Hey All... love this time of year when more time can be devoted to theory articles and methodology before the projections get fine tuned and draft strategy begins. I'm making a serious effort to understand the ideas behind the numbers this year... altho I'll probably still just end up taking your word for it.
If you could please explain which method is used to come up with the Mastersball values? I thought it used to be CER but I'm not sure anymore. Also, if you wouldn't mind listing most popular valuation methods (CER? SGP? etc.) utilized in the fantasy baseball industry and what they mean / differences between them. This would be extremely helpful.
I am also curious about projection systems but find valuation a bit more interesting since nobody can really see the future making projections kinda difficult...haha, just realized if we all believed that then everything would mean nothing
FWIW -- we're well beyond the fine tuning stage, our projections have been out since Nov 15. We've also been drafting since then. Anyway, I assume CER is my old category efficiency ratings which really isn't valuation methodology but a tweak to the method I use (which, FWIW, I no longer employ).
There are three primary valuation processes. Only one has a name universally recognized and that SGP (Standings Gained Points). Another uses standard deviations and the one I use distributes value based on the percentage of stats contributed to each category. Some call it PVM but that's not globally recognized.
SGP awards value based on how many standings points the player is projected to provide. Theoretically speaking, the system is very flawed but from a practical sense, it generates a usable ranking list.
StDev uses a baseline in each category, usually the mean or replacement level. Each player's stat projection is compared to the baseline in terms of how many StDev above or below the projection is. Value is then awarded on the percentage of total StDev each player has. The math is elegant but no one has been able to convince me the difference in StDev between players equates to the difference in how they affect a fantasy team.
My method measures the percentage each player contributes to each statistical category. Allowances are made for replacement etc.
I'm not a fan of the word value in general. To me it's better thought of as potential. You can't assign a static number to a projections that represents potential. As such, i don't get hung up on dollar value precision. Whether I find a particular system flawed or not is moot. The absolute is not important. Relative is what matters. This player is better than that player. Player A is worth $22 while Player B is worth $19 doesn't matter. It's contextual, but chances are Player A can help you more than Player B. To me that's where the focus should lie.
All of our methods are detailed in the Platinum content.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord
I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord
Always remember, never forget, never say always or never. - The Forum Funklord
You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord
FWIW I've done a rough calculation for the CER based on last season's NFBC Auction Championship leagues (nine leagues each with 15 teams).
I'm not 100% sure I did it right, and I've only done the hitters thus far, but I came out with the following category multipliers:
HR: 1.00
R: 1.13
RBI: 1.06
SB: 0.83
AVG: 0.98
For batting average (I used "extra hits" above replacement batting average) it cost essentially the same amount of money per point all the way from top to bottom with very little deviation. Essentially anywhere but first place (which was significantly higher) had the same cost per point, so I arbitrarily picked 3rd place to target. Maybe I should have aimed lower to account for variation. For Runs it was very clear that the best spot to be is right near the top either in first or second place. RBI, the best spot was second. HR was third, and steals was kind of a range of 2nd to 7th (I picked 5th). I may do the pitcher CER as well. If I do I'll post the results. The main takeaway here is that basically, you should take some money from stolen bases and spread it around, which is hardly groundbreaking.
@Caught Looking. Don’t discount the importance of projections. I think of player valuations as a vehicle to bring meaning to projections. Seeing the future with 100% accuracy is not really the goal of good projections. Our goal should be to develop player projections that present the most likely outcome for a player in the upcoming season and come close to reality, as often as possible.
An example of this is from last year with Jose Abreu. ESPN projected Abreu to have a .254/.412/.495 slash line. Instead, his 2014 results were .317/.383/.581. But I would contest that the projections were not wrong even though they proved to be incorrect because the projection logic was not flawed. These projections were based on the information available and for the Cuban defector there simply wasn’t enough information.
Once you are comfortable with your player projections, you can move on to establishing valuations and rankings.
@Todd Zola. I like what you said that value as a single number doesn’t reflect the overall profile of a players projections. This is especially true in roto leagues where stat balance is key. However, I don’t agree that dollar values don’t need to be precise. Establishing dollar values is really just an exercise in arithmetic and is really just a standard scale so our SGP or other method results have a meaning that’s easier to read and understand. And rankings are important for pick ‘em and auction leagues, but auction leagues need scale to see the relative gap between player value.
And you also bring up another very important point. You say that while value differences aren’t important, instead what matters is knowing that “Player A can help you more than Player B”. I completely agree, but, unfortunately, none of the referenced valuation methods can account for player value specific to a single fantasy team. Also, evaluating player value for a specific team needs to account for the talent remaining in a draft and the roster configuration of the team in question. None of the valuation methods can do this either.
draftadvisor_adam wrote:
However, I don’t agree that dollar values don’t need to be precise.
draftadvisor_adam wrote: but, unfortunately, none of the referenced valuation methods can account for player value specific to a single fantasy team. Also, evaluating player value for a specific team needs to account for the talent remaining in a draft and the roster configuration of the team in question. None of the valuation methods can do this either.
The latter is why I contend precision is a fool's errand.
My job is to build a roster with as much intrinsic potential as possible. To do that I need to balance the market price of the player with what I'm willing to pay.
The market price is driven by all these published values in a vacuum -- then the league's individual nuances tweak those to fit to that specific league.
All I care about is gauging what I feel the market will pay for the player and deciding if paying that amount adds ample intrinsic value to my squad or hopefully if I can get the player for less.
I don't care if my CVRC says a player is worth $14 or $16 -- I care whether the cost it takes me to buy the player adds commensurate potential to my team.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord
I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord
Always remember, never forget, never say always or never. - The Forum Funklord
You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord
First, Thanks everyone for your responses... I really appreciate the feedback and information.
Todd thank you for pointing out the different valuation methods... I think I was still confused in my head between projections and valuation methods. I think I understand what you're saying about finding value relative to the player pool and translating that to practical draft/auction success rather than a be all end all perfect number for price.
I'm still gonna leave constructing projections for now and probably just circle back later but I have a couple of follow up questions.
One... why are values based on the 69/31 or 67/33 splits when hitters typically make up 60.8% (14/23) of the starting roster? Is this based on the notion that there are many more valuable pitchers that are part of the player pool and are undrafted as opposed to hitters? I know that reliability is usually debated but after the last couple years I have a hard time using that after seeing so many poor performances on both the hitting and pitching side among the mid and upper tiers.
Second... in using PVM... if you have all the players projected stats... how do you come up with those who are used to comprise the player pool in order to get the totals of each stat? Since those totals are used to calculate the percent that each player contributes which then gives you his value... how can you come up with the "most valuable" players to get to those totals? haha, almost seems like a catch 22. And are undrafted stats for FAs figured into those totals at all?
Already looking forward to more of this discussion. Thanks again
Caught Looking wrote:One... why are values based on the 69/31 or 67/33 splits when hitters typically make up 60.8% (14/23) of the starting roster? Is this based on the notion that there are many more valuable pitchers that are part of the player pool and are undrafted as opposed to hitters? I know that reliability is usually debated but after the last couple years I have a hard time using that after seeing so many poor performances on both the hitting and pitching side among the mid and upper tiers.
This is convention, not anything bases in theory. In other words, it's empirically based on what most often occurs.
The reason, therefore, is moot -- it is what it is.
That said, technically, the split should be 50/50 but that's obviously not what occurs. I'm convinced the initial values were done incorrectly using 14/9 and those became accepted, then strategically it was determined that more should be spent on hitting and things settled at 69/31 for 5x5 leagues.
Second... in using PVM... if you have all the players projected stats... how do you come up with those who are used to comprise the player pool in order to get the totals of each stat? Since those totals are used to calculate the percent that each player contributes which then gives you his value... how can you come up with the "most valuable" players to get to those totals? haha, almost seems like a catch 22. And are undrafted stats for FAs figured into those totals at all?
By right it's an iterative process but that's too intense to program so I just use the top "X" in each category where X is the number of draft-worthy player then total up the category dollars then do a normalization to make it fit mathematically since just totaling leaves the pool off a little in terms of dollars and number of players so the final normalization step is really just scaling everything to make the size and dollars correct.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord
I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord
Always remember, never forget, never say always or never. - The Forum Funklord
You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord
I was just playing around with the idea of PVM in a small example but I realized I couldnt come up with a player's value based on the final category totals from my league in 2014 because there's obviously way more stats in the final totals than what is drafted on auction day.
Having said that... could you use those totals and just remove some portion of each category stats based on non-drafted players and then re-compute the player's values based on the remaining totals? If so, how does one figure out the prior season (or coming season) replacement player stats in each category?
Or is that thinking flawed and its more accurate/easier to do it based on what you just said basing it on the next seasons projections and just working on the projected stats for the best players?
I've messed around with all this and more and at the end of the day, all the different systems force the same number of players to total the same number of dollars -- the difference just isn't worth worrying about. The game theory application is so much more important to winning than a static dollar value.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord
I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord
Always remember, never forget, never say always or never. - The Forum Funklord
You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord
Ok... well if you come up with the player pool after using the top whichever totals for each category... arent you then using some players who dont make it into the positive player values?
i'm sure you know what i mean but for example Player X hits .200 with 30 steals and is in your top 196 (14 tms - 14 hitters) for stolen bases but after you get your projected category totals Player X has a negative value or some value that puts him below the draftable threshold at his position.
Is this irrelevant bc this player will be part of the Replacement Level player pool?
I agree with you on the big picture of game theory being most important with values being a guideline to work from. Having said that... do you differentiate your Value Splits from your planned Spending Splits?
Last thing... do any of the projection systems have a consistently small stadard dev and/or success rate? If you compared your projection system to any of the known ones which would you say its similar to?
Thanks again so much Todd and hopefully I'll stay active on the boards this year instead of turning into a pumpkin April 1st... altho Pumpking ale is magnificent. Yes, I fruit the beer... sometimes.
The manner I do the numbers is admittedly a concession to facilitate automated calculations but I can make an argument that different draft strategies may dictate players I rank outside of the draft-worthy pool are on some rosters.
I ran values the iterative way and this was and they were nearly identical -- not exact but certainly close enough to use.
As for testing and comparing projection systems I don't believe in it so I don't do it.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord
I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord
Always remember, never forget, never say always or never. - The Forum Funklord
You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord
Todd can you explain how the iterative process works to arrive at the desired player pool?
You earlier pointed out the other two value systems (STD and SGP) ... what is it about PVM that you believe is a better way of assigning value to players? I was reading up on STD for Avg or Above Replacement over the weekend and they seem pretty logical. I dont know what the pitfall of using the Avg player vs Replacement would be though
How do you adjust for replacement level production in PVM?
I have a couple of game strategy stuff for auctions which i'm gonna start a new thread for...thanks already for your comments.
Todd can you explain how the iterative process works to arrive at the desired player pool?
You earlier pointed out the other two value systems (STD and SGP) ... what is it about PVM that you believe is a better way of assigning value to players? I was reading up on STD for Avg or Above Replacement over the weekend and they seem pretty logical. I dont know what the pitfall of using the Avg player vs Replacement would be though
How do you adjust for replacement level production in PVM?
I have a couple of game strategy stuff for auctions which i'm gonna start a new thread for...thanks already for your comments.
With the FSTA meeting and traveling I'm about 5 days behind schedule. The best suggestion I can make is to look through the archives in Platinum -- it's all answered in there in various essays.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord
I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord
Always remember, never forget, never say always or never. - The Forum Funklord
You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord
Re: Composite Stats
with the PVM method...and I guess with all methods... is there anything that can be done to account for all the non-drafted stats that will be contributed? Naturally any method based on projections will have this issue if you're ranking the drafted player pool against one another such as with StdDev.
Maybe that is the biggest asset in using prior year standings as is done using SGP since it is attempting to measure how a player affects the composite standings.
Re: Replacement vs Avg
could you explain why most sites use value over replacement versus avg player? the idea of avg player seems a little wonky in general but I cant concretely dismiss it.
Gonna be honest -- you're nit picking things that aren't relevant. The number after the dollar sign is just a number -- it's not like PV=nRT -- you need to back off treating it as anything more than a small piece of the overall puzzle. It's not absolute -- the best use is comparing players on a relative basis.
if you're doing a draft based on stats that are in the book -- a retro draft -- then we'll get more specific.
But there's just so many factors that make any valuation system flawed it's not worth nit-picking over the equivalent of pennies.
It's so much better to spend time researching players and finding nuggets about each that are actionable and not trying to refine a dollar value to two decimal places.
Stepping outside the point I just made -- SGP has so many serious flaws that the notion of "perhaps the best thing about SGP" doesn't register with me.
I don't know that most sites use replacement over average -- and, I can't speak for them. That said, sticking to the absolute versus relative philosophy, either baseline ranks players relative to each other (as does SGP) so at the end of the day, they all accomplish what you want -- relative ranks.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord
I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord
Always remember, never forget, never say always or never. - The Forum Funklord
You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord
putting relative player ranks aside... how are you approaching team assembly this year in terms of philosophy for mixed auctions?
seeing positions that are key to pay full price such as catcher? how much of an overpay are the top handful of players worth? (trout, Kershaw, etc) is it good to avoid spending on stars in the OF in particular bc of the depth of alternatives?
in general im looking at paying for catchers, one elite RP and wait for bargains on productive OF. otherwise flush out the infield and SP with balance and production.
Perry wrote up a 15-team auction he recently did and i believe Lawr, Perry and Pasko just did a mock-auction that they may write up.
I'll be doing a couple of 15-team mixers for the NFBC.
I'll likely look to spend a little early on foundation pieces while gauging the market on starting pitching. Depending where the prices of starters fall, I'll decide how to approach the staff, fully ready to wait and build it using several in the $12-$18 range.
There's no trading in the NFBC so I'll look for 2.5 closers, probably with two in the second tier.
Then it's just going with the flow.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord
I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord
Always remember, never forget, never say always or never. - The Forum Funklord
You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord