The teams have won-lost percentages, allowing us to formulate a line’s starting point. We can use a formula to determine how many times a team with a winning percentage of .540 is playing a team with a .480 winning percentage is expected to win. It turns out to be a percentage we can turn into a money line. In this example, a team with a .540 WP vs. a team with a .480 WP is expected to win this game 55.98% of the time. We turn that into a money line that becomes -127. However, the starting pitchers have the most significant effect on a game. We will have to find a way to evaluate them accordingly.
The old-school guys have their lists and values on starting pitchers to add and subtract cents to the money line. So, a pitcher might be listed as a plus 5.25 and another a plus 1.25 to a line. They add the difference between these ratings to their lines. I believe that is virtually all opinion. Even though some of these guys are good with that, I think there are better ways to capture starting pitcher quality, and we’ll go over them in some detail here.
I do this in two ways. The first is to utilize a new Sabermetric measure called WAR (Wins Above Replacement). This stat is explicitly designed to compare two players in value. It is delivered to us regarding how they affect their teams regarding wins, so we can certainly use this! There are companies whose only job is to data mine MLB. I use them to project a pitcher’s WAR performance. The best part is that it is updated throughout the season; thus, it gives the current standard for this pitcher! We want to utilize each player’s current performance rather than overall performance. MLB has 162 games, so there are many performance peaks and valleys; using an updated WAR projection, we can capture those peaks and valleys! A projected WAR number will look like this: 2.7. This means this pitcher has 2.7 wins, which is better than an average pitcher. Because it is shown in terms of wins, we can adjust our earlier discussed money line with the most accurate measure of value possible. This is how Webbie Odds are created! I do this for every game, every pitcher, and every day. I believe this to be the most accurate winning percentage available, which becomes Webbie Odds that we can compare to actual bookmaker odds. We can easily find variances and then act on them OR do further research. I would like to do more qualifications. So, onto Starting Pitchers Data comparisons.
I have a spreadsheet that compiles various metrics. A crucial component of these metrics should be learned. We want to use predictive measures as much as possible! We are trying to predict performance, not assume performance based on past results. So you will not see metrics like ERA, Innings pitched, hits allowed, etc. Those measures are what has been accomplished with no bearing on what will happen today. This is again where Sabermetrics are applied to vastly aid us in predicting today’s performance. Disclaimer: This will not become a Sabermetric definition piece, so if you want more information about the meanings, calculations, and usages for these metrics, please go to FanGraphs and read all about them.
The spreadsheet shows teams and pitchers in the grey area. The following sections are all about Game Score. The following section is the quality of pitching via xFIP, SIERA, and FIP. The following section shows how batters compete against the pitcher regarding xWOBA. The following section shows the pitcher controlling the game from the mound in terms of K/9 and expected strikeouts for this game. The following section dissects the pitcher regarding the quality of pitches and location. The following section indicates the batters’ ability to make contact and what kind of contact they make. Now that we have compiled these, we can further judge how a pitcher will perform today. This, in turn, will allow us to form better, stronger, bettable opinions.
Game Score in and of itself is not predictive at all. It is a terrific form of qualifying past pitching performances that indicate the ability to get outs and limit runs and hits. It also credits strikeouts as part of the 3 True Outcomes in baseball: Strikeouts, Walks, and Home Runs. The goal is for a pitcher to dominate these areas, which then dominates batters’ abilities to produce runs. This is also FIP. However, understanding a pitcher will make roughly 30 starts; we can start to look at his Game Score per game and compare it to his season-long average, his last seven starts (1.5 months and his previous three starts to get a more predictable view as to what his performance today is likely to be. It is more of an art form but still has predictable tendencies that will significantly aid us. If a pitcher has performed above his season-long average over his last three starts, he is likelier to throw beneath that level today. The vice-versa is also true; if he has underperformed in his previous three starts, he will likely outperform in today’s game. It is based on the idea that he will get his short-term average score to his mean average for the season. If we know that a pitcher is likely to be better or worse than his average score and we know what his average score indicates as quality, then we can assume his performance today will be above or below that level. We can then utilize his metrics with an understanding that he will be better than those metrics or worse, which is NEVER baked into the odds from the sportsbook and will give me a tremendous edge!
xFIP, SIERA, FIP
These metrics are designed to give pitching performance more meaningful than ERA. They are all designed to evaluate overall performance. They are scaled and shown in a format similar to that of ERA, so they are comparable to understand. First is FIP and xFIP. FIP stands for Fielding Independent Pitching. It is the “three true outcomes” encapsulated into an ERA format. It measures a pitcher’s performance using strikeouts, walks, and home runs. xFIP is the same thing; however, the predicted result utilizing his measurable makes up FIP. FIP would be used instead of ERA, for example. SIERA takes more into account than FIP and is predictive. It uses the idea that strikeouts are good, even better than FIP, walks are poor, and batted balls in play are complicated. It looks at where balls are hit and their outcomes. For example, balls hit in the air versus balls hit on the ground are included in SIERA. SIERA is an excellent measure of pitcher quality. By encompassing these predictive stats of xFIP and SIERA, we can get a reasonable estimate of how this pitcher will perform in his next few starts. It may not be for today’s game, but the tendency is that he will start to move towards these metrics from where his stats indicate today. Again, these are not baked into the odds from the sportsbook.
xWOBA and L30 xWOBA
WOBA is a batter indication of how the hitter performs overall. It is a better stat because it not only encompasses batting average and slugging, but it also accounts for base percentage! In a nutshell, OBP and WOBA will head toward similar results, but WOBA is designed to show how players contribute to run scoring. Run scoring is essential to us as gamblers! xWOBA is the predictive form of WOBA. It uses the metrics that comprise WOBA and then formulates them to move toward a predictive view. Will these batters perform better or worse than currently? xWOBA guides answering that question. However, we are talking about pitchers. We can see how a pitcher has performed against the hitters he faced using these metrics. I use how the pitcher has done against the team he is opposing today and how he has performed over the last 30 days. It is essential to understand what the hitters on the team he faces have done against him. It is more important to weigh his last 30 days, though. A pitcher who has been doing well in the previous 30 days will likely continue doing well even if the players he is facing today have hit him well. The understanding is that an offense is only as good as the pitcher they are facing today, and if that pitcher is at his best, the offense will not be. Again, these metrics look at today’s game and the games shortly for this pitcher. They are not capturing past performances, but they are predicting future ones.
K/9
I use K/9 because it is innings-based and not number-of-batters-based. Strikeout percentage would be the measure used for batters faced. Although both will lead down a very similar path, I also want to calculate strikeouts for this pitcher for this game. Using K/9 helps with that calculation. I can look at how many innings this pitcher should play, his average strikeouts per nine innings, and the batter’s strikeout rate to create a formula to get a reasonable estimate of how many strikeouts this pitcher should have today. I also want to know how likely the team he is facing is expected to put the ball in play. The less likely a team puts the ball in play, the less likely they are to create opportunities to score runs.
Pitch Modeling
Pitch modeling is new but very adept at predicting the quality of pitches a pitcher throws. Stuff+ measures the characteristics of the pitches thrown. It includes release point, velocity, vertical and horizontal movement, and spin rate. It is a predicted measure of these results so that we can look at the quality of his pitches based on a good day. Pitcher+ captures those same characteristics but also includes the location of where these pitches are thrown. It consists of the handedness of the batters as well. These new metrics are detrimental to future understanding of how good pitchers are. At some point, they could make things like the walk-to-strikeout ratio a stat of the past with no meaning. Utilizing these metrics, we get another view of these pitchers that are not baked into the odds. The better I use this information, the more edges I get.
The following sections discuss groundball rates, balls in play, and hard-hit rates. These are not predictive metrics, but if “read” properly, we can extract useful information by utilizing the averages to return to the mean theories. We want to know if more balls are ground balls. Groundballs produce lets hits and certainly less damaging hits. Groundballs produce double plays as well. We want to support ground ball pitchers where we can. BABIP is very important when put into the law of returning to the mean. A batting average of balls in play will either be higher or lower than the league average. It can remain in those positions for periods; however, at some point, they will work back to the league average. If I know a pitcher has been “unlucky” with BABIP (meaning it is high), he is more likely to start having batted balls hit to fielders instead of going through for hits, especially if this pitcher is good at having a low hard-hit rate. The higher a hard hit percentage is, the less likely a fielder will get the chance to get to the ball as it goes through the infield quickly. Thus, we want to support pitchers with low hard-hit rates. Hard hit rates can also be an indication of damaging hits. Damaging includes extreme hits like a Home Run.
Using this worksheet, I can tie Webbie Odds (in my opinion, the most accurate odds for a game) into a more robust statement for a game. I can play on a Webbie Odds variance and use the predictive measure to exploit it further or even exploit other areas, such as a team total. I can use the worksheet as well. All of these tools provide an edge to me that is not offered in the odds at the sportsbook. They are all proprietary to me. Baseball provides many edges based on predictive data.
I want to mention that I have a premium side to this website. I publish my opinions, data, and wagers that I make on that page. If you want to try the premium page and get my game wagers, please visit this link: https://mlb-daily.com/payment-block-media-and-text/.
If you like to get any daily blog posts, use the free link to get the newsletter.