SBNation.com User Blog: splanket
http://www.sbnation.com/users/splanket
Posts made by splanket on SBNation.comOn the Significance of True Shooting Percentage
http://www.thedreamshake.com/2014/7/24/5935665/on-the-significance-of-true-shooting-percentage
splanketThu, 24 Jul 2014 23:05:37 -0400
<p>Hey y'all, I've been volunteering to do the math behind a few of my comments recently stating that "A difference in TS% of 9-10 points, when extrapolated to first options, all other things being equal, results in a difference in the neighborhood of 10-11 Pythagorean Expected Wins." Well, I'm bored (read: supposed to be studying for my Organic Chemistry 2 test) and figured I'd go ahead and show the math and post it here. Now, obviously, there are many, many more factors that affect the outcome of a season that just one player's true shooting percentage, but I'll get to that later.</p>
<p style="margin-top: 5px; margin-bottom: 5px; padding: 0px; font-size: small; line-height: normal;">First off, I'll start by explaining TS% a bit. The formula used to calculate it is (PTS/(FGA+.44*FTA)), and essentially the goal of the metric is to calculate a player's expected points per "shooting chance" (if you multiply TS% by 2, you get exactly that number) which is either a 2PT or 3PT shot (and if made and fouled, the and-1 free throw), or a trip to the line to shoot two. This should explain the presence of the .44 in there, it is a conversion factor for the league average rate of and-1's to trips to the line for two, since an and-1 is not actually it's own "shooting chance" but is tied to the made shot before it.</p>
<p style="margin-top: 5px; margin-bottom: 5px; padding: 0px; font-size: small; line-height: normal;">Anyway, on to the math. First, let's assume Player A is the first option on a team that is the absolute epitome of average. His TS% is exactly league average (around .530 each year), and his team goes exactly 41-41, scoring exactly 100.0 points per game while giving up exactly 100.0 points per game. Next, we'll assume Player B plays exactly the same as Player A, same amount of shooting chances, assists at the same rate, turns it over at the same rate, plays the same defense, and has identical team-mates. The only difference is, Player B shoots 9.0% better in terms of TS%.</p>
<p style="margin-top: 5px; margin-bottom: 5px; padding: 0px; font-size: small; line-height: normal;">Now, going off what I explained above, a difference of 9.0% in TS% means a difference of .18 expected points per "shooting chance." Using Steph Curry as a random first option example, he took 1383 FG attempts and 348 FT attempts, good for 1536.12 (we'll round to the even number) "shooting chances," or 19.69/game. Since we assume Players A and B both took the same amount of shooting chances and are both first options, we'll use this number. Player B gets .18 expected points more per shooting chance, when multiplied out that's 3.54 expected points per game more for Player B. Remembering that we said every single thing about players and their teams except the TS% was equal, that should mean that Player B's team should score 103.54 points per game while giving up 100.0 points per game.</p>
<p style="margin-top: 5px; margin-bottom: 5px; padding: 0px; font-size: small; line-height: normal;">Now that we've established all that, we can plug the results into a Pythagorean Wins formula. If you don't know what this is, it essentially takes a team's point differential and says how many games the team was expected to win over the course of the season. It is generally incredibly accurate to real wins, with more than half the league no more than one game off their Pythagorean Expected Wins. The formula most commonly used is Win%=((Points For per game)16.5 / ((Points For per game)16.5 + (Points Against per game)16.5 ). Since Player A's team scores 100.0 points per game and gives up exactly that many, their expected win% is 50.0% for a record of 41-41. Player B's team, however; scores 103.54 points per game while giving up 100.0. Plugging this into the formula, you get (103.54)16.5 / ((103.54)16.5 + (100)16.5 ), which comes out to an expected win% of 63.97% for a record of 52.45-29.55, or 11.45 games better than player A's team.</p>
<p style="margin-top: 5px; margin-bottom: 5px; padding: 0px; font-size: small; line-height: normal;">Now, obviously there are shitloads more factors affecting the game than just the best player on a team's TS%, but an 11.45 win difference is huge. That's the difference between the Twolves and the <a class="sbn-auto-link" href="http://www.sbnation.com/nba/teams/golden-state-warriors">Warriors</a> last year. If y'all have any questions on the math or want to know any more about any of this, feel free to let me know!</p>
<p style="margin-top: 5px; margin-bottom: 5px; padding: 0px; font-size: small; line-height: normal;"><strong>TL;DR: For a first option on a team, all other things being equal, each additional point of TS% is worth a little over 1 Pythagorean Expected Win.</strong></p>
"Offensive tackle Ryan Harris passed his physical and is officially signed with the #Texans now." -...
http://www.battleredblog.com/2012/9/2/3287602/offensive-tackle-ryan-harris-passed-his-physical-and-is-officially
splanketSun, 02 Sep 2012 16:45:40 -0400
<blockquote>
<p><p>"Offensive tackle Ryan Harris passed his physical and is officially signed with the #Texans now." - Tania Ganguli, Texans Reporter, via twitter</p></p>
</blockquote>
<div class="source"><p><a href="">https://twitter.com/taniaganguli</a></p></div>