Before the start of every series, barring my availability, comes these previews. They change in shape and format over time, and as they do I will do my best to update this primer. In order to conserve pixels (and make it friendlier to mobile/tablet browsers) and because of the sheer reptition in these previews, I utilize shorthand at times. My hope is that any questions stemming from how the post is organized or what any certain part means will be answered herein. Let's begin with the beginning.
MARINERS (55-64) | Δ Ms | TWINS (50-67) | EDGE | |
---|---|---|---|---|
HITTING (wOBA) | -105.2 (30th) | -5.5 | 15.8 (12th) | Twins |
FIELDING (RBBIP) | 42.9 (2nd) | 1.0 | -11.7 (21st) | Mariners |
ROTATION (TRA) | 14.4 (10th) | 5.2 | -68.2 (29th) | Mariners |
BULLPEN (TRA) | 1.9 (15th) | -0.4 | -13.2 (25th) | Mariners |
OVERALL (RAA) | -45.9 (21st) | 0.3 | -77.3 (25th) | MARINERS |
Here is the overview table. At the top row, you get the teams and their records at the beginning of the series. There is also a Δ Ms column. Δ (Delta) is a commonly used symbol in math to represent change. This column will tell you how much the Mariner values in each row have changed since the last preview. It's sort of a summary statistic of how well the Mariners hit, fielded and pitched over the last couple games.
The rows beginning with hitting. I use the basic wOBA formula, which currently does not include any baserunning events. Below hitting is fielding. I distrust defensive metrics that focus on individual players, especially when their exact inner workings are unknown. The system I use looks only at the team-level and attempts to be park neutral. There are two posts that outline the idea, this first one describes the methodology but leaves out the subsequent addition to control for park and scoring biases. The next two rows deal with pitching and are based around tRA. Graham provided a justification for tRA-like statistics and the actual metric flows directly from that, using expected outs and runs to compute a run average just like ERA does with actual outs and earned runs.
Batter | PA | P/PA | Slash line | nBB | K (sw) | 1B/2B/3B/HR | Sw% | Ct% | Qual+ |
---|---|---|---|---|---|---|---|---|---|
J Jaso* | 35 | 3.9 | .321/.457/.607 | 7 | 4 (2) | 5 / 2 / 0 / 2 | 38 | 84 | 150.9 |
J Montero | 44 | 3.1 | .318/.333/.455 | 0 | 5 (4) | 12 / 0 / 0 / 2 | 53 | 89 | 157.7 |
P/PA = pitches per PA [avg~3.8], nBB = uBB + HBP, Sw = swinging [avg~45%], Ct = contact [avg~81%], Qual+ = a measure of quality of batted balls [avg=100, higher is better]
The data here is totaled over the past two weeks (14 days) and I've cut off some of the rows for the sake of brevity. Normally this chart would include every hitter with at least 25 PAs over that previous time span.
Qual+ is the only column here that I hope is baffling. The rest should be straightforward if you've spent any time around this site before. My aim with Qual+ is to sum up in one number how good the batted balls have been. I didn't want to just break out here's how many ground balls, here's how many line drives, and so on. One, that would make the chart too horizontally large. Two, it's bereft of context. What's a good breakdown? Sure, line drives are good, but how often? Do a lot of line drives make up for a lot of pop flies? Those aren't questions I think many people have an innately comfortable handle on. I don't.
So what I've tried for now is to take all the batted balls (home runs included) and add up their expected run values. Based on run expectancy charts, you can find out how many runs the average ground ball or average fly ball or the average whatever generates. Those values are already used in TRA to figure out the expected number of runs a pitcher would have allowed. Here, I take that sum and divide it by the number of batted balls for the hitter (to make it a rate stat) and then divided that by the league average to put it on a relative scale. So think of it like OPS+, but for batted ball types only.
I don't intend this to be any kind of precise statistic. My only aim is to be able to see at quick glance if a batter has been roughly average, above or below and get a fuzzy sense of by how much. It's like a sanity check on the triple slash line; a hopefully improved version of the "look at line drive rate then compare to BABIP" method. So I wouldn't put too much weight on them
MARINERS | Δ Ms | TWINS | EDGE | |
---|---|---|---|---|
INFIELD | 26.5 (4th) | 0.8 | -33.0 (30th) | Mariners |
OUTFIELD | 16.4 (9th) | 0.2 | 21.3 (8th) | Twins |
RBBIP | 0.292 (1st) | .000 | 0.314 (20th) | Mariners |
OVERALL | 42.9 (2nd) | 1.0 | -11.7 (21st) | MARINERS |
Explainer |
Next comes fielding and though I do distrust individual fielding statistics, the way I compute mine are based on performance against batted ball types (see the link at bottom of table). Since those are fairly easily divided between the primary responsibility of infielders (grounders, bunts, half of the pop flies) and outfielders (fly balls, line drives and the other half of pop ups), I might as well offer that breakdown, right? A RBBIP* in the bottom half of the league gets represented in red.
* reached base on ball in play: BABIP + errors
15 AUG 12:40 |
||
FELIX HERNANDEZ | JEREMY HELLICKSON | |
chart | chart |
Often below the fold comes the game by game previews, highlighted by the starting pitcher charts. First you get the gamedate and time information and the names of the pitchers. An asterisk following the pitcher's name is standard baseball convention signifying a left-hander.
For the charts themselves, they may seem overwhelming and take some getting used at first, but they present a lot of data in a small amount of space. And have color. I like color. Across the top of the chart is information on each pitch that MLB classifies the player throws at least 5% of the time. After the type comes the frequency the pitch is thrown and after that is the average speed. So for example, "CH 21% 85" indicates that 21% of pitches thrown by this pitcher are a change up and that he averages 85 mph on his change up.
Below the pitch information are three bar graphs representing the quality of each pitch. The three measures of pitch quality are "Whiff%", the rate that hitters make contact against the pitch when they swing; "Strike%", the percentage of times the pitch is thrown for a strike; "GB Rate", the rate of ground balls on balls in play off that pitch.
Each pitch is graded against the league median for that particular pitch. That is, change ups are compared against all other change ups. The vertical axis are standard scores (similar to standard deviations), showing you roughly how far above or below the median each pitch is.
The final black bar at the right-most edge of each graph is the overall grade for the pitcher encompassing all his pitches. It is not an average of the individual pitch grades. As mentioned, those are graded on relative curves. The black bar is not. It compares the pitcher's total contact, strike and ground balls rates against all other pitchers.
Reliever | BF | Str% | nBB | Ct% | K(sw) | GB% | HR | Qual- | LI |
---|---|---|---|---|---|---|---|---|---|
T Wilhelmsen | 47 | 66.3 | 3 | 79.8 | 14 (9) | 50.0 | 1 | 88.2 | 1.7 |
O Perez* | 32 | 66.9 | 1 | 78.5 | 7 (3) | 33.3 | 0 | 79.4 | 1.4 |
Str% = strike rate [avg~63%], Ct% = contact rate [avg~78%], GB% = groundball rate [avg~45%], Qual- = a measure of quality of batted balls [avg=100, lower is better], LI = leverage [avg~1.2]
The final bit is a detailed look at the bullpens. These are similar to the batting breakdowns but obviously geared toward what I fancy in evaluating pitchers. Qual makes a repeat appearance but is now Qual- instead of Qual+. It is calculated exactly the same way but the minus sign is used to differentiate that, for pitchers, lower numbers are better than higher
Leverage index pokes its furry little nose in at the end. It is the average of all plate appearances by the reliever, not the average of the leverage when the reliever entered. It's another rough barometer statistic. You can use it to see a hazy outline of bullpen roles and usage. Since relievers are less used than hitters and I didn't want super small samples to make this pure noise, the data is over four weeks (28 days) rather than the two weeks for hitters. Also there's a cutoff of 20 batters faced.