a. in a h2h format, you only need to score ONE more "x" (hr, rbi, 2b, 3b, etc) in order to win the category. This makes Pierre's (and others) steals less valuable as you exceed the threshold needed to post the win. If you only need 9 steals for the week, 34 is wasted production.
b. Pitching is much more variable year-to-year than hitting- not only in cumulative totals, but also start to start. It's hard (i said hard, not impossible) to base a successful h2h team solely on pitching. you may get 15 saves and 10 wins in a week, and then post 0'fers in the next two consecutive.
c. If you DO decide to go pitching heavy in your draft, you will return MUCH higher value for your picks if you take the production hitters and then trade them for quality pitching. You would ideally want to draft hitters for the first two rounds, and then be the first person to take a pitcher in round 3.
d. If you DO decide to go pitching heavy in your draft, you must adjust your formulation to account for what I call the "feast vs. famine effect". If you have the top X pitchers and top X stealers on your team, you will require a lower number to assure victory. The other teams will not have those players their teams, and therefore, will not be as competitive in those areas. ALSO, you must account for the fact that some teams will then have Pujols, Longoria, Utley, etc. on the same roster, which elevates your required production in order to win the counting stats.
e. I do not recommend the pitching dominant strategy, as it is much easier to find high-production pitchers on the waiver wire throughout the year (think Daniel Hudson, Brandon Morrow, or a plethora of Saves). What would then happen is teams with high-caliber offenses would pick these guys up (since you wouldn't be) and your advantage in these categories is then diminished or negated.
very good points biesbol. im going to re-evaulate everything once i get more confident with my algorithm.
i like how baseball monster evaluates the stats through z-score. looks like im going to have to completely redo my calcs now
but yeah this would explain my problems with myalgorithm, this method shows a much lower effect that saves, triples, and holds have on overall rankings - bringing values back to more appropriate levels.
mr anderson... wrote:very good points biesbol. im going to re-evaulate everything once i get more confident with my algorithm.
i like how baseball monster evaluates the stats through z-score. looks like im going to have to completely redo my calcs now
but yeah this would explain my problems with myalgorithm, this method shows a much lower effect that saves, triples, and holds have on overall rankings - bringing values back to more appropriate levels.
thanks for the help (:
ill post an update when i get there
Welcome to the Cafe. Keep pluggin, your type of enthusiasm is great to see; I'm sure you'll both share and learn a lot here.
a. in a h2h format, you only need to score ONE more "x" (hr, rbi, 2b, 3b, etc) in order to win the category. This makes Pierre's (and others) steals less valuable as you exceed the threshold needed to post the win. If you only need 9 steals for the week, 34 is wasted production.
b. Pitching is much more variable year-to-year than hitting- not only in cumulative totals, but also start to start. It's hard (i said hard, not impossible) to base a successful h2h team solely on pitching. you may get 15 saves and 10 wins in a week, and then post 0'fers in the next two consecutive.
c. If you DO decide to go pitching heavy in your draft, you will return MUCH higher value for your picks if you take the production hitters and then trade them for quality pitching. You would ideally want to draft hitters for the first two rounds, and then be the first person to take a pitcher in round 3.
d. If you DO decide to go pitching heavy in your draft, you must adjust your formulation to account for what I call the "feast vs. famine effect". If you have the top X pitchers and top X stealers on your team, you will require a lower number to assure victory. The other teams will not have those players their teams, and therefore, will not be as competitive in those areas. ALSO, you must account for the fact that some teams will then have Pujols, Longoria, Utley, etc. on the same roster, which elevates your required production in order to win the counting stats.
e. I do not recommend the pitching dominant strategy, as it is much easier to find high-production pitchers on the waiver wire throughout the year (think Daniel Hudson, Brandon Morrow, or a plethora of Saves). What would then happen is teams with high-caliber offenses would pick these guys up (since you wouldn't be) and your advantage in these categories is then diminished or negated.
this is basically a formula for how to win any H2H league regardless of settings. draft power early and often. SP can be found or streamed throughout the year. saves can be found or just punted.
for 2 years in a row i have been good in one league where i went SP heavy. ended the year w a good record...but flopped in the playoffs. ill stay w big bats early and often from now on
ok guys. i got it all fixed now. converting to a z-score to analyze statistical categories compared to one another has brought the values to levels that would be expected. now i just need to determine if i want to include opportunity costs between selecting a pitcher or batters.
if i DO include opportunity cost the top 15 is dominated with pitchers (12 of top 15), but i think i will leave them separate and merge together. (opportunity cost involves taking in account that a pitcher would not be hitting and a hitter would not be pitching...perhaps a gigantic waste of time to determine looking back on it now. oh well).
here are my results from the 2010 season with 50 ab and 20 IP criteria (i just left it there for the time being, will update to reflect projected fantasy league totals; but that will have to wait for the time being since im more interested in the concept and results should only vary slightly)
5x5 scoring for 2010 1 12.68192829 Roy Halladay 2 12.07242087 Carlos Gonzalez 3 11.44200402 Adam Wainwright 4 11.34278164 Felix Hernandez 5 11.1539745 Albert Pujols 6 10.92919789 Carl Crawford 7 10.66289471 Joey Votto 8 10.46232759 Jose Bautista 9 10.13666201 Miguel Cabrera 10 9.403978713 Josh Hamilton 11 9.257757025 Ubaldo Jimenez 12 8.952493344 Jered Weaver 13 8.949167764 CC Sabathia 14 8.509499646 Roy Oswalt 15 8.371192895 Justin Verlander 16 8.360551355 Juan Pierre 17 8.301869539 David Price 18 8.284871304 Alex Rios 19 8.212493654 Hanley Ramirez 20 8.194900203 Jon Lester 21 8.150743093 Paul Konerko 22 8.064599145 Ryan Braun 23 8.032252426 Robinson Cano 24 7.950396129 David Wright 25 7.868709567 Matt Holliday 26 7.812136769 Chris Young 27 7.749303418 Cliff Lee 28 7.711743589 Tim Hudson 29 7.561329444 Troy Tulowitzki 30 7.542209218 Matt Cain 31 7.532356151 Chris Carpenter 32 7.509997156 Jayson Werth 33 7.443584695 Evan Longoria 34 7.415317438 Dan Uggla 35 7.329927551 Shin-Soo Choo 36 7.218301578 Clayton Kershaw 37 7.177311873 Hunter Pence 38 7.092220029 Vladimir Guerrero 39 7.037785074 Mat Latos 40 6.990122914 Corey Hart 41 6.93045683 Adrian Beltre 42 6.929064659 Tim Lincecum 43 6.8945477 22 Rickie Weeks 44 6.892049203 Cole Hamels 45 6.858037596 Josh Johnson 46 6.835913017 Drew Stubbs 47 6.756107575 Andrew McCutchen 48 6.715314466 Angel Pagan 49 6.695253285 Mark Teixeira 50 6.671520923 Trevor Cahill 51 6.6120488 31 Brett Gardner 52 6.602427044 B.J. Upton 53 6.466149218 Adrian Gonzalez 54 6.419292315 Shane Victorino 55 6.409313839 Brett Myers 56 6.404822039 Aubrey Huff 57 6.385658499 Alex Rodriguez 58 6.279708374 Ichiro Suzuki 59 6.234440493 Matt Kemp 60 6.224114928 Jonathan Sanchez 61 6.218259683 Ryan Howard 62 6.197591436 Rajai Davis 63 6.160343065 Adam Dunn 64 6.1316042 87 Nelson Cruz 65 6.10493738 Kelly Johnson 66 5.960885286 Delmon Young 67 5.897328593 Bobby Abreu 68 5.882821392 Clay Buchholz 69 5.867306262 Ryan Zimmerman 70 5.808160755 Nick Swisher 71 5.803186037 David Ortiz 72 5.73126878 Michael Bourn 73 5.721440688 C.J. Wilson 74 5.677022467 Vernon Wells 75 5.497818757 Michael Young 76 5.473371416 Colby Lewis 77 5.458766965 Bronson Arroyo 78 5.455569151 Francisco Liriano 79 5.435093812 John Danks 80 5.39040086 Andres Torres 81 5.381081578 Shaun Marcum 82 5.359194721 Gio Gonzalez 83 5.337135022 Heath Bell 84 5.327966775 Jose Reyes 85 5.32358691 Torii Hunter 86 5.314585201 Dan Haren 87 5.312260969 Ryan Dempster 88 5.283174485 Prince Fielder 89 5.258532905 Johan Santana 90 5.256805191 Tommy Hanson 91 5.255398926 Carl Pavano 92 5.215389376 Ted Lilly 93 5.143600948 Brandon Phillips 94 5.094440918 Brian Wilson 95 5.069592169 Hiroki Kuroda 96 5.032742578 Derek Jeter 97 5.016055827 Max Scherzer 98 4.99043132 Colby Rasmus 99 4.909117943 Scott Podsednik 100 4.902796318 Ervin Santana
interesting finds: my algorithm shows that top ranked in both formats roy halladay was 17% more valuable than the next listed player (cliff lee) in the 9x9, but only 3.8% more valuable than the next listed player (carlos gonzalez) in the 5x5 scoring format, even though 5x5 scoring showed more favoritism to pitching (7 pitchers in top 15, 3 out of top 4) than the 9x9 (5 pitchers out of top 28, top 2 overall)
next step for me is to try and streamline it so i can just select a button for the scoring categories rather than have to physically write the function in the cell and copy it down, and start inputing my personal projections. hopefully i wont have to make it a macro since i currently do not know how to write one, but i suppose i could learn if needed.
mr anderson... wrote:I am in the process of developing a fantasy baseball algorithm specifically designed to aid in draft day selection for my leagues custom format.
strategy wise, I'd punt SB and starting pitching. Just load up on power, and top of the line relievers. you could even go so far as to draft all of your bats first, then grab relievers.
I'd definitely punt SPs here but speedy leadoff hitters could contribute in R, H, SB, and 3B. I wouldn't use a high pick on one, though. Maybe secure a B.J. Upton-type in the 7th-9th, then a Rajai/Pods type in the 14th-16th, then waste 3 of your last 5 picks on lottery ticket leadoff burners like Desmond Jennings (though probably not Jennings himself because he'll probably get super-hyped come March).
Also, if you're planning to use this as a draft guide, I'd devalue triples considerably. They're few and far between and you easily could win with like 3 or 4 triples in a week. There's only a few players you can reliably draft for triples: Reyes, Craw, Granderson, Victorino, Rollins, and maybe Stephen Drew. And even a couple of these guys could easily end up with single digit triples for the year. It's just too random and unpredictable week-to-week.
Rocinante2: you know Rocinante2: its easy to dismiss the orioles as a bad team ofanrex: go on Rocinante2: i'm done Rocinante2: lmao
it's starting to shape up nicely. i had an oversight that i found this morning where i didnt account for the fact that low whips and era are actually desired, so i simply just calculated the z-score off of the inverse of those numbers (era of 4 = 1/4 = .250; era of 3 = 1/3 = .333), and the numbers are now accurately reflective.
i have projected 125 players, mostly upper tier but a few of the part timers and scrubs just to compare and have got some very interesting results in a 5x5.
the highest relief pitcher that my projections have computed is, or course, mariano rivera at 63. the other closers ive made projections for are: joakim soria (84), neftaliz feliz (98), craig kimbrel (102) [i have him mocked as the closer for now], heath bell (107), chris sale (113) [again, mocked as closer for now], so this is FAR from a complete list but more than likely there will only be 1 closer found within the top 100 for overall value. ive triple checked all my calcs and they appear correct to me.
in a 9x9, mariano rivera is ranked all the way to 101. it seems odd, yes, but when compared to top tier starters and everyday regulars the expanded categories apparently dilute the value of closers considerably.
my projected top 25 in 5x5 through today (ive projected all the big names so i dont believe im missing anyone obvious)
but the biggest surprise is that it generated adam wainwright a near 5.5% improvement over halladay in 5x5, but an 8.3% DECREASE in the expanded format based off, what i obviously feel would be, fair projections:
halladay: 34 starts, 242 IP, 230 H's, 77 ER, 37 BB, 210 K's, 19 W, 10 L wainwright: 34 starts, 238 IP, 200 H's, 70 ER, 50 BB, 204 K's, 18 W, 10 L
and albert pujols just dominates this game as evidenced by a 16.8% increase over number 2 ranked joey votto.
of course all these are based off my projections and for all you know im a middleaged housewife's pet kitten, but i still found this interesting and wanted to share. ill upload my calc incase anyone wants it but its holiday time and i dont have to work so that means i wont be working on this until monday at the earliest.
mr anderson... wrote: for instance: there were 1198 saves, 1993 holds, and 2410 wins for the entire imported player pool. my reasoning tells me that due to the disparity between these categories, a save will be twice as effective at winning its respective category versus a win due to it being twice as scarce, and subsequently that a relief pitcher with 5 wins and 10 saves will be 39.9% more effective (valuable) as a relief pitcher with 5 wins and 10 holds.
I disagree with this, respectfully. Since you also count W, K, K/9, and QS, I would think that SP weigh alot heavier on the outcome than a RP. I get that since there are less saves than they are wins, that saves could be more effective (in theory) but I think many may take this message the wrong way. I can easily see a beginner reading this and taking away that because saves are less abundant, it's better to have more closers than SPs. Also, your entire arguement and theory is based upon a certain statistical setting, so you would have to be playing under these exact same settings to follow this logic to a T. Great stuff though. "Baseball layed by the ambidexterous, understood by the poindexterous. "
mr anderson... wrote:I am in the process of developing a fantasy baseball algorithm specifically designed to aid in draft day selection for my leagues custom format.
strategy wise, I'd punt SB and starting pitching. Just load up on power, and top of the line relievers. you could even go so far as to draft all of your bats first, then grab relievers.
I'd definitely punt SPs here but speedy leadoff hitters could contribute in R, H, SB, and 3B. I wouldn't use a high pick on one, though. Maybe secure a B.J. Upton-type in the 7th-9th, then a Rajai/Pods type in the 14th-16th, then waste 3 of your last 5 picks on lottery ticket leadoff burners like Desmond Jennings (though probably not Jennings himself because he'll probably get super-hyped come March).
Also, if you're planning to use this as a draft guide, I'd devalue triples considerably. They're few and far between and you easily could win with like 3 or 4 triples in a week. There's only a few players you can reliably draft for triples: Reyes, Craw, Granderson, Victorino, Rollins, and maybe Stephen Drew. And even a couple of these guys could easily end up with single digit triples for the year. It's just too random and unpredictable week-to-week.
Hey torpedo. I live in Tampa and have had the priveledge of watching Jennings first hand the last few years during Spring training. Y'know what? I like what I see out of Dexter Fowler more.
i updated the calcs when determining standard score to remove double rounding in some categories (like batting average: i was erroneously averaging together all the averages, now it sums the total hits and calcs against total at bats & so forth)
i have also changed the way that it will calculate dynamic stats like Batting average, whip etc.
before it would take the standard score for at bats, the standard score for batting average, and then combine them together for an interpretive average. i didnt like the way i set this up so i starting looking at other resources (i have only ever taken beginning stats in college) and have set it up more appropriately.
the new calc for batting average is now based off how many more (or less) hits player a would have over the average player for that sample set.
player hits - (player AB*(sum of all hits/sum of all at bats)) = xBA
THEN it takes a standard score of the xBA figure, which is used in calculating a players overall value.
examples from 2010: who had a bigger impact for batting average: albert pujols , joe mauer, or robinson cano
pujols: 183 hits, 587 AB (0.312) mauer: 167 hits, 510 AB (0.327) cano: 200 hits, 626 AB (0.319)
right away we should see that cano should be higher than pujols because cano has a higher average with a higher number of at bats, but is cano better than mauer?
using the xBA calc: (note: 2010 totals for players over 50 abs: 40860 hits ; 155468 AB)
so we see that cano was, in fact, more valuable than mauer even though he hit .008 points lower with 116 additional at bats
the highest for 2010: josh hamilton at 49.860. the lowest for 2010: carlos pena at -32.205
whip and era are calc'd in a similar manner (ERA is how many more (or less) earned runs allowed per nine innings than the average of all the sample set ; and WHIP is how many more (or less) walks + hits allowed per inning pitched than the average of all the sample set) with subsequent standard score calculations
i have also done this for k/9, and k/bb in my expanded format xK/BB=(playerK/playerBB-(sum(K)/sum(BB))*IP xK/9=(playerK/playerIP*9)-(sum(k)/sum(IP)*9)*IP
here is an updated top 25 for all of 2010 in 5x5 scoring with the new additions/improvements:
55x5 rank 5x5 score 9x9 rank 9x9 score PLAYER TEAM Pos. 1 13.6866917 3 20.92770512 Carlos Gonzalez COL LF 2 13.60131232 1 25.46881839 Roy Halladay PHI SP 3 12.57392273 6 18.46443009 Adam Wainwright STL SP 4 12.3125607 7 18.12178694 Felix Hernandez SEA SP 5 12.15001309 5 18.68872 Albert Pujols STL 1B 6 11.84568676 4 20.59194843 Carl Crawford TB LF 7 11.74236896 8 17.82003741 Joey Votto CIN 1B 8 11.3308263 9 17.71931479 Miguel Cabrera DET 1B 9 11.0167557 12 15.93965621 Josh Hamilton TEX LF 10 10.12402112 11 16.4092605 Jose Bautista TOR RF 11 9.902292661 25 13.11079942 Ubaldo Jimenez COL SP 12 9.754996817 10 16.95794705 Jered Weaver LAA SP 13 9.7377004 17 14.11915931 Roy Oswalt HOU/PHI SP 14 9.431708325 13 15.34558301 Robinson Cano NYY 2B 15 9.189429512 51 11.28927844 CC Sabathia NYY SP 16 9.032184506 49 11.5183381 David Price TB SP 17 8.978000649 27 13.08885938 Paul Konerko CHW 1B 18 8.972419235 18 14.03923982 Ryan Braun MIL LF 19 8.913904023 15 14.52395408 Matt Holliday STL LF 20 8.827648922 68 10.32477108 Billy Wagner ATL RP 21 8.802271917 2 22.26009955 Cliff Lee SEA/TEX SP 22 8.757646029 35 12.63677635 Hanley Ramirez FLA SS 23 8.696458845 31 12.79119187 Justin Verlander DET SP 24 8.591864049 29 12.988592 Jon Lester BOS SP 25 8.5213779 45 11.7478594 Alex Rios CHW CF
this SHOULD be my last update so i can quit bothering you guys
just didnt want to screw anyone over with an inadequate program. im just going to continue working on projections and research calculating a dollar amount for auction formats. i can post if anyone is interested but otherwise this should be it.
Cafe Home
Fantasy Football
Fantasy Basketball
Fantasy Hockey
by biesbol » Wed Dec 22, 2010 10:02 am 
layed by the ambidexterous, understood by the poindexterous. "
