<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Valentin Schuh on Medium]]></title>
        <description><![CDATA[Stories by Valentin Schuh on Medium]]></description>
        <link>https://medium.com/@vschuh?source=rss-349b8b39602e------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*EsyACj5Vxv99sETXmyWo2Q@2x.jpeg</url>
            <title>Stories by Valentin Schuh on Medium</title>
            <link>https://medium.com/@vschuh?source=rss-349b8b39602e------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 01:18:29 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@vschuh/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com" rel="hub"/>
        <item>
            <title><![CDATA[AAA(+) and RAI: Baserunning aggressiveness in two and a half stats.]]></title>
            <link>https://medium.com/@vschuh/aaa-and-rai-baserunning-aggressiveness-in-two-and-a-half-stats-4329aac12625?source=rss-349b8b39602e------2</link>
            <guid isPermaLink="false">https://medium.com/p/4329aac12625</guid>
            <category><![CDATA[sabermetrics]]></category>
            <category><![CDATA[softball]]></category>
            <category><![CDATA[sports-data]]></category>
            <category><![CDATA[baseball]]></category>
            <category><![CDATA[sports-analytics]]></category>
            <dc:creator><![CDATA[Valentin Schuh]]></dc:creator>
            <pubDate>Mon, 12 Oct 2026 20:43:28 GMT</pubDate>
            <atom:updated>2026-10-12T20:50:44.201Z</atom:updated>
            <content:encoded><![CDATA[<p>As a coach, an opponent’s baserunning behavior is one of the easiest aspects to prepare for. Whether in baseball or softball, you can adjust your defense to be ready for steals through pickoffs, backpicks, or even pitchouts.</p><p>However, the common metrics for baserunning, such as stolen base percentage, or more advanced metrics like wSB or even UBR, do not really help in knowing whether someone steals <em>a lot</em>. Even the raw number of stolen base attempts (SB+CS) doesn’t really quantify how aggressive someone is — at least not without sufficient context of the league or the player. For batting stats, wOBA captures the tendency of a league in terms of batting, because using isolated OPS was not enough to understand a player’s tendency within their environment. For the same reason, stolen base attempts fail to capture a player’s <em>actual</em> tendency.</p><p>Now there is already <strong>wSB (</strong><a href="https://library.fangraphs.com/offense/wsb/"><strong><em>Weighted Stolen Base Runs</em></strong></a><strong><em>),</em></strong><em> </em>which, if you look at its formula, already compares the player to the league tendency:</p><blockquote>wSB = (SB * runSB) + (CS * runCS) — (lgwSB * (1B + BB + HBP — IBB))</blockquote><p>However, besides the already mentioned issue that it rewards success and doesn’t analyze a player’s tendency, there is another flaw in its simplicity. It essentially looks at every plate appearance where a player reaches first base and treats each as one uniform opportunity to steal. A very simple formula that is good at estimating a baserunning <em>value</em>, but it really does not help if you want to know whether a player is <em>going</em> to steal.</p><h4>AAA — Attempts above Average</h4><p>So, to answer the question of whether a player is going to steal a base, we need to look at it in a different way. For that, I have created a metric called <strong>Attempts Above Average (AAA)</strong>, which is exactly what it is named: a decimal number telling how many stolen bases above average a player has attempted.</p><p>How do we define the average? We look at a league and every situation in terms of outs, base occupancy, and which runner we want to know about. We group a few base situations and then have a set of matrices that look like this:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2F3wb0a%2F2%2F&amp;display_name=Datawrapper&amp;url=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2F3wb0a%2F2%2F&amp;image=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2F3wb0a%2Fplain-s.png%3Fv%3D2&amp;type=text%2Fhtml&amp;schema=dwcdn" width="600" height="206" frameborder="0" scrolling="no"><a href="https://medium.com/media/a7e99a3e14ed2b70ac287ba218bede97/href">https://medium.com/media/a7e99a3e14ed2b70ac287ba218bede97/href</a></iframe><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FmfpDm%2F3%2F&amp;display_name=Datawrapper&amp;url=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FmfpDm%2F3%2F&amp;image=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FmfpDm%2Fplain-s.png%3Fv%3D3&amp;type=text%2Fhtml&amp;schema=dwcdn" width="600" height="169" frameborder="0" scrolling="no"><a href="https://medium.com/media/788aaad993ca5565eac8f09b470f57c1/href">https://medium.com/media/788aaad993ca5565eac8f09b470f57c1/href</a></iframe><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2F79VMG%2F3%2F&amp;display_name=Datawrapper&amp;url=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2F79VMG%2F3%2F&amp;image=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2F79VMG%2Fplain-s.png%3Fv%3D3&amp;type=text%2Fhtml&amp;schema=dwcdn" width="600" height="169" frameborder="0" scrolling="no"><a href="https://medium.com/media/c6b69cd2026a136cd7850a9740283444/href">https://medium.com/media/c6b69cd2026a136cd7850a9740283444/href</a></iframe><p>This example of steal rates in the Austrian Softball Bundesliga illustrates why the situation matters. As a runner on 1st, you are expected to steal between 30 and 60% of the time in a 1st-and-3rd situation, depending on the outs, while as the only runner, you would only be expected to steal about 15 to 18%.</p><p>With the matrices for the league average, you then simply look at each baserunning situation for a runner, multiply the steal rate by the quantity to get an expected number of attempts. The sum of all expected attempts is then deducted from a player’s actual stolen base attempts to make up the final Attempts Above Average.</p><p>To see how this works in practice, let’s look at a hypothetical runner, <strong>Player A</strong>, and see how their decisions stack up against the league baseline.</p><p>Over the course of the season, imagine Player A finds themselves on first base with no outs and no other runners exactly ten times. According to our matrix, the league average steal rate for this specific scenario is 17.48%. Now, let’s say they also find themselves on first base in a 1st-and-3rd situation with no outs on five different occasions, which is a highly aggressive trigger situation across the league with a 58.97% steal rate.</p><p>We can calculate exactly how many times an average runner would attempt to steal given those exact situations:</p><blockquote>Expected Attempts = (10 opportunities × 17.48%) + (5 opportunities × 58.97%)<br>Expected Attempts = 1.75 + 2.95 = 4.7</blockquote><p>Based purely on the league environment, the math expects them to take off roughly 4.7 times.</p><p>However, Player A is quite aggressive. They actually took off five times in the first scenario and four times in the second, resulting in nine actual steal attempts. To find their Attempts Above Average, we simply subtract the expected baseline from what they actually did:</p><blockquote>AAA = 9 Actual Attempts — 4.7 Expected Attempts = +4.3 AAA</blockquote><p>Based on this metric alone, knowing they had 9 attempts and that this is almost twice as many as expected, we can unequivocally say this runner is very aggressive and likely to steal… But <em>how</em> likely?</p><h4>AAA+ — Attempts Above Average+</h4><p>During a game, a coach might ask exactly that: “How likely is the runner to steal?” and since AAA is a cumulative statistic, it’s counting. A player who’s on base more will have more opportunities to steal, which would also influence their AAA. A much more aggressive runner may have fewer Attempts Above Average than someone who simply gets on base more and is conventionally aggressive.</p><p>For this purpose, similar to stats like wRC+ or OPS+, we simply divide the actual attempts by the expected attempts and multiply the result by 100 to come up with <strong>Attempts Above Average+ (AAA+)</strong>. That way 100 is the league’s average, while 200 would indicate the runner steals twice as aggressively as the league average runner.</p><p>In our scenario, the calculation looks like that:</p><blockquote>AAA+ = 100×(9 Actual Attempts / 4.7 Expected Attempts) = 191 AAA+</blockquote><p>Player A steals about 91% more than the average runner.</p><h4>RAI — Relative Aggressiveness Index</h4><p>A third way to quantify someone’s aggressiveness is to answer the question “in how many of the situations did they try to steal”, for which you could simply use their steal rates, but that’s not quite telling a story. A player can have a low steal rate simply because they are often blocked from stealing, for which we created the matrix to begin with. So, to calculate their steal rate, I came up with the <strong>Relative Aggressiveness Index (RAI)</strong>, which considers that matrix. But rather than the average steal rate, we compare the player’s steal rate to the highest individual steal rate of the league in each situation (with at least a certain number of situations).</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FpOWzY%2F2%2F&amp;display_name=Datawrapper&amp;url=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FpOWzY%2F2%2F&amp;image=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FpOWzY%2Fplain-s.png%3Fv%3D2&amp;type=text%2Fhtml&amp;schema=dwcdn" width="600" height="187" frameborder="0" scrolling="no"><a href="https://medium.com/media/46f611cbf24a79a2e04a6058bbe26368/href">https://medium.com/media/46f611cbf24a79a2e04a6058bbe26368/href</a></iframe><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FoEMQh%2F3%2F&amp;display_name=Datawrapper&amp;url=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FoEMQh%2F3%2F&amp;image=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FoEMQh%2Fplain-s.png%3Fv%3D3&amp;type=text%2Fhtml&amp;schema=dwcdn" width="600" height="150" frameborder="0" scrolling="no"><a href="https://medium.com/media/206c32a67e13f3894b4523410b1f571c/href">https://medium.com/media/206c32a67e13f3894b4523410b1f571c/href</a></iframe><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FnkTaS%2F2%2F&amp;display_name=Datawrapper&amp;url=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FnkTaS%2F2%2F&amp;image=https%3A%2F%2Ffd.xuwubk.eu.org%3A443%2Fhttps%2Fdatawrapper.dwcdn.net%2FnkTaS%2Fplain-s.png%3Fv%3D2&amp;type=text%2Fhtml&amp;schema=dwcdn" width="600" height="150" frameborder="0" scrolling="no"><a href="https://medium.com/media/e6f66dd6140666beb0e18897d875a5cf/href">https://medium.com/media/e6f66dd6140666beb0e18897d875a5cf/href</a></iframe><p>Interestingly, here, using the maximum, the highest likelihood of a runner stealing 2nd base with no one else on is considerably higher than any other situation, even 1st-and-3rd. This makes sense because an R1-only situation is much more dependent on a runner’s ability to steal, which means slow runners are dragging the baseline down much more than they do for 1st-and-3rd situations.</p><p>Now re-introducing our hypothetical Player A, with five attempts as R1-only in ten opportunities and four attempts in five 1st-and-3rd situations, both with no outs. The RAI baseline for these situations is 85.71% and 58.97%, respectively. To calculate RAI one needs to find the denominator for each situation, the maximum expected attempts, and sum them for one final denominator, which represents the maximum number of attempts one can expect. For RAI, you then simply divide the actual attempts by that denominator.</p><blockquote>RAI Denominator (maximum expected Attempts):</blockquote><blockquote>R1-Only Denominator: 10 opportunities*85.71% R1-Only Baseline = 8.57 <br>1st-and-3rd Denominator: 5 opportunities*58.97% 1st-and-3rd Baseline = 2.95</blockquote><blockquote>Total RAI Denominator = 1st-and-3rd Denominator+R1-Only Denominator Total RAI Denominator = 8.57+2.95 = 11.52</blockquote><blockquote>RAI = Attempts / Total RAI Denominator:<br>RAI = 9/11.52 = 78.1%</blockquote><p>Player A has a <strong>78.1% RAI</strong>. This confirms our earlier observation that this player is highly aggressive and steals in almost 80% of the reasonable situations.</p><p>Quantifying aggressiveness requires separating <em>intent</em> from <em>opportunity</em>. AAA does this by showing the exact number of extra steals a player attempts compared to an average runner in the exact same situations. RAI takes it a step further, scoring that intent against the league’s absolute maximums. A player might have a low AAA simply because they hit at the bottom of the order and rarely reach base, but a high RAI reveals that the moment they do get on, they are an immediate threat. Combined, they turn baserunning tendencies from a guessing game into a measurable scouting tool.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4329aac12625" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Framing and Its Value in Softball and Baseball]]></title>
            <link>https://medium.com/@vschuh/framing-and-its-value-in-softball-and-baseball-b430b1895d53?source=rss-349b8b39602e------2</link>
            <guid isPermaLink="false">https://medium.com/p/b430b1895d53</guid>
            <category><![CDATA[baseball-analytics]]></category>
            <category><![CDATA[sabermetrics]]></category>
            <category><![CDATA[baseball-statistics]]></category>
            <category><![CDATA[baseball]]></category>
            <category><![CDATA[softball]]></category>
            <dc:creator><![CDATA[Valentin Schuh]]></dc:creator>
            <pubDate>Sun, 26 Oct 2025 19:53:54 GMT</pubDate>
            <atom:updated>2025-10-26T20:21:42.064Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5zcA3aB21cISTPtK3wJYOw.jpeg" /></figure><p>In softball and baseball, <strong>framing</strong> is the subtle art of stealing strikes — making a pitch just off the plate look like it caught the corner. It’s a catcher’s sleight of hand, deceiving the umpire into a favorable call. While its value is undeniable, it’s often understated precisely because it’s <em>nearly</em> intangible. Especially in leagues and tournaments where the exact location of pitches is not widely recorded (which is essentially every league that is not the MiLB or MLB), it is <em>seemingly</em> impossible to even tell who is good at framing and who is not, let alone determine the value of it.</p><p>In this article, we will approximate a framing statistic, quantify its value, and examine some examples in European softball and baseball.</p><h3>How to calculate it:</h3><p>The biggest problem with leagues in Europe is that they do not record the exact location of pitches, which at first makes it seem impossible to calculate a statistic that tells how good a catcher can <em>disguise a pitch’s location</em>. However, with a sufficient sample size, you can look at a pitcher’s tendency and, based on that, make an estimate of how many strikes they will throw. And that is exactly how you can calculate it.</p><p>Pitchers have different tendencies — some are prolific strike-throwers, some tend to be more wild. These are tendencies that marginally change over the course of a short span of years. A pitcher having a 30 percent called strike rate will not start throwing 40 percent the next year, and vice versa. In fact, once a pitcher has found their approach, their called strike rate will not move by more than 2% over a single year.</p><p>Looking at a pitcher’s career, their last 500 Plate Appearances or the last 2000 Pitches, you can make a good estimate of how many strikes they should throw under normal circumstances. (When defining the sample size, keep in mind you need enough sample size to make an approximate picture of a pitcher’s tendency but not too much as to where a pitcher’s tendency from 4 years ago still impacts the data) To make the tendency even more telling, we only look at a pitcher’s tendency outside of the battery we try to analyze, so their strike percentage to everyone except for the specific catcher.</p><p>Based on that data, you really just need to count the number of pitches thrown by that pitcher to that catcher and multiply it by the strike rate we determined before, and you have an <strong>expected number of strikes</strong>. Now, you take the <strong>actual strikes</strong> thrown to the catcher and subtract the expected value, and you will have an amount of strikes called above or below expectations, which we can attribute to framing. Mathematically put, the formula looks something like this:</p><h4>Formula for <strong>Framing Strikes Above Average (FrSAA):</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/563/1*LE4m9GEZnWzFcfR_jMcxxQ.png" /><figcaption>Formula for Framing Strikes Above Average (FrSAA)</figcaption></figure><p>Where <em>c_ps </em>and <em>c_pb </em>stand for the pitched strikes and balls to the catcher, <em>t_ps </em>and <em>t_pb </em>stand for the pitcher’s total pitched strikes and total pitched balls, respectively, to all catchers, and <em>i</em> represents the pitcher.</p><p>Now, rather than just looking at a single pitcher, you take the sum of each pitcher the catcher has caught in a tournament (to get the framing in a tournament).</p><p>To make this look like a statistic where 100 is the average and 110 means the catcher frames an additional 10 strikes on the expected 100 strikes, you have to divide the catcher’s total actual strikes by their total expected strikes, and then multiply that ratio by 100. The adjusted formula looks like this:</p><h4>Formula for Framing+ (FRM+):</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/504/1*bQxfEDGUwwyPVbuWPDhwJA.png" /><figcaption>Formula for Framing+ (FRM+)</figcaption></figure><h3>The Value of Framing</h3><p>Okay, so a catcher can steal a few extra strikes. <em>So what?</em> To understand the real-world impact, we have to attach a run value to their work. Stealing 10 strikes in a tournament <em>feels</em> important, but how many runs does that actually save? To answer that, we first have to answer a more basic question: what is one strike <em>worth</em>?</p><h4>The Value of Strikes and Balls</h4><p>To find this value, we calculate the average marginal run difference for a strike and a ball. This is done by first finding the average run expectancy for every count. We use <strong>wOBA (weighted On-Base Average)</strong> for this, which measures the average runs created per plate appearance. By finding the average wOBA for all PAs that <em>pass through</em> a specific count, we can find the marginal value of any pitch.</p><p>For example, if the average wOBA for a PA at 0–0 is 0.300 and the average wOBA for a PA at 0–1 is 0.250, the marginal value of that first strike is <strong>-0.050 runs</strong>. In other words, the expected value of this plate appearance dropped by 0.05 runs with that strike. The batter is expected to produce 0.05 fewer runs in this plate appearance.</p><p>A final, single strike value and ball value are then created by calculating the weighted average of all these different marginal pitch values based on how frequently each count occurs. Taking all differences from the counts and calculating the average difference a strike or a ball produces.</p><h4>The Value of a Stolen Strike</h4><p>Deceiving an umpire into calling a ball a strike is worth the absolute difference in each of those situations. Expanding on the formula for the framing strikes above average, the new formula would look like this:</p><h4>Formula for Framing Run Value (FrRV):</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/770/1*mv_OMfBKKCdgfLwvcy-lsg.png" /><figcaption>Formula for Framing Run Value (FrRV)</figcaption></figure><p>So, instead of, for example, 0.05 runs a ball would give the batter you created -0.05 runs (both real-world values) for the batter, which would mean an absolute difference of 0.10 runs created by one stolen strike. If you stole 10 strikes, you produced 10 times this difference, equaling a whole run already. This value quickly becomes quite substantial.</p><h3>The Best Framers in Europe</h3><h4>Softball</h4><p>Let’s put the formula to work. One of the clearest examples of an elite framer in European softball is <strong>Katerina Vackova</strong>, the backstop for the Czech national team and Tempo Praha. The data shows she’s not just a good receiver; she’s a run-saver. Over 2,913 recorded pitches, the model expected her pitchers to get 769.8 called strikes. Instead, Vackova coaxed 885 out of the umpires — a staggering 115 extra strikes. This gives her a career <strong>FRM+ </strong>of 115 (that is, 115 strikes on 100 expected strikes), and a <strong>FrRV</strong> of <strong>12.3 runs</strong>. Of those 12 runs she stole, nearly all came from her last two Extraliga Seasons (6.5 in 2024 and 5.5 in 2025) and the 2025 European Championship (1.6).</p><p>We can break down these three tournaments by looking at the five pitchers she caught most frequently:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JvMwN9QAW9Yxbh-Yz5ovDA.png" /></figure><p>The data clearly shows a very consistent trend: of the five pitchers she caught, all five of them had elevated numbers of strikes called. The pitcher she caught most frequently was <strong>Anna Štenglová,</strong> of whom she received 1324 pitches. Of those pitches, roughly 321 were expected to be called strikes; however, she received <strong>almost 100 additional strike calls,</strong> which accounts for more than 10 runs stolen. The others, while caught less, also show this trend of receiving additional called strikes, strengthening her case of being a great framer.</p><h4>Baseball</h4><p>Over the last three seasons in the Baseball Extraliga, another Czech catcher stood out as an impeccable framer. Trebic Nuclears and Tempo Praha’s backstop <strong>Martin Zelenka</strong> has caught a total of 2071 pitches with an expected 648 called strikes to his actual 725, which amounts to <strong>77 extra strikes</strong> and almost 8 runs stolen, and a career <strong>FRM+ of 112</strong>. Most of this can be credited to his 2023 Extraliga Season, in which he <strong>stole 5.5 runs</strong>. An additional <strong>two runs</strong> came in the two years that followed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lo2ggiCMUNA0rtyS6obk8g.png" /></figure><p>In his pitcher breakdown, looking at the nine pitchers he most frequently caught, we can also clearly see a consistent trend of always receiving more strike calls than expected. <strong>More than 42 of the stolen strikes</strong> came from the top 4 pitchers in this list. But regardless of who he caught, it always resulted in more strikes than expected, further proving his framing skill.</p><h3>How to Use and What to Look For in These Stats</h3><p>These stats, given the lack of actual location of the pitches, work completely on the theory of normal distribution through sample size. This means a sufficient sample is necessary, which refers to a pitcher’s tendencies, the catcher’s total pitches called (ball or strike), but also ideally several tournaments of a catcher. A catcher posting above-average numbers for several tournaments, several pitchers, having seen sufficient pitches, is <strong>no coincidence</strong>; however, if one of these parameters is unchecked, it may indeed be a coincidence.</p><h4>Lurking Variables</h4><p><strong>Pitcher yips</strong></p><p>Over the course of a season, a pitcher is going to have around as many performances above their average as below, going by the rules of normal distribution again. Though a tournament may only last a single week, and especially in softball, these tournaments may still have 10 or more games, creating more than a sufficient sample size in terms of pitches recorded, albeit an insufficient sample size in terms of gamedays. A pitcher may simply have a bad week in which they throw more wildly than their career tendency indicates, which subsequently hurts the catcher’s framing statistic. The way to counter this is to validate a catcher’s framing by examining the pitcher’s numbers for that tournament — if they post normal numbers in terms of walk rate, (called) strike rate, or whiff rate, you can also expect their catcher’s framing statistics to be valid.</p><p><strong>Timing</strong></p><p>Another common issue would be a situation in which a pitcher doesn’t pitch for a long time in the database. For example, Pitcher A pitches a tournament in 2021, then goes off record, pitching in a league that does not have pitch-by-pitch data, and then comes back in 2025 with a completely refined approach and tendencies. In those four years, the pitcher may have improved their command significantly, throwing more strikes. This situation, which isn’t too uncommon, would result in elevated framing statistics for the catcher as the career tendency of the pitcher is incorrectly low, based on their numbers from years ago, which aren’t accurate anymore. These numbers may be discarded as a whole, as there is no proper way to determine the impact a catcher has on the improvement of this pitcher.</p><h3>Conclusion</h3><p>Ultimately, in a baseball and softball world increasingly dominated by high-speed cameras and pitch-tracking data, it’s easy to assume that leagues without that technology are left guessing. But as this model shows, that isn’t the case. By intelligently using the data we <em>do</em> have — namely, pitcher tendencies over a large sample — we can build a reliable tool to identify and quantify a catcher’s defensive value. While stats like FRM+ and FrRV must be used with care, accounting for the variables that can skew the data, they provide a powerful new lens for evaluation. For teams, it means a way to find hidden value; for catchers, it’s a way to finally get credit for the runs they save, <strong>one stolen strike at a time</strong>, which at the end of the season can amount to an entire win stolen.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b430b1895d53" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[To Swing or not to Swing —The Value of Taking or Swinging at Pitches.]]></title>
            <link>https://medium.com/@vschuh/to-swing-or-not-to-swing-the-value-of-taking-or-swinging-at-pitches-9f5598afb974?source=rss-349b8b39602e------2</link>
            <guid isPermaLink="false">https://medium.com/p/9f5598afb974</guid>
            <category><![CDATA[sport-analytics]]></category>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[softball]]></category>
            <category><![CDATA[sabermetrics]]></category>
            <category><![CDATA[baseball]]></category>
            <dc:creator><![CDATA[Valentin Schuh]]></dc:creator>
            <pubDate>Mon, 21 Oct 2024 09:31:38 GMT</pubDate>
            <atom:updated>2025-10-26T19:35:05.150Z</atom:updated>
            <content:encoded><![CDATA[<h3>To Swing or not to Swing — The Value of Taking or Swinging at Pitches.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LLS1MgMxqYwiJ2pnTOnXYg.jpeg" /></figure><h3>Introduction</h3><blockquote><em>Should a player always take a 3–0 pitch regardless of how its thrown? Should the batter have their mind made up before the pitcher even comes set?</em></blockquote><p>That is a debate colleagues and I recently found ourselves in. My argument was based on linear weights. The linear weights are calculated with the run-expectancy changes and give each result a value representing the increase of expected runs between the moment of it happening and the end of the inning. A walk is valued at only around <em>0.55</em> runs, the least among all offensive results. On a 50% walkrate, that would result in merely <em>0.28 </em>runs, significantly lower than any hit would have (which ranges from roughly 1 to 2 runs). Assuming the batting average on 3–0 swings is around <em>0.300</em>, the run production on swings would at least challenge the run value of <em>0.28 </em>on takes. We did not exactly come to a conclusion, as we were lacking the data, but it did motivate me to create a statistic to display take and swing values of players in 3–0 counts.</p><h3>3–0 Values</h3><p>In order to calculate the 3–0 values, I’d only require a quick SQL prompt on my database to give me the results of swings or takes on 3–0 counts (hits, walks, strikes, outs). A 3–0 take for a ball is worth the value of a walk, and a swing is worth the linear weights from singles, doubles, triples, and home runs. (which are different from league to league.) I took the total values divided by the total situations, and that gave me the average run production on swings and takes, and that was virtually it.</p><p>So quickly done, I figured I might just change the prompt and get the results for each other count as well. And immediately I realized the drastic miscalculation I had done. The take-value on each count that did not have three balls was zero or less, because I only accounted for walks and hits and disregarded the value of balls and strikes completely; I valued them at zero.</p><h3>The Value of Balls and Strikes</h3><p>After some days of brainstorming, I came up with a way to get the value of a ball or a strike thrown. I looked at the linear weights and their calculation and how they take the average change of run-expectancy for all their results. In a plate appearance, your fight is against the out, so rather than the run-expectancy, we take the out expectancy for each count and look at the average change of expectancy on strikes and balls. A quick demonstration by the example of an out-expectancy table:</p><pre>Out Expectancy Table<br>┌──────────┬────────┬────────────┬────────┬──────────┬────────┬─────────┐<br>│ Count    │ Out%   │ Count      │ Out%   │ Count    │ Out%   │<strong> Avg K </strong>  │<br>├──────────┼────────┼────────────┼────────┼──────────┼────────┼─────────┤<br>│ 0-0      │ 0.122  │ 0-1        │ 0.103  │ 0-2      │ 0.473  │ <strong>0.176 </strong>  │<br>│ 1-0      │ 0.102  │ 1-1        │ 0.115  │ 1-2      │ 0.361  │<strong> 0.130  </strong> │<br>│ 2-0      │ 0.065  │ 2-1        │ 0.114  │ 2-2      │ 0.354  │<strong> 0.145 </strong>  │<br>│ 3-0      │ 0.022  │ 3-1        │ 0.096  │ 3-2      │ 0.348  │<strong> 0.163  </strong> │<br>│ <strong>Avg Ball</strong> │ <strong>-0.025</strong> │ <strong>Avg Ball</strong>   │ <strong>-0.002 │ Avg Ball │ -0.031</strong> │<strong> Total K </strong>│<br>│          │        │ <strong><em>Total Ball</em></strong> │ <strong><em>-0.019</em></strong> │          │        │<strong><em> 0.153  </em></strong> │<br>│          │        │            │        │          │        │         │<br>└──────────┴────────┴────────────┴────────┴──────────┴────────┴─────────┘<br><br><br><br><br><br></pre><p>In the table, we have the out-expectancy (as “Out%”) listed for each count. This represents how many pitches resulted in an out in each count. 2-Strike counts have significantly higher out-expectancy, given the strikeout chance; similarly, 3-Ball counts have lower values with the walk-opportunity in place in those counts. All the way on the bottom and on the right side, there is the average <strong>change </strong>for respective columns and rows. The average value of those changes represents the strike/ball value. But beware: that value is the out-expectancy change rather than the run-expectancy change. So a strike, for example, increases the out-expectancy by <em>0.153</em> (so about 15%). For our calculation, we require the run-expectancy change, however. That is calculated simply by multiplying that value with the outvalue of the linear weights (wobaOut), which will be between <em>-0.18</em> and <em>-0.28</em> (in the mlb it is roughly <em>-0.26</em>; in Europe it is usually lower). In our example, we use the value <em>-0.19</em> (as this is the average outvalue in my database), which sets up the following formulas and results:</p><blockquote>0.19 ⋅ 0.153 = -<strong>0.02907 </strong>strike-value</blockquote><blockquote>-0.19 ⋅ -0.019 = <strong>0.00361 </strong>ball-value</blockquote><p>And just like that, balls and strikes have values. As already mentioned, each tournament/league has different linear weights, so it makes sense to calculate those values individually for each league and add them to the linear weights.</p><h3>Calculating the Swing-Take-Matrix (STM)</h3><p>Now, with the final pieces added, the value of swings and takes for each count can be calculated. The formula is quite simple but long, as you have to account for each result multiplied with its linear weight and that for each count. Finally, a formula for a count could look something like this:</p><blockquote>(Balls ⋅ Ball-Value + BB⋅ wobaBB + Strikes Looking ⋅ Strike-Value + Strikeout Looking ⋅ wobaOut) / (Balls+Strikes looking + Strikeouts Looking) = Take Value</blockquote><blockquote>(Strikeout swinging + In Play Out) ⋅ wobaOut + ((Strike swinging + Foul) ⋅ Strike-Value + 1b ⋅ woba1b + 2b ⋅ woba2b + 3b ⋅ woba3b + HR ⋅ wobaHR)/(Strike swinging + Strikeout swinging + Foul + In Play Out + hits) = Swing Value</blockquote><p>Walks and strikeouts both get valued not as balls or strikes but only as walks and outs to avoid counting them twice. The wobaXX variables represent the linear weights. Again, with each league having their own linear weights, every player has a different value in each league, which makes sense but isn’t practical. So ideally you would take the average from a player over all their competitions.</p><h3>How to use the STM</h3><p>Generally, the use of the STM should be straight forward; a higher take value indicates a situation in which you should take, and a higher swing value indicates a good swing opportunity. Still, there are two different methods to use this matrix to your advantage:</p><h4>A Batters Example:</h4><p>One of those methods is for batters. Use it on your batters to see their struggles or strengths. Or you can take a look at opposing players matrices to tell your pitcher how and when to attack the batter.</p><pre>Jose Leonardo Ferrini Colmenarez:<br><br>Take Value<br>┌───────────────┬─────────┬─────────┬─────────┐<br>│ Balls/Strikes │ 0       │ 1       │ 2       │<br>├───────────────┼─────────┼─────────┼─────────┤<br>│ 0             │ -0.0084 │ -0.0037 │ -0.0148 │<br>│ 1             │ -0.0061 │ -0.0050 │ -0.0193 │<br>│ 2             │ -0.0094 │ -0.0052 │ -0.0191 │<br>│ 3             │ 0.1646  │ 0.3482  │ 0.4065  │<br>└───────────────┴─────────┴─────────┴─────────┘<br>Swing Value<br>┌───────────────┬────────┬────────┬────────┐<br>│ Balls/Strikes │ 0      │ 1      │ 2      │<br>├───────────────┼────────┼────────┼────────┤<br>│ 0             │ 0.0795 │ 0.0651 │ 0.0495 │<br>│ 1             │ 0.1235 │ 0.0989 │ 0.1067 │<br>│ 2             │ 0.1522 │ 0.0871 │ 0.0935 │<br>│ 3             │ 0      │ 0.1790 │ 0.1184 │<br>└───────────────┴────────┴────────┴────────┘<br>Swings and Takes<br>┌──────────────┬──────────┬─────────┬─────────┐<br>│ Balls/Strikes│ 0        │ 1       │ 2       │<br>├──────────────┼──────────┼─────────┼─────────┤<br>│ 0            │ 330|1030 │ 249|352 │ 65|150  │<br>│ 1            │ 205|368  │ 221|274 │ 113|208 │<br>│ 2            │ 73|169   │ 159|162 │ 126|168 │<br>│ 3            │ 0|87     │ 65|105  │ 98|107  │<br>└──────────────┴──────────┴─────────┴─────────┘</pre><p>Those are the values of Jose Leonardo Ferrini Colmenarez, who plays for San Marino baseball and who has represented Italy in the 2021 European Championship. I chose him somewhat randomly based on the fact that he has quite a nice sample size of more than 1300 plate appearances. And quite immediately I noticed that in those 1300 PAs he has seen 87 counts of 3–0 and never swung even once, resulting in a swing value of 0 for that count. Overall, his swing rate of 35% is about 5% higher than the Serie A average in 2024; his swing value is significantly better than the league average, but his take value is by average <em>0.185 </em>worse than average. Most polarizing are his 3-ball decisions. His swing values are well above average, but his take values are well subpar. His biggest loss is on the takes at 3–0, which indicates that he is too conservative on those counts, not surprising with no swings in 87 chances. His biggest advantage is on 3–1 swings, where he is still swinging 9% less than average but generates <em>0.1226</em> more value on his swings than his league. Also, his take value is average in that count. You can make the assumption that this is his best count.</p><h4>A Pitchers Example:</h4><p>What I find more usable is the pitcher’s perspective, though. How does a pitcher attack and perform in a given situation? This can be a huge advantage as an opposing coach, knowing when to give your players a take-sign or not. Again, let us look at an example. This time, let’s do softball:</p><pre>Lisa Hop:<br><br>Take Value<br>┌───────────────┬─────────┬─────────┬─────────┐<br>│ Balls/Strikes │ 0       │ 1       │ 2       │<br>├───────────────┼─────────┼─────────┼─────────┤<br>│ 0             │ -0.0083 │ -0.0049 │ -0.0273 │<br>│ 1             │ -0.0088 │ -0.0054 │ -0.0282 │<br>│ 2             │ -0.0093 │ -0.0065 │ -0.0230 │<br>│ 3             │ 0.2531  │ 0.3934  │ 0.3948  │<br>└───────────────┴─────────┴─────────┴─────────┘<br>Swing Value<br>┌───────────────┬─────────┬────────┬─────────┐<br>│ Balls/Strikes │ 0       │ 1      │ 2       │<br>├───────────────┼─────────┼────────┼─────────┤<br>│ 0             │ 0.0215  │ 0.0180 │ -0.0615 │<br>│ 1             │ 0.0456  │ 0.0343 │ -0.0144 │<br>│ 2             │ 0.0158  │ 0.0487 │ -0.0150 │<br>│ 3             │ -0.0520 │ 0.0758 │ 0.0011  │<br>└───────────────┴─────────┴────────┴─────────┘<br>Swings and Takes<br>┌───────────────┬──────────┬─────────┬─────────┐<br>│ Balls/Strikes │ 0        │ 1       │ 2       │<br>├───────────────┼──────────┼─────────┼─────────┤<br>│ 0             │ 728|1533 │ 493|604 │ 203|297 │<br>│ 1             │ 352|545  │ 470|437 │ 300|367 │<br>│ 2             │ 110|194  │ 233|242 │ 264|263 │<br>│ 3             │ 4|98     │ 106|116 │ 174|151 │<br>└───────────────┴──────────┴─────────┴─────────┘</pre><p>These stats belong to Dutch pitcher Lisa Hop, this time not chosen too randomly. Hop is one of the better pitchers in Europe, and those values we just calculated will show that, but first, to understand how good she is, let’s look at some rudimentary stats of hers. Over 4 years in the Dutch league, she posted a 1.57 ERA in almost 500 innings, a batting average against of .176, and 661 strikeouts on 239 walks — that is quite alright to say the least. Now let’s look at the STM in comparison. Her average take value is <em>-0.0167</em> “better” than average, which means batters create <em>0.0167</em> runs per pitch less against her. Batters also generate an astonishing <em>0.092</em> runs less on swings. That is a lot. Looking at those stats as a coach, the best advice against her would be to not face her. There is not a single count in which batters are performing at least average against her. In take-values, they perform marginally below average until it gets to 3-Ball takes, in which they then produce <em>0.0554</em> runs less than against league average. In swinging, there is only a single count in which batters come even close to average; that is the 3–1 count. Swinging in that count generates about <em>0.0758</em> runs, only <em>0.0197</em> runs below average. On 2–2 counts, batters generate negative <em>0.0150</em> runs swinging; the average for that count is a positive value of <em>0.1654</em>, a whopping difference of <em>0.1804</em>. She owns that count, and you should certainly not swing.</p><p>In conclusion, <em>Lisa Hop is a solid pitcher.</em></p><h3>Comparing</h3><p>What you might have noticed is that in my examples, I always compared the statistics to the league average. That is something I would recommend when using abstract statistics, which they are. A common way to go about that is using the difference of player value and league value plus 1 and that times 100 ((value-lgValue + 1) * 100). That gives you a value with the base of 100, which is used for wRC+, for example. 100 would represent the league average; everything above is better and everything below is worse. In this scenario, a 1000-base might even be appropriate.</p><h3>Summary</h3><p>I would argue this is the most innovative statistic for live coaching that I have yet produced in my database. The knowledge of when it’s good to swing or not is very valuable. Being able to confidently give the take-sign can be a huge advantage. Additionally, quatinfying strikes and balls is a nice extra that might come in handy for different statistics.</p><h4>So Swing or Take?</h4><p>Now I have answered all questions but the initial ones, so before I conclude this article, let me do that: The statistics vary strongly from player to player and also tournament to tournament; however, for larger sample sizes and higher-tiered leagues (such as the Serie A or the Hoofdklasse in Italy and the Netherlands), it is safe to say that you <strong>should not </strong>swing on most 3–0 pitches. Among players with at least 300 PA in their career, the average take-value is <em>0.262</em> runs better than the swing value. That difference increases with one strike and even more with two strikes up to <em>0.361</em>. 3-Ball counts are the only counts in which taking is more valuable than swinging, which makes sense, given those are the only counts in which you can walk and produce more than just the ball value on a take.</p><h3>Afterword</h3><p>Thanks for reading. I hope you enjoyed this read, found it interesting, entertaining, or even funny (?). If there was something unclear or if I did not consider something, let me know.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9f5598afb974" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Saving Runs with your arm and it’s reputation. A different way to evaluate Catchers]]></title>
            <link>https://medium.com/@vschuh/saving-runs-with-your-arm-and-its-reputation-a-different-way-to-evaluate-catchers-8b271b3a4978?source=rss-349b8b39602e------2</link>
            <guid isPermaLink="false">https://medium.com/p/8b271b3a4978</guid>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[baseball]]></category>
            <category><![CDATA[sports-analytics]]></category>
            <category><![CDATA[softball]]></category>
            <category><![CDATA[statistics]]></category>
            <dc:creator><![CDATA[Valentin Schuh]]></dc:creator>
            <pubDate>Tue, 08 Oct 2024 19:55:46 GMT</pubDate>
            <atom:updated>2024-11-08T13:29:46.335Z</atom:updated>
            <content:encoded><![CDATA[<h3>Saving Runs with your arm and it’s reputation. A different way to evaluate catchers in Baseball and Softball.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DgFkKhDHSIavvD1VSgaOkg.jpeg" /><figcaption>Elias Kreska at the U23 European Baseball Championship 2023 — Photo credit: Myself</figcaption></figure><h3>Foreword</h3><p>The strive for the statistical edge by sports analysts will never be satisfied, especially in the sports of baseball or softball, where statisticians and mathematicians regularly come up with new ways to measure the value of players or teams. May it be a statistic representing overall value or one focusing on a certain aspect of a player, such as hitting, fielding, or running. The amount of different ways to quantify one’s performance is astonishing.</p><p>While I might not qualify as a professional statistician, I at least fall under the term “<em>statistic-geek” </em>and regularly think of new methods or implement already existing methods to improve my database of global Baseball and Softball statistics. But to find an aspect that is unexplored is incredibly unlikely, and it would be foolish to claim I am the first one to come up with the two <em>new </em>methods that I will introduce to you. I will however elaborate on them with the formulas I used and my thoughts behind them.</p><p>First, however, what is the aspect I am talking about even? The answer to that question is <em>catchers, </em>more precisely their defensive value in preventing baserunning. A catcher is probably the most valuable defender (if we count pitching separately) on the diamond. Blocking, picking off, and throwing out runners is why catchers have more defensive impact than any other position. The latter one is what both statistics will focus on. One is the immediate value of catching runners, and the other is the value a catcher prevents with their reputation or the respect of the opponents.</p><h3>Weighted Caught Stealings</h3><p>The name of the first statistic is <strong>weighted Caught Stealings,</strong> or short <strong>wCS. </strong>And there might already be some who know exactly how this statistic works simply by the term <em>weighted</em>, which is used in several other statistics, such as the counterpart <em>wSB</em><strong> </strong>(weighted Stolen Bases), which works exactly like <strong>wCS</strong> — just from the other perspective.</p><p>Catching a runner is valuable, but how valuable and how often do you need to catch a runner to be considered good statistically? One can use a stat that looks at a success rate of caught stealings (or <em>SB%</em> <em>against</em>), that would at least tackle the second question but also only in consideration of the sample size of the catcher.</p><h4>Calculating wCS</h4><p><strong>wCS</strong> needs three constants: the league base stealing success rate, which we call <em>lgSB%</em> moving on; the other two are the value of a stolen base and the one of a caught stealing; we call them <em>runSB</em> and <em>runCS</em>. <em>runSB</em> is calculated looking at the average change of the run expectancy on a stolen base based on the <a href="https://library.fangraphs.com/principles/linear-weights/">run expectancy matrix,</a> while <em>runCS</em> is calculated with the following formula:</p><blockquote>runCS = -(2 · RunsPerOut + 0.075)</blockquote><p>both according to <a href="https://library.fangraphs.com/offense/wsb/">fangraphs</a>.</p><p>With those three constants, the actual calculation can begin. You take a player’s total attempts against, and with the <em>lgSB%</em>, you calculate the expected caught stealings. The difference between actual caught stealings and expected caught stealings is multiplied by the sum of <em>runSB</em> and the absolute value of <em>runCS</em>. I chose to use the difference of those two as it represents the two different outcomes — a catcher turns a stolen base into a caught stealing.</p><p>So let’s look at the 2024 MLB — the <em>lgSB%</em> is roughly 79%, the <em>runSB</em> is 0.200 and the <em>runCS</em> is -0.405. Will Smith had 96 attempts against him, which gives him 20.16 expected CS based on the <em>lgSB%</em> (0.21 · 96), he had 32, which results in the following final formula:</p><blockquote><em>(32 – 20.16) · 0.605 = 7.1632</em> runs caught from Will Smith in 2024.</blockquote><p>To figure out that Will Smith is a great catcher in terms of catching runners, you would not have needed all this complicated math. The simple <em>SB% against</em> him is well sufficient to make that assumption. The strength of <strong>wCS</strong>, however, is that for one, it gives you a real number — 7 runs (and some change). And that number is especially valuable in interleague comparisons; I don’t mean AL/NL but rather comparing leagues outside of the MLB where basestealing tendencies and success rates can be vastly different, having a clear impact on the <em>runSB</em> and <em>runCS</em> values. <strong>wCS</strong> takes this into consideration and gives you an (almost) tendency-neutral value. In the following table, the difference is illustrated by the example of a few of Europe’s best baseball and softball leagues in the year 2024:</p><pre>Baseball<br><br>      Tournament        Nat   games   runSB   runCS    lgSB%    SB | CS   <br> --------------------- ----- ------- ------- -------- ------- ----------- <br>  Serie A Baseball      ITA     496   0.207   -0.486   77.9%   972 | 276  <br>  Hoofdklasse Honkbal   NED     222   0.225   -0.547   81.2%   608 | 141  <br>  Extraliga             CZE     155   0.240   -0.444   84.8%   498 | 89   <br>  Croatian Baseball     CRO     116   0.231   -0.689   90.7%   611 | 63   <br>  Baseball D1           BEL     112   0.250   -0.662   88.0%   322 | 44    <br><br>Softball<br><br>      Tournament        Nat   games   runSB   runCS    lgSB%    SB | CS   <br> --------------------- ----- ------- ------- -------- ------- ---------- <br>  Serie A1 Softball     ITA     151   0.192   -0.531   82.4%   267 | 57  <br>  Hoofdklasse Softbal   NED     142   0.268   -0.626   78.8%   272 | 73  <br>  Softball Ladies D1    BEL     119   0.213   -0.844   91.9%   409 | 36  <br>  Extraliga žen         CZE      96   0.178   -0.544   77.0%   211 | 63  <br>  Softball Bundesliga   AUT      62   0.221   -0.932   85.8%   169 | 28  </pre><p>As visible on tables, the values for <em>runSB</em> and <em>runCS</em> vary quite substantially from 0.178 to 0.268 for <em>runSB</em> and -0.444 to -0.932 for <em>runCS</em>, a 51% and 101% increase, respectively. Getting caught stealing in the Austrian Softball Bundesliga is twice as bad as getting caught in the Czech Baseball Extraliga, for instance. As already established, <em>runSB</em> is calculated by looking at the run-expectancy change of all <em>SBs</em>; therefore, it is no surprise there is only little correlation between <em>runSB</em> and <em>lgSB%</em> (25% roughly) but significantly more correlation between <em>runCS</em> and the <em>lgSB%</em> sitting around 60%, as <em>runCS</em> is calculated by <em>RunsPerOut</em>, which is immediately impacted by caught stealings.</p><h4>Summary</h4><p>Looking back at the statistics of someone like Will Smith, an analyst reading those statistics would likely tell his base coaches to be conservative against Will Smith. As a result, next year Will Smith would probably have significantly fewer attempts against, which will have a negative impact on his <strong>wCS</strong>, and that makes sense as he will not be catching baserunners like he did previously. However, base stealing is a valuable part of offense, and Smith might not catch runners, but he is still preventing them from stealing, simply by his reputation — and that is also valuable.</p><h3>Respect Value</h3><p>The value of preventing runs with the respect the opponents have for you sounds a bit strange at first, but will hopefully make sense soon.</p><p>The formula to calculate the <strong>Respect Value </strong>is significantly more difficult than <strong>wCS </strong>for one reason, and that is that in <strong>wCS</strong> you analyze what happened, while for <strong>Respect Value </strong>you will need to look at what <em>did not but could have </em>happened. You will need to look at all base-stealing opportunities, which is already quite difficult as you will need a full database of play-by-play data of a tournament — rather than just box score metrics like <em>SB</em> and <em>CS</em> and the linear weights that are publicly accessible for the MLB via <a href="https://www.fangraphs.com/guts.aspx?type=cn">fangraphs</a>.</p><h4>Calculating the Respect Value</h4><p>Once you have that available, it will need you two constants: the attempt rate <em>AR</em>, which is calculated by taking the attempts divided by all possible attempt situations, and secondly, the attempt value <em>AV</em>. The value is calculated by taking the sum of the <em>SB%</em> multiplied with the runSB and the <em>CS%</em> (which is <em>1-SB%</em>) times the <em>runCS</em>:</p><blockquote>av = (SB% · runSB) + ((1 - SB%) · runCS)</blockquote><p>Before we dig into the individual calculation, what defines a stolen base or attempt opportunity? Generally, if a runner has an empty base in front of them, they may steal. But what if it is occupied? You would not immediately be able to steal, but if that runner in front of you is going, you can, and it will count as an attempt as well. Whatever one may define as opportunity, it is important to stay consistent. If you look at a 1st and 2nd situation and disregard the runner on 1st as a potential runner (given they are a trailing runner), you may not count a double steal as two attempts (as you would only have one opportunity but two attempts).</p><p>Now it is time to look at the individual stats; neccessary are only the total amount of attempts against one and the total attempt opportunities. With the same principle as before, we calculate the performance of a player above-/or below average. So take the difference of individual attemptrate (attempts/attempt opportunities) and league attemptrate. That result is multiplied with the attempt value, which we established beforehand. The final result can be both negative and positive, with 0 displaying the average as for <strong>wCS.</strong></p><p>Here is the leaderboard of the Women’s Softball Euros 2024:</p><pre>           Player                 Team        Games as C   Respect  <br> --------------------------- --------------- ------------ --------- <br>  Mariah Elisabeth JAMEYSON   Switzerland              5      1.15  <br>  KAITLYN MARIE CALEEN        Ukraine                  7      1.01  <br>  Eliška SLANINOVÁ            Czechia                  6      0.97  <br>  Cara EDWARDS                Great Britain            8      0.83  <br>  Rebeka VIGH                 Hungary                  3      0.79  <br>  CHELSEA LYSIKATOS           Greece                   4      0.72  <br>  Elisa CECCHETTI             Italy                    6      0.71  <br>  Gitte JANSSEN               Belgium                  5      0.62  <br>  Emma SYRJÄLÄ                Finland                  7      0.47  <br>  Chiara ENRIONE-THORRAND     France                   7      0.39  <br>  Paula DUVNJAK               Croatia                  4      0.38  <br>  Laura WISSINK               Netherlands              6      0.27  <br>  Lea KARNER                  Austria                  5      0.26  <br>  Sofia ZEPEDA                Greece                   5      0.23  <br>  Mèlissa MAYEUX              France                   2      0.20  </pre><p>What this tells us can be summarized shortly as ‘Kaitlyn Caleen saved roughly one run in her 7 games by her opponents conservative approach to stealing against her’. But how exactly did we get to 1.01?</p><p>The attemptrate <em>AR</em> in this tournament was 9.37% (251 attempts on 2680 chances), the <em>runSB</em> and <em>runCS</em> values were 0.348 and -0.584, respectively, and the successrate of steals was around 83% (209 successes to 42 fails). This gives us the following formula for the attempt value <em>AV</em>:</p><blockquote>0.83 ⋅ 0.348 + 0.17 ⋅ (-0.584) = 0.190 AV (without roundings, it’s about 0.192).</blockquote><p>So those are our constants <em>AV</em> and <em>AR</em>; her individual values are 120 chances and 6 attempts against. With those values, we can insert into the formula:</p><blockquote>((120 Chances ⋅ 0.094 AR) — 6 Attempts) ⋅ 0.192 AV = 1.01 Respect</blockquote><p>And just like that, we have calculated the runs saved by respect.</p><h3>Final Summary</h3><p>The statistics <strong>wCS </strong>and <strong>Respect Value </strong>can both be used individually or combined to value your own or your opposing catchers. Calculating them requires a large set of data, which will always be the hardest thing to acquire. If you do have that, you will still need a good bit of coding.</p><p>The <strong>Respect Value</strong> for one has more purpose in judging and evaluating players but is more or less useless in a game. It is a nice statistic to tell how good and important a catcher is to their team.</p><p><strong>wCS, </strong>on the other hand, can be used to scout catchers and their ability to throw out runners and is probably the best tool for that exact purpose. It is also rather easy to calculate once the linearweights are established (otherwise, default values could do a similar job).</p><h3>Other uses</h3><h4>Pitchers</h4><p>While the primary idea behind the statistics might have been evaluating catchers, one can not ignore the factor of a pitcher in both of those statistics. You can easily attribute a <strong>wCS </strong>value to a pitcher as well and actually include pickoffs (which I have not for catchers). Moreso, you can take a look at a pitcher’s respect values, which can tell something about their velocity of pitching and ability to pick off.</p><h4>Battery</h4><p>If you can make it for both pitchers and catchers, you might as well do it for both combined and create statistics for each combination to find the best battery in that regard.</p><h4>Runners</h4><p>As mentioned before, runners already have <strong>wSB,</strong> which is the exact opposite of <strong>wCS. </strong>However, <strong>Respect Value </strong>is something that can be used for a runner as well and can be interpreted as aggressiveness — how much does a runner steal compared to the average? This can be simply based off of chances, or it can also be based on catchers and their respect, with the expected attempt rate being the average of all catchers the runner faced. The naming for that is still up for debate (Aggresiveness, Anti-Respect, Disrespect,…?).</p><h4>Teams</h4><p><strong>wCS </strong>and <strong>Respect Value,</strong> have little value looking at a team, as teams don’t use so many different catchers and to look at each individual catcher is much more valuable. However, looking at the cumulative Runners-<strong>Respect Value</strong> (which I’ve mentioned above), you can tell how aggressive or conservative a team is, and that might be the most valuable statistic of all that I just mentioned.</p><p>There are probably dozens of more ways to utilize those two statistics and you can write an analysis on all of them, but I will not do that in thise article. It is likely they will be featured in future articles however.</p><h3>Afterword</h3><p>I hope you enjoyed reading this article and I’d love to hear opinions and ideas on utilizing these statistics and/or finding different methods for <strong>wCS </strong>and <strong>Respect Value</strong>. The strive for the statistical edge by sports-analysts will never be satisfied after all.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8b271b3a4978" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[wRC+ or wRC — One of MLB’s most popular stat and it’s overlooked sibling]]></title>
            <link>https://medium.com/@vschuh/wrc-or-wrc-one-of-mlbs-most-popular-stat-and-it-s-overlooked-sibling-f123cbbe7131?source=rss-349b8b39602e------2</link>
            <guid isPermaLink="false">https://medium.com/p/f123cbbe7131</guid>
            <category><![CDATA[mlb]]></category>
            <category><![CDATA[baseball]]></category>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[sports]]></category>
            <category><![CDATA[analytics]]></category>
            <dc:creator><![CDATA[Valentin Schuh]]></dc:creator>
            <pubDate>Wed, 20 Dec 2023 17:14:08 GMT</pubDate>
            <atom:updated>2023-12-20T23:21:39.814Z</atom:updated>
            <content:encoded><![CDATA[<h3>wRC+ or wRC — One of MLB’s most popular stat and it’s overlooked sibling</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XEt6Eka20HSIK-b2NMHe5Q@2x.jpeg" /></figure><p><em>wRC+ or wRC what is the better stat</em>. A question a stat-nerd or someone who explores Sabrmetrics a lot might want to answer immediately. <em>Obviosuly wRC+ is better, the “+” stands for an extra step therefore it must be better, right?</em> But this question is a lot more philosophical than it looks and should raise some more questions before using a statistic to evaluate a player. But before we get to the bottom of that, let us look at both stats thoroughly.</p><h3>What are wRC and wRC+ actually</h3><p>Both are stats that fall under the term <em>sab(e)rmetrics</em>, which was coined by Bill James and stands for <strong>S</strong>ociety for <strong>A</strong>merican <strong>B</strong>aseball <strong>R</strong>esearch — Metrics. You could write several articles about sabrmetrics alone, but that is not the goal of this article, so to keep it short, sabrmetrics are statistics based on complicated formulas that are meant to show a players value, quality and whatnot in simple numbers. Most famously WAR (Wins Above Replacement) is <strong>the</strong> sabrmetric. It is a cumulative statistic to put a players overall value in one stat, based on three other sabrmetrics.</p><p>So what does wRC(+) represent? Best way to answer this is by simply decrypting the acronym. <strong><em>W</em></strong><em>eighted </em><strong><em>R</em></strong><em>uns </em><strong><em>C</em></strong><em>reated (</em><strong><em>+</em></strong><em>)</em> is what it stands for and this might already answer it. In better words the statistic is meant show how many runs a player creates.</p><p>Lets look at the formula for wRC:</p><blockquote><em>wRC = (((wOBA-League wOBA)/wOBA Scale)+(League R/PA))*PA</em></blockquote><p>And for wRC+:</p><blockquote><em>wRC+ = (((wRAA/PA + League R/PA) + (League R/PA — Park Factor* League R/PA))/ (AL or NL wRC/PA excluding pitchers))*100</em></blockquote><p>Easy right? Of course anyone whos not proficient in sabrmetrics will have a hard time decrypting these formulas, but it actually isn’t all too difficult just stick with me for a bit. wRC essentially takes a players wOBA above league average and adds the average runs per plate appearance in the league. This is a cumulative statistic and therefor gets multiplies with the plate appearances of a player. The only question that remains is, what is wOBA.</p><p><strong><em>W</em></strong><em>eighted </em><strong><em>O</em></strong><em>n-</em><strong><em>B</em></strong><em>ase </em><strong><em>A</em></strong><em>ver</em>age, short wOBA, is a statistic developed by Tom Tango in his book called <em>“The book: playing the percentages in baseball”. </em>It takes each outcome of a plate appearances, gives it a factor and divides that with the total amount of plate appearances. The factors are supossed to represent the likelyhood of scoring a run, for example Home Runs have a factor of 2.014 in the latest version, which you could interpret as <em>“A home run is worth a bit more than 2 runs”</em> (which makes sense considering there might always be a player on base).</p><p>Knowing what wOBA is, we can summarize wRC as:</p><blockquote>A statistic that represents how many runs above average a player produced over the course of a certain span.</blockquote><p>Now let’s look at wRC+. At first sight this looks way more complicated than just<em> an extra step,</em> like I promised, but it actually isn’t. wRAA is not entirely different to wRC, it has the same formula, only without the league norming (putting the value compared to the rest of the league) and the plate appearances. The league norming is actually done in the same step in this formula, and the stat is not cumulative, therefore it does not need the mulitplication with plate appearances (as wRC+ is<em> per P</em>A). The extra step that remains is the norming to park factors. As not every park has the same dimension, not every park has the same probability of runs being scored. Park factors are usually based on 5-year trends and simply analyze the tendencies. After the parks are taken into account, we only divide the result with the average wRC per batter (compare it with the league average) and multiply it with 100. This means if you create runs like the average player, you will have a wRC+ of 100.</p><p>In short the formula means: wRC+ takes wRC multiplies it with the park factor and compares it to the league average wRC per PA. The result is displayed on scale with the basis of 100, and is <strong>not </strong>cumulative.</p><p>To put this into words:</p><blockquote>wRC+ is statistic that represents how many runs you would produce compared to the average player in a neutral park on a scale with the basis 100</blockquote><h3>Quality over Value?</h3><p>The most important difference between those stats is that one is cumulative the other one not, therefor independet of plate appearances (with 100 being average). So what is the better stat? If we are going by mainstream-baseball-analytics, you would say wRC+ obviously, nobody uses wRC, while wRC+ is beocming one of the most popular stats. But then again, WAR is a far more popular stat than its (almost) PA-Independet counterpart WAR/162, which is simply the WAR a player would have in 162 games.</p><p>The actual question should not be <em>wRC or wRC+?</em>, but <strong><em>Quality or Value?</em>. </strong>And there can’t be a general answer to that, there is no <em>better </em>stat, there are situations where a cumulative stat (a Value-Stat) is more appropiate, and there are situations where it is more fitting to use non-cumulative stats (a Quality-Stat).</p><p>This is a question that evidently is not limited to wRC and wRC+, if you take a moment, just to think about any stat in baseball, football, basketball even in business it applies. The essential question is:</p><blockquote><em>Do you want someone who’s the best at what they are doing, or someone who delivers the most value</em>.</blockquote><p>In sports this means:</p><p>Do you want someone who throws 300 yards per game but goes 20 for 35 or someone who throws 250 yards and goes 15 for 20.</p><p>Do you want a player who goes 12–31 on the field with 35 points or someone who goes 11–14 with 26 points.</p><p>At last, do you want a player with the highest wRC+ or the player with the most wRC.</p><h3>When to use what</h3><p>I cannot tell you when to use what statistic, however there situations where it’s objectively better to use one over the other. For me, it makes sense to use wRC in retrospect on a fixed time. That means, you look at a full season, a month or even week, for me it matters how much value you produced, rather than how <em>good</em> you were. A team benefits more from a players actual value, rather than their <em>potential </em>value (had they played all games).</p><p>An example would be the 2023 Season and the MVP Voting:</p><p>If youre asking yourself, who is the best batter in the mlb, then wRC+ should be the stat to use. The answer then, will be Shohei Ohtani, not Acuña, and not really close either. But Ohtani doesnt rack up the most value, that would be Acuña, and that is also not particularly close. Ohtani actually ranks 6th among the mlb in wRC, given his lack of plate appearances. Also worth mentioning, the 3rd highest wRC+ among qualified batters was achieved by Corey Seager, who ranks 11th in wRC, given he has only 119 games to his name this year.</p><p>This outstanding wRC+ was the reason why Seager finished 2nd in the AL MVP voting only behind Ohtani and ahead of his teammate Marcus Semien. Semien had more value in batting, fielding and base running, because he played a significant amount more than seager. Seager was considerably <strong>better </strong>in batting than Semien.<strong> </strong>But MVP stands for most <strong>valuable </strong>player and not <strong>best</strong> player. Had it been latter one, Mike Trout should’ve won a few more of these.</p><p>In the NL it was less of a difference in value- and quality-stats. Of course you could’ve made a case for Betts as he had signifact better defensive value, but overall Acuña had more value in batting and base running which ultimately won him the award — not undeservingly.</p><h3>In Summary</h3><p>In order to adequately evaluate it is important to not just use proper statistics but also to understand them and know how to use them. The general public is not a sabrmetrician and doesn’t know about the formulas of stats and it doesn’t understand the difference between a stat with or without a “+” — even though it might be a crucial difference. Statistic websites such as <a href="https://www.fangraphs.com/">FanGraphs </a>or <a href="https://www.baseball-reference.com/">Baseball Reference</a> offer detailed glossaries about their stats, the formulas and information about how to use them.</p><p>In a Baseball world where statistics are getting more and more cryptic, glossaries can be a good guide to understand what the <em>baseball youth</em> is talking about. Or you can just stick with the good old batting average, RBI’s and OPS.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f123cbbe7131" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>