It’s Alive: Price Guide 2013

The Last Player Picked Price Guide is back online:

http://lastplayerpicked.zxq.net/

CAIRO is up already (with Kris Medlen as the #4 pitcher?).

Updates to follow as more projections are released.

About these ads

92 thoughts on “It’s Alive: Price Guide 2013

  1. Mays, what happened to the old site ? In any case, you have a wealth of great articles on it that you posted. Any plans to put them back up ?

  2. Mays! great to have you back. what happened? I’m sure many of us were worried about you!

    One question – did you have Quality Starts in prior versions of the price guide?

  3. Mays – very glad you’re back. Love the Price Guide.

    I realize that CAIRO comes from SG, but any thoughts on why it’s projecting CarGo for 150 AB’s?

    • probably the simple answer is that it’s early and not enough eyeballs have looked at these projections to correct mistakes like that. Cairo is at version .1

      • I double-checked, and the error is in SG’s spreadsheet (in both v0.1 and 0.2)…

        I’m guessing he mixed in one or more of the minor league Carlos Gonzalezes… (Just quadrupling the projection to make a full season still ends up pretty weak.)

      • Good point – that got me to double-check v0.2 (which is probably what I should have done from the beginning). You’re right about the playing time error in his spreadsheet, but he projects CarGo at .292/.354/.490, which seems pretty impressive for one of the minor league CG’s. His playing-time-adjusted CarGo comes out to 598 PA’s, 76 R, 20 HR, 86 RBI, 10 SB (and of course, .292/.354/.490).

  4. Mays,
    Tried to upload some custom projections and got an error. Downloaded known-good .csv files from 2012 Steamer and got the same error:

    Warning: move_uploaded_file() [function.move-uploaded-file]: Unable to access ../../cusds/fuhbrwfcBatting.csv in /www/zxq.net/l/a/s/lastplayerpicked/htdocs/upload.php on line 50
    Dataset Upload

    Files upload failed.

  5. I think the Streamer projections are out now, at least partially, Razzball has them.

    Is there anyway we can get a holds category on the projections this year?

    • Yeah, I was just looking at Steamer. It’s going to be a bit of a hassle to convert it to something usable (They usually use the same mlbamids that I do, but the Razzball version has players with a mix of BIS IDs and some other ID.)

      • They are not even on the steamer blog yet. Maybe this is just a special format for Razzball and they will officially launch their own later.

  6. anyone else having trouble uploading the latest steamer projections? I keep getting errors, can’t figure out what the problem is. I’m even pasting the data into old .cvs that I know work.

  7. Guys, I found the bug with Carlos Gonzalez and have fixed his projection. He got conflated with a few other Carlos Gonzalezs. His new projection is 608 PA and .300/.360/.517, which I think is more reasonable.

    • “Preset” fills in the form below with the roster/scoring settings for a standard league at ESPN, Yahoo, etc. Just trying to save a little clicking effort.

  8. Mays – I have a number of questions (and some disagreements) about your valuation method.

    1. From the csv fiels, how do you get from “total” (the players total Z-score) to adjTotal, then to dollarValue?
    2. Why are you considering positional scarcity? The goal of an owner is to maximize your total Z-score at an auction. Not sure why you change around the value of a player’s Z-score because of the position the player plays.
    3. Why do so few players come out valued at $1 when in reality there are many more players that go at $1 than what your calculations imply. In a typical auction around 2 hitters per each team are taken at a dollar.

    • I’m not Mays, but I’ve been using his method on my own for a while (because my league uses QS as a category)
      1. To get adjTotal, you subtract the lowest replacement level from the player from the positions he qualifies for. Replacement levels are negative, so this number is higher. To get dollarValue, you just need to calculate marginal dollars per marginal point, which is ((salary cap * # of teams) – (# of teams * minimum bid * number of starters) / (sum of adjusted total for all starting players). This number is then multiplied by each player’s adjTotal to create their dollarValue.

      For instance, my league has 11 teams, each with a $260 cap, with a $2 minimum. That’s $2860 (260 * 11) of players salaries. But we have 23 players per team, so replacement level salaries eat up $506 (2*23*11) of those dollars. That leaves $2354 marginal dollars. Marginal points is the sum of the adjTotal for every player whose adjTotal > 0. Because having an adjTotal of 0 is the definition of replacement level. Looking back at 2012 stats, my league had 859.42 points of hitting, and 230.41 points of pitching, for a total of 1089.83 points. The ratio of dollars to points is then 2354 / 1089.83, or 2.16. A players projected salary then is his adjTotal*2.16 + minimum bid.

      2. You need to maximize Z-score *while fielding a legal team with respect to roster limits*. I am in a 2-catcher league. That means 22 catchers need to go in the auction. If you don’t include positional adjustments, you will not have 22 catchers who rank above replacement level. This in turn would skew catchers projected dollar values (the 22nd catcher taken is by definition worth minimum salary, and the value of the other catchers is defined by how much better than this catcher they are).

      3. You’re correct that this is a limitation of this method, in that only the last player picked at each position is listed at the minimum. If Mays wanted to, he could do something similar to Football Outsiders KUBIAK projections and ask you to define how many players per team would go for the minimum, then the bottom X players would go for that, and you’d just have to recalculate marginal points for the remaining players. But this is a minor thing in the long run. If you want to add a few dollars to the top players values because you knew you’d be saving a few dollars with the bottom players, you could do so.

      • I will look into #1, thanks. I think #2 and #3 go hand-in-hand here, but I don’t think #3 is minor at all, it has material impact on dollar values. If you have only six spots (one C, 1B, 2B, 3B, SS, and OF) in Mays projections at $1, when in reality, a 12-team league has ~24 guys who go for $1, that’s a lot of money that gets displaced upward, because its not just the 24 guys, its all of the players above them. The 25th worst guy in May’s calculations might be listed as $4 but in reality he’s barely worth $2. That money has to be displaced as well.

        I am of the belief that positional scarcity should be ignored in the calculations, but a user should be aware that those players that go for $1 but are actually worth much less than $1 which is, in essence, an overbid. Separately, If you have a league full of Ryan Brauns and Jose Molinas, the “worst” OF should not be worth $1 if his output is much different than, say, a Jose Molina. Now of course the real world is different than theory, but nonetheless your worst outfielder might not necessarily be worth $1.

      • Mike D,
        I think you’re conflating some of the game theory of the auction with actual dollar values. Just because you can get a guy for $2 at the auction doesn’t mean that he’s not worth the $4 of “points” he’s projected to earn for the season. When you’re building yourself a price guide, you might want to shift some dollars from the bottom to the top, if you expect more $1 guys to be purchased than actually exist. So you can have a target or maximum price, which factors in some bidding nuances, and then an expected value. I learned the hard way last year that I could have “overspent” on some top talent, and ended up with more overall value.

  9. Dr. Pepper – how much do you find adding QS makes a difference? Unless I’m going crazy, Ian older version of the Price Guide had QS, and when I switched to it from my normal 7×5 rules (to go to 7×6) I didn’t see huge difference at all? The good SP gained $1 or $2, RP went slightly down.

    if I switch to 65/35 price split and use 7×5 rules, the prices come extremely close to how my league drafts with 7×6 rules

    • Mays’ implementation of the QS was spotty, which is probably why he removed it now. I believe for projections, he just used the previous years QS, which is going to be inaccurate for players who have changed roles or were injured for a good part of the year. I’ve projected QS based on a regression I did that showed the relationship between QS%, R/GS, and IP/GS.

      That said, there doesn’t end up being that much difference. When I switch from 6×6 with QS to 5×6 without them, the % of money going to hitting goes from 71% to 73%, which is mostly from the SP losing a couple of dollars. If you calculate, and then do a 2% readjustment to reallocate those dollars back to pitching, then you should be fine. The ranking of SP changes a little bit, but it’s probably within the margin of error for everything (considering we’re dealing with projections) anyway.

      • Ha, I remember finding someones regression to predict QS and using it. Maybe that was you? I actually remember now that I took Steamer Projections, calculated QS on my own using that regression output, and uploaded to the price guide. But yeah same conclusion as you…it doesn’t change a ton, enough that I could do without it and just do some manual fudging in excel afterwards

    • Optimal Hitter / Pitcher ratio is simply the ratio of total adjPoints of all drafted players. For instance in my 2012 projections, sum of adjPoints for all drafted hitters is 859. Sum for all drafted pitchers is 230. Ratio is 79/21. 859 / 1089 is 79,% and 230/1089 is 21%.

  10. Regarding players going for $1…

    I think that is as much about fluid auction dynamics as pure valuation. In my league last year, we had way more $1 players than we had ever had before. Our rules didn’t change at all, it’s just for some reason people went much more “stars and scrubs” than ever before. There were mad bidding wars for stars which left tons of guys available at the end for $1. other years people were calmer

    • True, but it is an important variable that affects all players’ prices that I think should be accounted for, otherwise, what’s the point of the Price Guide? There’s theory, and then there’s a realistic real world scenario which is what users are looking for.

  11. Thank you very much for this website as I enjoy using it every year. Question on keepers: If I ‘edit’ a player and type in a keeper value, does that have any effect on the listings? I didn’t see anything change. Not sure if prices should change, if a notification should appear showing a player is kept, etc.

  12. I’m relieved to see you’re back and got things running again. I left a donation for you, but I noticed it went to the email you had associated with your previous site. Hopefully you get the money. It isn’t much, but it is a token of my appreciation for the indispensable site!

    Let me know if you have trouble picking up the donation.

  13. Hi Mays – is there a way to programmatically upload/attach projection csv files? It would be AWESOME if we could put a parameter in the url that is a link to a file in a dropbox/skydrive type service. Any thoughts on how/if this could be done?

  14. Hello. Does the “donate” button work on the price guide? I won my league last year, and a key part of my strategy was using your price guide to identify market inefficiencies. I’d like to give back. Thanks.

  15. This is a great tool…I have used it for a couple years now…thanks!
    Im having a bug issue with keepers. I enter the “kept at” price in the box and press save changes, but when the chart rebuilds, none of the prices have changed. I tried with both firefox and chrome and with both the cairo and steamer projections…same story with both. Thanks for checking into it!

  16. I really wish you had holds as an option. It’s basically impossible to get a somewhat accurate value for anybody in the league without them…especially pitchers, obviously.

    • IP affects the rate stats for pitchers, the same way PA affects OBP. This is as intended. Think about it this way: which player will provide more value to your ERA at the end of the season: a 3.0 ERA pitcher who pitched 60 innings, or a 3.0 ERA pitcher who pitched 200 innings? Since your ERA is calculated at the team level, having a better than average ERA in a larger sample size is more valuable. Read Mays’ summary of how the price guide works at http://web.archive.org/web/20120402034209/http://www.lastplayerpicked.com/how-the-price-guide-works-part-i-standard-scores/ . Pay special attention to the section on rate stats, and you’ll see that this is how the Price Guide has always done it.
      Or think of it like this. When figuring out how valuable a player’s ERA is, the question we are really asking is “How many runs less than average will this player allow?” Average ERA among drafted pitchers in your league is projected to be 3.73. So if Stephen Strasburg were an average pitcher, he would allow 73 earned runs in 177 innings. Instead he’s projected to allow only 57 earned runs, so he’s 16 earned runs better than average. If Craig Kimbrel were an average player, he would allow 26 runs in 64 innings. Instead he’s projected to allow 14, so he’s 12 runs above average. The spreadsheet is making the same calculations but changing them to dollars for your auction. WHIP works the same way. Technically maybe K/BB should be multiplying by walks, since that is what is in the denominator, but I’m not 100% sure on that. I will switch K/BB if you wish, but ERA and WHIP are working exactly as intended.

      • I’m in a league which uses: W, ERA, WHIP, K, K/BB, S. 10 teams, 9 Pitchers. The Price guide SEEMS to way over rate closers (using Steamer). Papelbon = F. Hernandez at $25. Your IP effect described above would seem to contradict this valuation. Yeah Papelbon has slightly better rate stats, but Felix is is going to pitch 121 more innings. Is LPP wring or am I missing something?

      • Tom, in doing some calculations, it looks like you’ve got a league where you can start 9 P, regardless of whether they are SP or RP? That can tend to throw the price guide out of whack.

        Here’s the thing. The price guide does its calculations and decides that based on your league’s settings, if everyone was being smart, they would draft 3 or 4 SP and 5 or 6 RP. Then it figures out values based on that assumption. If that assumption isn’t true, and I’m guessing it’s not, that throws the calculations off. In this case, you need to adjust the price guide based on what everyone else is actually going to do.

        I would guess that most people in your league are going to draft 6 SP and 3 RP. I would set the pitcher positions in the price guide to that. Then understand that RP are actually more valuable than they are projected for, and it’s okay if you pay a little more than projected.

    • Thanks so much for your help!

      So it sounds like you agree that having more RP’s than SP’s might be a correct strategy, right? I’m far from a fantasy expert, but starting more RP’s seems to only help you in Saves and some in ERA he certain expense of W’s and K’s. What am I missing?

      Also there is no allocation place to but number of bench spots available on LPP. Shouldn’t this be in there or should I just use total number of players displayed to accommodate this.

      If you have another minute.. I’ve played in roto league for years – now in 10 team H2H each category. Positions are 1 each IF, CI,MI, 5 OF, UTIL, 9 Pitchers, 3 bench, 1 DL spot. Cats are HR, RBI, SB, OBP, TB, E and W, ERA, WHIP, K, S, K/BB. Any thoughts on draft strategy? Biggest questions are how to fill bench (hitters vs pitchers), how many SP’s vs RP’s and should plan on streaming given that there are 7 transactions allowed per week and daily lineup changes?

      • Like I said down below, the thing is, in roto your strategy sometimes needs to change based on what your league’s conventional wisdom is.

        Let’s get one thing clear first. Most RP do provide positive value in ERA, WHIP, K/BB, and obviously saves. It may not be more value than a top SP gives, but it’s certainly more value than say an average SP (who can be a drain in your rate stats, the same way Adam Dunn is a drain on a team’s AVG).

        As for what you should do, the way I see it, you have two possibilities to gain an advantage over your competition:
        1. Punt wins and strikeouts: Let’s imagine some scenarios. If you go 3 SP/6 RP, I think you’re going to finish last in W and K. If you’re going to finish last in W and K, why draft any SP at all? Draft 9 RP. The beautiful thing about this is that if you have 9 RP, they don’t even all have to be closers. If you can get 4 closers, the other 5 can be guys like Kenley Jansen, Sean Dolittle, Jake McGee, Sean Marshall, setup guys who have excellent ERA, WHIP, and K/BB and should only cost you $1 each, because no one else will want them. If you can get 4 closers pretty cheaply, combine them with 5 setup guys who make the minimum, you can finish with 42 points pitching for not much money. Throw all that extra money at trying to get across the board good hitting and if things go well, you can win your league. Remember that the price guide is telling you what it thinks people are worth, not necessarily what you should be trying to pay. You should only pay $1 more than others are willing to pay. This will probably get your league to change the rules next year ;)

        2. Start 1 RP more than the average team does. If they start 3 RP, you start 4. Maybe try to make your SP slightly better than you normally would. You’ll probably finish slightly below average in wins and strikeouts, but that is balanced by you finishing above average in the other 4 categories.

      • Maybe I’m analyzing this wrong, but I took Steamer projections and looked at the top 81 pitchers. Ranked by whip I found that an equal number were SP’s vs RP’s. Ranked by K/BB there were actually a few more SP’s. So it seems like punting K’s and W’s only guarentees you S’s and ERA which means you could just end up splitting pitching cats every week. So what am I missing now?

        Also, these questions are all about the same league – H2H, each category, daily lineup changes, 7 waiver pickups/wk. The only playing time restriction is that your pitchers must pitch a minimum of 10 inning/week (no season max or mins). I’ve been asking roto style questions since this is NOT a points league and I was told to rank like roto. What would you think of drafting 1 ace, 3 #2, 5 RP’s, and streaming SP’s with great matchups (and 2 start weeks) all season? Might give me shot at W’s. I have 3 bench slots – in this strategy do I roll the dice and use those 3 for streaming SP’s (hoping that no one has a major injury on offense)? I guess this might come to haunt me in the playoffs where my opponent will surely try to block my streams.

      • I’ve just realized the possibility that your league might have an innings minimum. If it’s set to 1000, like ESPN does, then you might have to do 3 SP and 6 RP just to reach it by the end of the season.

      • As to your other league, I don’t have much experience with a H2H daily start league, because I don’t have the time to do one myself. With these types of leagues, it depends on what restrictions there are. If there’s a season-long maximum for IP or PA, you want to be below it until the final day of the season. If there’s a weekly maximum, you want to be below it until the final day of the week. Usually if you go over it, any stats you accrue for that day still count. If there is a maximum of GS from SP, same thing. If there’s 162 GS allowed from each hitting position, well that’s a tough one. If you can’t get the kind of players who rarely get days off (and I don’t think you could get a whole team of those guys), then I’m not really sure how you do it.

      • Ok, so h2h categories league. I can’t really corroborate your numbers on WHIP and K/BB. Anyone who has a negative dollar value in that category is below average, will hurt you in that category, and therefore is someone you’d like to avoid on your roster. In that regard, I show 38 pitchers that are above average in WHIP (among top 250 pitchers): 23 of them are RP and 15 are SP. And K/BB is an even split with 18 RP and 18 SP. I’ll grant that this is still not the slam dunk I imagine it would be. But remember that we’ve only got 18 SP above average in K/BB, so the majority of SP on people’s rosters are below average, which means they cancel out the benefits of the above average guys. While some of the RP who are above average in K/BB are non-closers who you can get for league minimum.

        Look at it this way: A team who is average in a stat will have their dollar values in that stat add up to 0. $ERA is dollars of ERA value above average, $0 of $ERA means being average in ERA. If you have a 9 man pitching roster of Betancourt, Nathan, Street, Jansen, Benoit, Uehara, Marshall, Doolittle, Mujica, your total pitching values are: $-32 Wins, $17 Saves, $16 ERA, $19 WHIP, $-30 Strikeouts, $19 K/BB. Being $16 above average in each of those stats, I feel safe projecting you to win those four categories most weeks (assuming those players perform to projection, of course). And remember because most of them are $1 pickups, you’ve got more money to spend on hitting than the average team, which hopefully is enough to pickup the win in 4 hitting categories.

        But it was just a theoretical way to gain an advantage over your opponent. If you don’t want to do that, you can just do the same SP/RP breakdown that others are going to do, with the expectation that you are going to be better at identifying talent than they are.

        The problem with streaming SP is that they tend to not be as good, so they are probably going to have worse rate stats. If you win Wins but lose WHIP because your SP, you haven’t really gained anything.

        I’m doing some thinking about winning categories in a h2h league. Hold for a later post.

      • Thank you so much for your time and energy on this topic. I look forward to the follow up. This could be my most interesting draft ever.

      • Alright, so I ran the numbers for 2012 using your league’s settings. And then I calculated the correlation between scoring categories for starter quality talent.
        * A big positive number means a player who is good at the one stat is also good at the other stat. For instance the correlation between Wins and Strikeouts is 0.89. Pitchers who are good at wins are very likely to be good at strikeouts. Pitchers bad at wins (RP) are bad at strikeouts. Obvious stuff.
        * A number near zero means that there’s almost no relationship between the two stats. The correlation between SB and AVG is 0.02. That means being good at one stat doesn’t tell anything about being good at the other stat. There are players who have a lot of SB who have a great AVG (Mike Trout), there are players who have a lot of SB who have a crappy AVG (Shane Victorino), there are players who don’t get any SB who have a great AVG (Adrian Beltre), and there are players who don’t have SB and have a crappy AVG (Ike Davis).
        * A big negative number means that being good at one stat means you’re most likely bad at the other stat. The correlation between SB and RBI is -0.45. Most people who are good at SB are bad at RBI and vice versa (your Rajai Davis, Michael Bourn, Ben Revere types).

        The idea is that it tells you the stats to focus on, because being good at those stats mean you’re probably good at other stats.

        On offense, I think I might ignore SB: RBI and HR correlate very well, and there’s a pretty good correlation with both of them as well. While errors and average don’t really correlate with anything, so that means you can succeed without hurting other stats. Whereas SB has a large negative correlation with RBI and HR. This means it’s hard to be good at both SB and RBI (or SB and HR) at the same time.

        On pitching, Ws and Ks correlate very well with each other, but Ws has a negative correlation with everything else. This means that pitchers who are good at wins are most likely to be kind of bad at ERA and WHIP. This corroborates my strategy of focusing on RP, where you get Saves, ERA, WHIP, and K/BB. Notice also the very good correlation between ERA and WHIP.

        So if you ignored SB, W, and Ks, focusing your money on the other stats, you might have a better chance at winning 9 categories a week. But it’s only a theoretical way to proceed.

      • Wow! Thanks again for all the hard work. You have an extremely interesting idea – punting W’s, K’s and SB. I’m going to try it with a mock draft or two and see what I come up with. I really think it might work to get me into the playoffs. Here’s my question though…good teams get in the playoffs. The best probably have at least 2 of the top 10 pitchers. I wonder if they can easily win 4 out of 6 pitching cats by just playing their 2 aces. Any 2 of Lee, Strasburg, Kershaw, Greinke, Hamels, Verlander or Bumgardner are probably going to beat me in everything but ERA and Saves. A 1 out of 3 chance to have a 2 start week from even one ace is going to give more teams a chance to beat me. Do you think it’s worth drafting one ace relatively early to 1)keep him off the board and 2) force my opponent to have 3 Aces? Also, should I go hard after Kimbrel as my anchor? He’s projected to be so much better than all the other RP’s I think he might be worth it.

  17. There doesn’t seem to be any changes when putting in dollar values for keepers. In the past those players would be frozen and an inflation dollar column would appear. Right nothing changes.

  18. The H2H won’t let me take out keepers. When I choose that box, a line appears under the player that says, “kept at” with an empty box following it. When I enter a dollar value and click “save changes” it reloads but still has the player listed as if nothing happened.
    Thanks!

  19. Ok, I’m really stuck on the iterations piece (as described here by Mays: http://web.archive.org/web/20120419083123/http://www.lastplayerpicked.com/how-the-price-guide-works-part-iv-iterations/).

    If I have stats for 300 pitchers, and I want to find the top 110 (11 team league, 10 pitchers per team), WHY can’t I run the standards for all 300 pitchers, peel off the top 110 (re-run the z-scores so they’re relative to the 110 and not the broader pool), and then just call it a day?

    The price guide regularly shows 7 iterations to arrive at the top pitchers… and I can’t for the life of me understand what that value that process involves compared to the one I outlined above.

    Can one of the big brains on the board to step-in here (Dr. Pepper, I’m looking at YOU).

    Happy St. Patrick’s Day, all…

    Kevin

    • The issue is that the valuation of categories among the entire pitcher population isn’t necessarily the same as the value of categories among the draftable talent. And it’s the value among the draftable talent that we’re really concerned with.
      For instance, I tried it with a spreadsheet I had made up that uses the LPP method, and your method was overrating closers. I think this makes sense because my pitcher list only has 38 pitchers projected to have saves out of 499. So when you calculate the z-score of saves among all 499 guys, the average and standard deviation are both lower. Therefore when you sort based on those z-scores, closers move higher up the list than they should be. With the iterations, instead of figuring out how valuable those 38 guys are among all 499 guys, we eventually figure out how valuable the 30 RPs who are worth getting drafted are among the 110 pitchers who will get drafted, which is what we care about. The same is true of every category, but the effect is less noticeable in categories that have a smaller standard deviation.

      • Dr Pepper – putting it in the context of Saves (relative to a 499 pitcher pool) really helps it make sense. I’m stuck on implementing the algorithm. If I start with 353 pitchers in my candidate pool, can you show how to step through the iterations as outlined by Mays in the archive: “The Price Guide’s solution is to perform the valuations iteratively. Each time it processes, it feeds the top players from the previous iteration into the current one. It keeps going through that process until the results from a previous run are identical to the current run. At that point it has found the optimal player pool.”

        Do I:

        1. Take the first 110 pitchers and calculate their z-scores
        2. Take the next 110 pitchers, calculate their z-scores; and then from this second run replace the bottom 30 from the first run with the top 30 from the second run.
        3. Take the next 110 pitchers, calculate their z-scores and then from this third run, replace the bottom 30 from the list with the top 30 from the third run.
        4. Take the last 23 pitchers, calculate their z-scores and then from this fourth run, replace the bottom 20 from the list with the top 20 from the fourth run

        I’d think after this, I’ve run through every player, and have the “best” of each group. I would then take the players that were dropped in each pass and run through them in 110 player blocks, again taking the top 30 (is there a better number?) until there is no change?

        Is there a more clear way to articulate this so I can implement it programmatically? Any assistance would be greatly appreciated.

        For what it’s worth, I threw a proof-of-concept (not production code) together at http://sablelab.azurewebsites.net/pricing – that shows the output of my current process. You can see how it’s currently valuing batters and pitchers…

        Thank YOU for your clarity and assistance.

        Regards,

        Kevin

      • I do all my calculations in a spreadsheet using functions, but I’ll explain what I do and hopefully that illuminates the process.

        I’ve put a spreadsheet online at https://docs.google.com/spreadsheet/ccc?key=0AtKaVXpzf0CgdHJBV2dXdHg3b3RJcXZTbklkZUQzOXc that I made in order to see if there are any free agents worth adding to my team before the season started. It’s not full-featured because I don’t care about dollar values anymore, but it’s probably instructive for figuring out projected value (converting that to projected dollars is trivial).

        I’m going to go over some things that you probably already understand, just be thorough.

        My league is an 11 team league with 8 pitchers and 15 hitters. So my starter-quality talent is the top 88 (11*8) pitchers and 165 (11*15) hitters. You’ll notice that for each counting stat I calculate the average and standard deviation among the starter quality talent. For rate stats, I calculate the average in that rate stat among the top quality talent (the top 88 pitchers are projected for 5,388 earned runs over 13070 innings, so average ERA among them is 3.71 [9*5388/13070]). Their xERA (ERA converted to a counting stat), then, is (3.71-their ERA)*their IP. Stephen Strasburg is projected for 2.88 ERA, so he is .83 better than average per inning (for ERA and WHIP, a lower number is better. For AVG and OBP for hitters, a higher number is better, so the calculation is slightly different), .83 per inning over 177 projected innings gives an xERA of 147.

        Then we calculate the z-score for each counting stat.
        Strasburg is projected for 14 wins. The top 88 pitchers average 10.6 wins, and the standard deviation of wins among them is 4. So Strasburg’s z-score for wins is (14-10.6)/4. This works the same for every other player, not just the top 88. And this works the same for every other scoring category. Then we take the sum of all scoring categories (column Z of the pitchers tab).

        Now we need to calculate the positional adjustment so that the top 88 pitchers have an appropriate number of SP and RP (and the top 165 batters have appropriate numbers of C, 1B, 2B, SS, 3B, and OF). The short answer is that you need to make it so that the last starter at each position is worth 0 adjusted points and everyone else at that same position gets bumped up accordingly. In other words, the adjusted value of a SP is his z-scores sum (column Z) minus the value of the last drafted SP. And the adjusted value of an RP is his z-score sum minus the value of the last drafted RP. And the same with hitters. I do this manually because my league’s settings are liberal enough that I only need to make adjustments to C and RP. I haven’t given a lot of though to how it would work algorithmicly

        Now here’s the key to your question about iterations. If players are already listed so that they are sorted by adjusted total (adjusted total decreases for each time you move down the list), then you’re finished. If they’re not (and it would be a miracle if they were) then you need to sort by adjusted total and then go back to the top and do it again. Because if the identity of the top 88 pitchers (or the top 165 hitters) changed, that means the average and standard deviation changed, which means the z-scores changed, which means the positional adjustments may have changed. You keep doing this until the sort stabilizes.

        Finally you can figure out the marginal $ to marginal point exchange rate and figure out dollar values.

  20. Dr. P: I’m in a league which uses: W, ERA, WHIP, K, K/BB, S. 10 teams, 9 Pitchers. The Price guide SEEMS to way over rate closers (using Steamer). Papelbon = F. Hernandez at $25. Your IP effect described above would seem to contradict this valuation. Yeah Papelbon has slightly better rate stats, but Felix is is going to pitch 121 more innings. Is LPP wring or am I missing something?

    • I thought for sure that I responded to this already, but I don’t see it here. Testing out some settings, it seems to me that your league doesn’t have any SP/RP restrictions?

      I’m in a league that’s the same way, and this will sometimes mess the Price Guide up. The reason is that it is calculating optimal pitcher breakdowns far different from what people in your league are actually going to do. That’s the funny thing about roto, your strategy is sometimes dependent on your opponents strategy. The price guide thinks with your settings (4 categories that favor RP [relievers always do better on rate stats, plus saves]), each team should have 3 SP and 6 RP, and the dollar values are based on that assumption.

      But it’s more likely that others in your league do a traditional 6 SP and 3 RP. If that’s what everyone else in your league is going to do, that’s how you should set the guide, with the understanding that RP are probably worth a few dollars more than that, so you shouldn’t be afraid of going over with them.

      • Uh…yeah…I was trying to figure out how this blog reply button worked and I didn’t know how to get my post associated with the right thread. If you could look at my follow up questions in my post related to the other thread, I would appreciate it. Thanks!

  21. The H2H price guide using Steamer still has Hanrahan at Pittsburgh and his numbers are different from the roto projections. Are those the latest projections? Will this be updated? Thanks.

  22. This method assumes that the predictions for each stat are equally accurate. I doubt this is true. Steals, for example, are skewed, leading to large z-scores for a few players who steal more than about 30 bases each year (and negative z-scores for the majority of batters who steal 5 or fewer), and, more importantly, I suspect that steals predictions are less accurate than, say, hits (or OBP or OPS if you use those stats). In pitching, it’s probably even worse; I’d guess that there is far more variability in wins predictions than there is in K predictions. Saves and holds are even harder to predict. It seems like weighting the component z-scores for whatever stats your league uses by year-to-year correlations for each variable would address this source of error. So, for instance, if the correlation (r) between, say, actual and predicted hits for 2012 was .9, the hits z-score would be multiplied by .9. Has anyone ever done that? Does it change the rank order or dollar values very much? Any thoughts on this would be appreciated.

  23. sadly it looks like Mays has largely abandoned the site. it’s too bad as I’ve been an LPP user for a bunch of years, have donated a couple times. I hate to do this, but for anyone looking for an LPP like valuation system, check out rotochamp. you have to pay, but it’s much like LPP used to be like in 2009/10 – lots of different projections, a nice composite projection with good projected playing time methodology, and a working inflation calculator. I have no affiliation with them, and it will cost you $20 or something, but I think it’s well worth it.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s