Automated Value Calculator values seem low for closers

[alleyoops]

I'm new to using the AVC (formerly PAUL), so maybe I just have something set up wrong. I don't see a lot of explanation of the Pool Size Factor, so maybe that's it. I ran it for a mixed 5x5 14 team league, changing the Pool Factor to .72 to get 126 for a pool size. The numbers look "real" to me, with the exception that across the board, all closers appear to be $5 or more lower in value than I expected (and lower than at other sites).

I have all categories weighted at 1. Is there something in the calculator that devalues SV, relative to what I see elsewhere?

Note: The tool says to post in the Automated Value Calculator thread, but I couldn't find one.

[cwk1963]

Maybe it has to do with the hitting/pitching allocation? How do the SP values look?

[Todd Zola]

I'm new to using the AVC (formerly PAUL), so maybe I just have something set up wrong. I don't see a lot of explanation of the Pool Size Factor, so maybe that's it. I ran it for a mixed 5x5 14 team league, changing the Pool Factor to .72 to get 126 for a pool size. The numbers look "real" to me, with the exception that across the board, all closers appear to be $5 or more lower in value than I expected (and lower than at other sites).

I have all categories weighted at 1. Is there something in the calculator that devalues SV, relative to what I see elsewhere?

Note: The tool says to post in the Automated Value Calculator thread, but I couldn't find one.

There is a difference between what players are worth and what they usually go for in auctions. Assuming the settings are right, chances are your league is willing to pay more for saves than raw value indicates. Another possibility is we may be more conservative with the saves projection than other sites for the closers you are looking at.

In general, the shallower the league, the less valuable saves and hence closers are. Check out the pitching value grid and note hoe the value for closers goes from 4x4 only to 5x5 only to 15mix to 12 mix.

All that said, one of the nice things about the CVRC is you can alter the weights to get values your league is likely to use -- as well as altering the hitting/pitching split.

[Todd Zola]

For comparison purposes, I play in a 14-team mixed keeper league. Here are some salaries from last season -- inflation runs 30%.

Rivera, Nathan -- 25
Ryan 22
Jenks 19
F Cordero 17
Lidge, K-Rod 18
Putz, Saito 16
Street 14
Soria, Valverde 13
Papelbon 8 - kept as a keeper when he was drafted as a SP a few years ago

for 2009, we have Papelbon at 20, Mo and Nathan at 18, Soria 17, then 12 guys between 10 and 14 -- pretty representative of what happens in this league, when inflation is factored in.

[alleyoops]

It isn't the pitching/hitting split, as I set to 31% in the AVC, and used that in Rotolab (with HQ projections).

It isn't the projection numbers themselves, per se. For example, Mariano Rivera's projections at the other site are worse than here, across the board, yet his calculated value is $23 there, but $16 here. They project more W, same SV, more K, lower ERA, lower WHIP.

The values that I mention in the first post are not expected league bid values, but calculated values in Rotolab. Rotolab does have a number of things that you can change to influence the calculated values, but even tweaking them somewhat extremely, I can't get Rivera's value to go down much at all.

TG Fantasy Sports has a calculator that produces $25 for Rivera, with projections worse than here in W and SV, better in K, ERA, and WHIP.

This tells me that there is something very different in how the values for Rivera are calculated here, versus these other two sites. Since it appears that it's consistent that all closers are lower here, I figure it has something to do with SV, and how they're processed with respect to the value calculations.

I guess another possibility would be the projections on the other pitchers in the pool being vastly different (thus making Rivera more or less valuable relative to the total pool). I doubt that is the case, though.

Does anyone else here use Rotolab, and are you seeing consistently higher values for closers in 5x5 there, versus here?

[Todd Zola]

Much of this will be explained in a couple of valuation methodology essays which will be posted by the weekend.

If the alternate valuation methods are not what we use, that alone will explain the discrepancies.

If the method is similar to ours, the difference might be where replacement is set with respect to the pool, affecting the number of useful saves that are valued, in percentage to the overall pool.

Big picture, the important thing is to set the category weights such that the values are more representative of how your league values saves. This is one of the pratfall of valuation in general, conventional methods can only go so far.

Let's do some math...

There are 30 closers about 5$ too low, that is $150 shifted to the save category. There are 126 pitchers in this pool, so that leaves about 96 non-closers. This $150 is taken from this pool of 96 non-closers, meaning each loses a buck or two of value if it is applied linearly, but much like inflation, it is not applied linearly, more value is lost from the top-end, almost no value at the low and something in between in the middle.

This is a crude way of doing it, but an approximation could be the to 32 non-closers lose $100 while the middle 32 lose $50, with those at the very top losing a bit more than those in the lower end of the top-32.

100/32 is about 3, so I would expect the top guys to lose a little more than 3 bucks.

Taking the CVRC and changing the saves weight to 2.1 gives the following values to select pitchers to use as an example, their regular value in ().

Johan 27 (32)
CC 26 (31)
Peavy 22 (26)
Kazmir 12 (14)
Volquez 11 (13)
Lowe 10 (13)

Papelbon 25 (20)
Nathan 22 (18)
Ryan 17 (13)
Fuentes 17 (12)

So the closers are now +4/5 and the top end starters lost -4/5 and the middle starters a couple bucks.

At this point, I would probably use the fact that some back-end pitchers you like for $4-$6 will probably be available for $1-$3, especially in a league this shallow. I'd take that extra handful of money and put it towards the top-end starters, bring their value back to what it was before, as you likely won't buy more than 1 top-end guy.

[cwk1963]

Much of this will be explained in a couple of valuation methodology essays which will be posted by the weekend.

If the alternate valuation methods are not what we use, that alone will explain the discrepancies.

If the method is similar to ours, the difference might be where replacement is set with respect to the pool, affecting the number of useful saves that are valued, in percentage to the overall pool.

Big picture, the important thing is to set the category weights such that the values are more representative of how your league values saves. This is one of the pratfall of valuation in general, conventional methods can only go so far.

Let's do some math...

There are 30 closers about 5$ too low, that is $150 shifted to the save category. There are 126 pitchers in this pool, so that leaves about 96 non-closers. This $150 is taken from this pool of 96 non-closers, meaning each loses a buck or two of value if it is applied linearly, but much like inflation, it is not applied linearly, more value is lost from the top-end, almost no value at the low and something in between in the middle.

This is a crude way of doing it, but an approximation could be the to 32 non-closers lose $100 while the middle 32 lose $50, with those at the very top losing a bit more than those in the lower end of the top-32.

100/32 is about 3, so I would expect the top guys to lose a little more than 3 bucks.

Taking the CVRC and changing the saves weight to 2.1 gives the following values to select pitchers to use as an example, their regular value in ().

Johan 27 (32)
CC 26 (31)
Peavy 22 (26)
Kazmir 12 (14)
Volquez 11 (13)
Lowe 10 (13)

Papelbon 25 (20)
Nathan 22 (18)
Ryan 17 (13)
Fuentes 17 (12)

So the closers are now +4/5 and the top end starters lost -4/5 and the middle starters a couple bucks.

At this point, I would probably use the fact that some back-end pitchers you like for $4-$6 will probably be available for $1-$3, especially in a league this shallow. I'd take that extra handful of money and put it towards the top-end starters, bring their value back to what it was before, as you likely won't buy more than 1 top-end guy.

This is a PERFECT example of why I and, I'm sure most everyone else, is here. Todd and Gary don't just put their valuations/projections/profiles out there for the public to purchase and walk away counting their cash. They take the time to explain the how's and why's to the best of their ability when asked a question. Furthermore, Todd's a chemist by trade and I'm not. The explanations are not done in chemistry mode where I would not understand a single thing. Keep up the great job guys. That's why this site rocks.

[alleyoops]

Big picture, the important thing is to set the category weights such that the values are more representative of how your league values saves. This is one of the pratfall of valuation in general, conventional methods can only go so far.

Thanks for the detailed explanation, Todd.

I think there was a discussion last year on the point above. That is, is it desirable to go into the auction with "true" values, values adjusted based on your league's tendencies, or both? If Rivera will produce only $16 in "true" value, I really would like to know that, but I'd also like to know that my competition will likely be willing to pay $23 for him (adjusted for league tendencies, or the number commonly seen in magazines and at the majority of sites). Both things matter. This is probably worth a new topic for discussion, though, since it's not specific to the closer values.

If, in fact, closers are worth only what the AVC here calculates, it tells me that the league(s) I'm in are overvaluing them, and I might be well-advised to avoid them in the auction and spend more elsewhere. That's probably not a bad idea in general, since SVs are easier to find during the season than some other categories, but if they're truly worth less than the prevailing wisdom in my leagues, it's even more attractive. Agree?

[Todd Zola]

If, in fact, closers are worth only what the AVC here calculates, it tells me that the league(s) I'm in are overvaluing them, and I might be well-advised to avoid them in the auction and spend more elsewhere. That's probably not a bad idea in general, since SVs are easier to find during the season than some other categories, but if they're truly worth less than the prevailing wisdom in my leagues, it's even more attractive. Agree?

To be completely honest, this has been the trendy mantra for a couple of years now, especially in relatively shallow leagues.

But it really depends on your league tendencies. If EVERYONE waits, the play may be to jump in early, I blogged about this and call it the PAPELBON PLAN. The reason being that some leagues have grown very adept at stock-piling the speculative closers, decreasing the amount of in-season saves available.

I usually decide on an amount I want to pay for a closer, and get the best available closer at that cost.

In the AL and NL only NFBC auctions, I decided I would spend $20 on a closer.

In the NL, I threw out Billy Wagner at $20, heard crickets and had my closer.

In the AL, I threw out Papelbon and dropped out at $20, he went for almost $30. I didn't want to pay that much, settled for Huston Street and Joe Borowski for about $25 combined.

I'll be defending my NL title in March.

I finished 7th in the AL

[rotodog]

I have used Rotolab and PAUL in the same season in the past. I do know that Rotolab allows for MANY TWEAKS on top of the simple league type and allocation splits (70/30 etc..).....They do allow weighting, valuation method (SGP etc..) and other tweaks. I know you said you tweaked it, but I am just making sure it is tweaked equally with PAUL....

It could also be the projections. I have not looked at projections for closers, but it appears to me that this year has MORE teams with shaky closers and 2 guys behind him that could get some saves. Saves are a one trick pony and if one set of projections is spreading the saves around and one set not, that could represent the value differences too..

Here is what I would try. I would not be so concerned about TRUE VALUE or EXACT projections. I would try to tweak rotolab by reducing the weighting for saves and giving that reduction to another more stable category. i would try tweaking it so that the rotolab values are closer in line with PAUL. Make sure the method isnt SGP also. Once you get them a little closer, the ones that TRULY stick out as different between the two systems, i would check the projection differences. As a closer, most of his value comes in saves. just an 8-10 saves difference could significantly alter value between 2 systems i think.

If one closer is being proctected for 36 saves one place and 28 the other, that may be the difference too.

Once I tweaked the lab, I honestly would not worry about TRUE value or EXACT value. You would be better off if you simply looked for pockets of value within the context of other closers. Whether Rivera is worth 16 or 25 dollars is not relevant to me much. I want to see the pockets of value... Look at closers with value in the same area as other closers. Perception will make some closers go for more or less than another closer. This is where you exploit the value. If one closer is worth 22 dollars and another 21 dollars , one may very well be perceived as a lower end closer and will cost less in relation to the other closer or closers....A players value isnt exactly what some program says he is worth...Perception is value on draft day..A player is worth exactly what someone is willing to pay for him at any point in time. Exploit the perception and get a 21 dollar closer for 7 dollars less than the 22 dollar closer.....

Of course that is an extreme example, but I always need to see a players value in context of the other players and exploit the perceptions.

[rotodog]

I also Never pay for closer myself in actively traded keeper leagues. FA and dump trades will get you in the middle of the pack easily and you can spend that money on monster offense....I build an offensive lead and trade offense for less offense and middling closers....Works every year.. Just my style...But if closers are not available via FA because the league is deep or closers are not traded during the season, then you need a closer or two on draft day. If a closer is going super cheap, I will grab him...but it rarely happens Closers need to be managed ..

[cwk1963]

If, in fact, closers are worth only what the AVC here calculates, it tells me that the league(s) I'm in are overvaluing them, and I might be well-advised to avoid them in the auction and spend more elsewhere. That's probably not a bad idea in general, since SVs are easier to find during the season than some other categories, but if they're truly worth less than the prevailing wisdom in my leagues, it's even more attractive. Agree?

To be completely honest, this has been the trendy mantra for a couple of years now, especially in relatively shallow leagues.

But it really depends on your league tendencies. If EVERYONE waits, the play may be to jump in early, I blogged about this and call it the PAPELBON PLAN. The reason being that some leagues have grown very adept at stock-piling the speculative closers, decreasing the amount of in-season saves available.

I usually decide on an amount I want to pay for a closer, and get the best available closer at that cost.

This is exactly how my league has been for a number of years. Hence, I am going to be looking to implement the Papelbon/Soria/Nathan plan and try to get one of them very early.

[viper]

Although I don't recall this problem per se, I remember removing all non-starters from the list of pitchers I used for PAUL. In my only leagues, I know what closers typically go for so I set up pitching dollars to equal my allocation less what I would spend on closers. I am doing 10 team "only' leagues so I reduced the required SP count to 70 or 7 per team. It seemed to closer to norm dollar value for SPs. It also fit into my thoughts of essentially adding the dollars needed for closers to my hitting dollar while I also required any closers purchased come from that pool of money.

[alleyoops]

I have used Rotolab and PAUL in the same season in the past. I do know that Rotolab allows for MANY TWEAKS on top of the simple league type and allocation splits (70/30 etc..).....They do allow weighting, valuation method (SGP etc..) and other tweaks. I know you said you tweaked it, but I am just making sure it is tweaked equally with PAUL....

It could also be the projections. I have not looked at projections for closers, but it appears to me that this year has MORE teams with shaky closers and 2 guys behind him that could get some saves. Saves are a one trick pony and if one set of projections is spreading the saves around and one set not, that could represent the value differences too..

Here is what I would try. I would not be so concerned about TRUE VALUE or EXACT projections. I would try to tweak rotolab by reducing the weighting for saves and giving that reduction to another more stable category. i would try tweaking it so that the rotolab values are closer in line with PAUL. Make sure the method isnt SGP also. Once you get them a little closer, the ones that TRULY stick out as different between the two systems, i would check the projection differences. As a closer, most of his value comes in saves. just an 8-10 saves difference could significantly alter value between 2 systems i think.

If one closer is being proctected for 36 saves one place and 28 the other, that may be the difference too.

Once I tweaked the lab, I honestly would not worry about TRUE value or EXACT value. You would be better off if you simply looked for pockets of value within the context of other closers. Whether Rivera is worth 16 or 25 dollars is not relevant to me much. I want to see the pockets of value... Look at closers with value in the same area as other closers. Perception will make some closers go for more or less than another closer. This is where you exploit the value. If one closer is worth 22 dollars and another 21 dollars , one may very well be perceived as a lower end closer and will cost less in relation to the other closer or closers....A players value isnt exactly what some program says he is worth...Perception is value on draft day..A player is worth exactly what someone is willing to pay for him at any point in time. Exploit the perception and get a 21 dollar closer for 7 dollars less than the 22 dollar closer.....

Of course that is an extreme example, but I always need to see a players value in context of the other players and exploit the perceptions.

Thanks for responding, rotodog.

First off, it isn't the actual projection numbers that cause the differences. The Rivera example above proves that. His projected numbers are better across the board at MB, but the calculated value is much lower.

In rotolab, I initially had the settings at what I used for this league at last year's draft, which I had tweaked to get the numbers close to what I thought the pitchers would go for last year. I had the SV factor at .4, with the others between .75 and 1.25. I had lowered the SV factor last year because the SV values seemed high then.

This year, I tried changing from PVM to SGP (not changing any SGP factors) and the closers were still high (compared to here). Then I tried going back to PVM, setting the non-SV factors to 1.0, and leaving the SV at .4. Still high. Finally, by lowering the SV factor to .2, I got numbers similar to here.

This seems pretty extreme. It indicates to me that the value calculation model in Rotolab and the model here treat SV very differently, in some way. Todd has given a couple of possible explanations. It really can't be determined for sure without looking at the code in both models, and that isn't going to happen.

For now, I'm going to take Todd's suggestion, look at what pitchers went for in that league last year, and make Rotolab come as close as possible to those numbers.

But I still have a problem, even after that, in the actual auction. Do I "believe" the MB values, and take those into consideration in my bidding? If so, I probably wind up with no closers (which may not be a bad thing). If I bid to just below the Rotolab numbers, as I've tended to do in the past, I may be overpaying for closers (if the MB values are "real") or I may be getting small bargains (if the Rotolab numbers are "real").

In the end, it's all linked in with my planned strategy in the league, and the ever-changing auction dynamics, so it's just another variable in the big picture. But the analytical part of me wants to know.

[rotodog]

Ok,

Here are my insights as I had used Rotolab for 5+ years. It is a great tool, but I honestly think too many people (including myself) have used it as a 100% fool proof plan at the draft.

In this thread, viewtopic.php?f=7&t=119

I tried to explain what I mean by this. It doesnt address your exact question, but will shed some light on it in Vipers 1st post and my follow up.

I was going to use no tools this year and just wing it with paper and calculator....I still plan on going that way, but I did go ahead and spend 30 bucks for this sites tools as I feel they are worth my money. I however will not be using the draft tracker or any other software at all. It has nothing to do with money at all...I simply feel we get locked into "prices" or "value" too much on auction day. People all want to get guys undervalue and get profit. Problem is that if you go this route and do not bid 1 dollar higher than rotolab says to, you are most likely not cashing a big check in October.

In the thread I linked I mentioned an instance where a guy can show as -$2 value, but still have value for someone. What you are seeing in rotolab is a players value, using projections in the context of a whole season. It is flawed to begin with. Even if I give you 2008 exact stats and exact values for your league, if you use dollar values to do a draft after the season, a smart auction drafter can still beat you.

Now I have been to drafts when 3-4 different owners are using Rotolab. Yet, the end result every year varied with these other owners. In the 3 years I was involved with this league, I won twice and took second place. the guy that won the other wasnt a rotolab user either. One guy asked if 4 guys use rotolab, how do you get 4 very different results? We all are essentially seeing the same exact stuff at the same time? How does one win consistantly and 3 others not? My problem with rotolab is that the other owners always had a mindset similar to what you mentioned. Bidding to prices or not? They all bid to prices, or even the inflated prices. but tended not to go over because they wanted to draft to targets and win the league on APRIL 1st, not OCTOBER 1st..... I knew when to spend the extra and when not to I guess.

My advice is to use it as a guide. A tool to help value players and keep track of the draft. A toll to find pockets of value at different positions and exploit them to save your money. Dont get so caught up with the actual value. If Pujols or hanley would cost 50 bucks in your league and rotolab says he is worth 40, every player essentially has to be valued on the fly with the 50 dollar ceiling in mind and then descend from there. If not, your team will look like the Oakland As and at some point, I feel you might (not you per se') have more money than talent. I have paid 50+ for a player many times ..THE BEST PLAYERS ONLY. I overpaid early on purpose to set the tone. Rotolab said I overpaid and bought at a loss, but when players like mike Cameron and Prince Fielder go for 40-45 dollars because 2 owners have too much money left, you will be glad to own Hanley and Pujols for 50 each.... That make sense?

I know you wont try this , but maybe I would run all my league valuations etc, then go to the grids or reports area and print out your lists and leave ROTOLAB at home.... It will be liberating.

I will be going back to the same league with rotolab users this year after taking a year off. I will not be using it and I am wondering if the same guys will use it or more will use it.