Um, Dave? Pretty sure I understand what’s going on here, but your projected final winning percentages are different for teams with the same record. Given that you’re projecting wins, and not including a decimal point, I would think you might want to recalculate winning percentage based on your projected record. Or report projected wins and losses with a decimal point, so it was more clear what’s happening here.
Also, I hate to be the guy that complains about my team, but I must ask how Pablo Sandoval has more projected playing time than David Wright. Wright’s been quite healthy every year except 2011 while Sandoval has missed significant time each of the past two years.
Also, would there be a benefit from adding DH time to the NL teams for the 20 or so games they play against the AL. Similarly, should the PA projections for the AL DHs be quite so high, seeing as quite a few games won’t feature a DH.
Right, but my point is this is exceptionally confusing to the casual Fangraphs visitor. They see St. Louis and Atlanta projected with the same number of final wins, and then a different winning percentage. If you were to report the number of wins to the first decimal place, instead, it will ameliorate that problem considerably.
Can you sort by division? I just tried to click on the NL Central and it wasn’t able to limit the page to just NL Central teams. Sorry to be the one to complain, great feature to the site I look forward to exploring it more!
Awesome features. Some equivalent of the leverage index for an entire season for a season seems like it would be possible now with this and totally cool. Beers are on me tonight!
Comment by All Balls No Brains — May 9, 2013 @ 11:35 am
I think the solution being proposed is to first use the actual percentage, then round to the nearest win, then go back and change the percentage to reflect that w/l record. If that is not right, it should be.
Comment by Tyler K Patterson — May 9, 2013 @ 11:39 am
For the sake of making things more understandable I think it would be good if the depth chart pages state that they are Rest of Season projections. It says that on the menu dropdown but if you go straight to one of the pages (like say from the links in this article) that isn’t clear.
The main differences between forecasting systems is going to be the amount of regression applied to different inputs and different aging curves. Steamer has spent a lot of time working on their pitching projections, tweaking the amount of regression based on velocity, and incorporated the infield fly/z-swing data that Dan Rosenheck wrote about a few months ago. ZIPS uses velocity as a factor too, but they probably aren’t handling it in the exact same way.
Playing time projections are inherently going to be a little subjective. They’re probably not perfect, and we’ll tweak them every day in order to try and make them as accurate as possible. Sandoval will likely be projected for a little less playing time when you wake up tomorrow.
2) When sorting any specific column (W%, WAR, wOBA, etc.) the first click is sorting from lowest to highest, which in most cases puts the worst team/player at the top. I’d imagine reversing the sort would be more intuitive.
3) I’ve noticed a strange projection from a player who only has a Steamer projection (no ZiPS) listed on their Player Card. Steve Pearce has a ROS line of .248/.331/.418 with a .326 wOBA, but on the Orioles Depth Chart he’s listed as .250/.308/.333 and .286 wOBA. Also, Russ Canzler has a .311 ROS Steamer wOBA, but .296 Depth Chart value. On the other hand, Chris Dickerson has both ZiPS and Steamer ROS projections on his Player Card (.293 & .318), but his DC value is just .294. From the players with more playing time there also seems to be a slight difference from the median or weighted average of ZiPS and Steamer projections, but not as severe as those I’ve mentioned.
I’m not so bothered by it in the first place, but is it wrong to just put a note at the bottom indicating that: “Pythagenpat outputs winning percentage, not number of wins. The number of wins is rounded to the closest whole number.”
Speaking of Carlos Gomez … your depth chart has him for 500+ PAs remaining, while both the ZIPS and Steam ROS projections have 300+ remaining. So where, exactly, are you getting your depth chart information from? Or, more to the point, is there a more central place to get it than walking through 30 individual team pages?
The playing time forecasts from ZIPS and Steamer are based on regressed historical playing time data, so if a guy wasn’t a starter in previous years, he won’t be forecast for starter’s playing time in 2013. That’s one of the advantages of manual playing time forecasts – we know things that the algorithm doesn’t, so we can make adjustments based on changes to a player’s role or changes within his team’s roster makeup.
The depth chart information is entered by a team of our authors, who are maintaining them on a daily basis. Those depth charts can be viewed on a team or position basis from these pages.
I ask because I export FG data to compile my own fantasy-league specific projections on a weekly basis. I can get ZIPs/Steamer ROS data easily enough from your projections pages, but exporting anything from these awesome new depth charts is going to be … unwieldy. Any chance this sort of thing makes it’s way to your more general projections page?
With regard to playing time, National League DHs are projected for far more playing time than is actually possible; Giants DHs, for instance, are projected to come to the plate 237 times. Given that National League DHs are generally terrible, the effect seems to be to skew the overall projections for National League teams downward.
This is amazing stuff. Thank you Fangraphs. However, I have to echo the first couple of posts. This really is not a projected W/L for the rest of the reason if your not going to ‘sim’ it over the actual remaining schedule. This is more of a Power Ranking as it assumes each team plays an equal / balanced schedule the rest of the way. This will especially effect teams that played a heavy Astro or Marlin schedule to date.
It began with the forging of the baseball websites.
Three were given to the fantasy players, neurotic, fanatical, and most obsessive of all fans.
Seven were given to the stat-heads, great minds and thinkers of their mothers’ basements.
And nine, nine websites were given to the common fans, who, above all else, desire favorable coverage of their favorite teams.
But they were, all of them, deceived, for another website was made: In the land of Arlington, in the fires of the internet, the dark lord Appleman forged in secret a master website, to control all others. And into this website, he poured his statistics, his analysis, and his depth charts and standings. Fangraphs: one website to rule them all.
I agree – when following team links I was unsure what I was looking at. Are these rest of season projections, year to date (I know it isn’t this because of the PA/IP numbers) or is it end of season projections. Or better yet – a link for each so a viewer can choose. At the very least put a title on the page.
To put a finer point on it, right now when I mouse over “Standings” I see the Blue Jays listed in first place in the AL East. (I know that’s because the team names are sorted alphabetically, but an alphabetical sort does not seem to be the optimal method of sorting teams when the site implies that one will be viewing the standings.)
One thing that is really helpful from a fantasy standpoint is batting order. I don’t know how much more work that would require from you folks. But without it, well, I’ll be inclined to instead visit a site that does include that. For examples, Rutledge, Simmons and Revere are 3 guys quite useful at the top of their lineups, much less so at the bottom thereofs.
Well, I’d actually be very interested in testing that. Since they’ll be changing every day, it would be great if someone could screenshot them now and see how they do at season’s end. (Tango is perhaps doing so as we speak.)
Other point of this – what is replacement level for DHs these days? It seems that virtually every pinch bat in the national league (the guys making up the DH list for the most part) is worse than some theoretical AAA player. That just can’t be right.
On another note, how is Evan Gattis projected to have a higher wOBA than Adam Dunn, but he is negative while Dunn is positive? There is definitely some weirdness going on at DH.
“Who is playing second base for the Marlins these days?”
But these depth charts don’t actually answer that question. They answer “what kind of production will the Marlins get out of second base the rest of the way, and from whom?”. Looking at the page I can’t tell if it’s Valaika, Solano, or a platoon of them both.
Love the new addition, but that section is more “projections sorted by position” than actual depth charts. I’ll still rely on Jason Martinez for actual, current depth charts (which tells me that someone named Derek Dietrich is their starting 2B).
I hate to keep making criticisms because overall this is amazing, but I have to agree here.
I was curious to see what the FG consensus* was on Dietrich’s playing time the rest of season, but was disappointed that the depth charts didn’t even show getting him any playing time, despite the fact that he made his MLB debut last night. This seems like exactly the sort of thing that should have been updated this morning if there really were daily updates being made.
*Or is it just one person for all teams, one person for various subsets of teams? If so, who is doing the daily updates?
And might I suggest that you figure out one of those classic FanGraph solutions for showing platoons, as well? Boston seems like a good example right now, where Nava will usually start against RHP, and Gomes will usually start in his place against LHP.
Any chance you can give us those combined zips and steamer projections? It would be really helpful if under the projections tab we had an option for a combined zips/steamer projection list.
I wonder what would be the best way to make it apparent that it does sort by depth chart. Maybe a column on the left where a player is labelled as “Starter” or “Reserve” or “Platoon vs LHP” or “Platoon vs RHP.” I think that would be pretty versatile, as it would help clarify situations like, say, a player starting at 1B vs RHP and at DH vs LHP.
I understand that the ZIPS/Steamer projections are influenced at least in part by the player’s season to date stats but is that also true for minor leaguers who are expected to play later in the year? In other words, are the projected stats for current minor leaguers affected by their performance in the minors so far in 2013?
I second that. Once that is done, from a fantasy standpoint the final missing link would be player position eligibility (based on Yahoo or some other logic) in an exportable format. If that data were available it wouldn’t be hard to build something like a roto trade analyzer (based on z-scores) that would project RoS roto value (auction $ and rank) for every player. I think fantasy gamers would love a resource like that. On the downside it could level the playing field for all those fish out there.
Christmas in May….seriously the best site on the web.
Comment by fantasysportshawk — May 10, 2013 @ 8:49 am
Dave – to echo some comments above, it would be awesome if the “synthesized” projections used in these depth charts were available in the projections section for sorting / exporting. For those of us who download RoS projections for use in our personal fantasy calculators it would be very valuable, since these projections not only blend Steamer and ZiPS but also have manual playing time estimates (a key addition).
An obvious example is someone like Carlos Gomez; because he was a part time player the past few years, is projected RoS for only 351 PA by ZiPS and 373 PA by Steamer. The “synthesized” projection has him at a much more reasonable 534 PA RoS given that he is a full-time starter.
This is an outstanding new feature of FanGraphs, to be sure. That said, there are some real head-scratching results. I find it difficult to picture the Astros winning at a .410 clip for their remaining games. Why would they improve by such a large margin? There is nothing in their performance to suggest such a dramatic improvement, especially against consistently superior competition dominating the rest of their schedule.
Also, how much does recent performance alter the projections? E.g., the Bluejays just won four games in a row, and scored in double-digits in each of the past three games. When you factor that in with Jose Reyes returning likely in July, Josh Johnson returning very soon, etc., they should improve substantially over the remaining games, but they are expected to play .500-ball in the rest of the way? I’m not saying they will win the division at this point, but I don’t think that 80 to 85 wins is out of the question with the talent they have returning to their normal career curves.
Comment by Ivan Grushenko — June 12, 2013 @ 1:46 pm
I’ve no idea what the point of this comment is, but now that you’ve brought it up, the real surprise would be if some team like 2012 Oakland, which hadn’t been picked to do anything, and hadn’t had a good first half, all of a sudden won everything in the second half and made the playoffs. If I had to pick a candidate to do that, I might pick the Cubs or Padres.
Comment by Ivan Grushenko — June 12, 2013 @ 1:53 pm
I cannot help but notice that as the season moves along the discrepancies between what the projected numbers and the actual numbers are turning out to be.
In particular I am looking at the Blue Jays. For example, in projections you have Casey Janssen with 39 IP (at 21.2 IP now), Brett Cecil with 33 IP (at 34 IP now), Steve Delabar with 39 IP (at 34 IP now), JA Happ with 17 IP (at 33 IP)… I could go on but you get the idea. So there is a strong disconnect between the productions and the actual numbers, at least on the pitching side.
The other problem I am seeing that the the projected WAR seems off as well (I am staying away from the pitching side as I know that can be more inconsistent than batting). Jose Bautista, is projected at 3.0 WAR with 369 PA(he has 2.5 WAR and is at 266 PA) Edwin Encarnacion 2.1 WAR with 407 PA( he has 1.6 WAR and 291 PA), Maicer Izturis 0.2 WAR with 260 PA (he has -1.4 WAR in 188 PA).
It just seems that the projections sharply disagree with reality doing simple comparisions. I hope that is Izturis can generate 1.6 WAR in the next 70 PA, but I don’t see that happening. I don’t think that the projects should just ignore it either.
This is a great idea, but as a casual user of the site, I was disappointed there is not a glossary of the column headers at the bottom for reference. RDIF and RS/G do not appear to be on the Glossary page either. I’m sure most everyday users of the site know for certain what these represent, but I am not certain. Thanks for considering.
Data in the depth charts is wrong for a good portion of the players. For example, Nick Franklin triple slash in depth chart 247/307/372. This is not even close to correct, even if it included minor league numbers for the year. True mlb slash currently at 309/397/509. This is just one of many completely wrong datalines on the mariners depth chart. Whats up with that???
I just went throught hte same thing as mrlewis above. Since we’re around the halfway point of the season, the numbers, at first looked like they were supposed to be season-to-date. It would be very easy to add a heading/note to the top of the page indicating the Depth Charts area is *projected* for the rest of the year, and perhaps a link back to this original post introducing them. Another suggestion (apologies if someone already made this suggestion above) would be to include the actual season-to-date pitching WAR and batting WAR along side the projected rest of season WAR. Would be interesting to see. Thanks for the great work!!
Comment by tribe Fan in SF — June 26, 2013 @ 2:16 pm
Funny how the run differential for the entire season favors the Reds but the Pirates are projected to finish one game ahead…Cincy needs one more outfielder, but that is the easiest thing in the world to get on the cheap. I have to believe the run differential should spell at least 90 wins for the Reds and a Wild Card if they cannot catch the Cards. Why the love for the Pirates I wonder?
Just wanted to chime in. Love the depth charts, great to see this broken out this way.
Question: How are PA counted for pinch hitters? For example if John Mayberry comes in to pinch hit for Delmon Young, and then stays in RF, that PA should be under RF. If he comes in to pinch hit for the pitcher, and does not stay in the game, that should be counted different. IMO that should be under a pinch hitter section.
Would like to see games played for each section, but that may clutter things up.
Why do the Projected ROS team records differ in the full standings (all 30 teams) vs the division standings? e.g. the Red Sox are projected to finish their final 3 games 2 – 1 (.574) on the first table but 1 – 2 (.431) on the second.