled by the Owlcroft name of the measure used: TOP for runs scored (Total Offensive Productivity) and TPP for runs yielded (Total Pitching
Productivity).
Just to clarify: there are no adjustments of any sort applied to the team stats used. What we are doing in the Table is applying the “raw” Owlcroft formulas for calculating real-world runs from real-world stats.
There is, of course, another way of looking at the Table. If you have not accepted as yet that baseball analysis really is as reliable as, let’s say, nuclear physics in its results, this display gives you a chance to look over just what numbers the analyses do put out. As we keep saying over and over on these pages, this kind of analysis is just as precise as the statement that a tossed coin comes up heads half the time: the results are “true” in the exact same sense, in that the longer the data run—coin tosses or games played—the more accurate the predictions become, ultimately approaching (but, not even in theory, ever quite precisely reaching) perfect accuracy. Putting it yet another way, we know awfully well the precise impact of virtually every stat in baseball on the ultimate bottom line—the scoring of runs and the winning of baseball games. (You can, and should, also look in on the graphic demonstration of accuracy page.)
The Actual Wins column (medium blue) is self-explanatory. The other five columns, subsumed under the broad heading Wins Projected From (blue), are what the title implies: they are the number of games won, out of games played so far, that we would normally expect from some given pair of runs-allowed/runs-yielded numbers. Here again, “normally expect” means the same as it does when we say we “normally expect” a tossed coin to come up heads half of the time.
We have calculated and displayed projected wins from two sets of scored/yielded pairs: the team’s actual, real-world, in-the-newspapers R/OR results, and also the calculated TOP/TPP values as they appear in the columns to the left. The R/OR results merely reflect the games-won formula’s predictions from the real-world results; the much more interesting TOP/TPP results reflect the entire baseball-analysis apparatus operating on the panoply of various individual stats to produce a projected games-won result, by first projecting runs scored and allowed, then projecting wins from those projected runs.
Next to each of those two wins projections, we show, for convenience, the degree by which the team’s actual wins record differs from the projection in question; we call that difference the “error,” but that is not a very descriptive term in this instance, since nothing is actually “in error” at all.
The way we have done the “errors” columns, a positive number (no minus sign in front of it) means the projection is over expectations by that many games based on the source runs data (that is, the real or the projected runs); a negative number means that the projection is under wins expectations for that source basis. We also added a second TOP/TPP errors column labelled Seasonal, which is the TOP/TPP wins “error” pro-rated out over 162 games; in other words, if the team kept on playing at the same performance level for the rest of the season and had exactly the same level of luck over that period as it has so far, that number is how many games ahead of or behind what the true quality of their play justifies they would actually end up at.
Again, for emphasis: the + and - signs refer to the projections, not the teams’ performances. To reckon the teams, just mentally reverse the signs: a seasonal-total wins projection of +2 means the projection was 2 wins high, so conversely the team was (or rather, is expected to end up) 2 wins under expectation.
As you look over the first part of the main Team-Performance Table (as explained above), you will see that in some cases the projected runs are almost exactly equal to the actual runs, while in others there may be a substantial difference. The same applies about the two games-won projections. Moreover, sometimes the two wins projections will be exactly the same and other times they will differ by a fair amount. So that you can get a good feeling for the overall closeness of the results, we have set forth some summary values for the overall errors.
In general, for any kind of deviation from expectation you want to know two basic things: how large or small is the average size of the error, and how close to correct is the overall average error.
To better visualize those, think of a person shooting a gun at a standard “bull’s eye” type of target. The average size of the error will be the averaged distance of each bullet hole from the dead bull’s-eye; after all, a bad shot could scatter bullets all over the place, none even near the bull’s eye, but if they make, roughly, a large ring around it, he could say, truthfully, “Well, on average I was just about dead on.” The size of error measurement shows up misrepresentations like that.
On the other hand, a shooter might make a pattern of bullet holes that is very, very tightly spaced, but centered on a point a moderate ways off the bull’s eye; he could truthfully say that the average amount by which he missed wasn’t very large at all—even though he was consistently biased off the true target. The overall average error measurement shows up that kind of “miss.”
Here is what the Error Sizes table looks like and means:
“Error” Type (calculated vs. actual)
Absolute size of “Error”
Percentage size of “Error”
Cumulative “Error” (+ and - cancel)
Batting Runs, from TOP
16.6 runs
2.44%
-12.87 runs
Pitching Runs, from TPP
15.7 runs
2.21%
-11.9 runs
All Runs, from TOP&TPP
16.15 runs
2.33%
-12.39 runs
Games Won from R/OR
2.63 wins
---
0.17 wins
Games Won from TPP/TOP
2.93 wins
---
0.07 wins
In the “error” listings, we show both kinds of error for both the TOP and TPP values, as well as for each of the types of games-won projections. Keep in mind that while all of the work we do is interesting, the brass ring on this carousel is getting games won from the assorted individual offensive and defensive/pitching stats. Thus, the numbers for the TOP/TPP games-won projections are the crucial element; if they are acceptably correct, all the other errors are immaterial except in an academic sense.
The average error size of the games-won projections, working from actual data from 3,656 team-seasons, is a mere 3.77 games (exactly 3.7664113785558 games), a satisfying result. (More than half of the errors were 4 runs or fewer; the few large erros pull the overall average up.)
That wins-error figure is little greater than the bare 3.25 games of the games-won formula itself. That is, if we used the games-won formula on actual (not projected) Runs and Opponents’ Runs, it alone would give a 3.25 games error. So why isn’t the error when using projected Runs and Opponents’ Runs (2.39% error) more than a mere half a game greater? It’s because some of the errors in each formula tend to somewhat cancel out.
We will leave it to you to judge—keeping in mind that the error sizes all go down as the season wears on and thus aren’t at their true values until the season is over—whether or not modern baseball analysis can tell how many games a team will win from just the usual line of stats as published in the papers.
Keep in mind that while the formulas used here include use of a great many baseball stats—the full laundry list is at-bats, walks, hits, total bases, sacrifice bunts and flies, hit batters, stolen bases, and caught-stealings—nearly the same accuracy can be had from only a small subset of that list: at-bats, walks, hits, and total bases. Using just those four stats, we can get an average predicted-runs error, over the same 3,656 team-seasons, of 3.54%. That’s above the 2.39% error rate of the full formula, but only by about 1%—really not bad at all for a “quick and dirty” approximation.
Having come this far on the site, you may be wondering where to look next. We recommend that before you cut to the daily data pages, you look over our discussion of rating pitching, so that the results we present will have meaning for you.
) in the upper right of this page.