How to figure out each players category by category value?

[jbluestone]

I am trying to do a little research on my HTH league, where I want to look at how much of a players $ value is in each category.

my league is about 55% offensive budget with a $380 cap. So i did $380*.55 = $42 for offensive budget (rounded up). So i put in an $84 budget and set the balance to 50% value to make sure that $42 were allocated to the hitters. I then checked only the offensive category of interest... i did this for each category, and then compiled the numbers in excel... however, the separate category values do not add up to the the same value if i just run a players value with all 5 categories checked.

for example here are these 3 players

PlayerID	Lastname	Firstname	OPS$	SB$	RUNS$	RBI$	HR$	TOTAL FMV	actual FMV
813	Cabrera	Miguel	12	1	7	11	10	41	64
2542	Braun	Ryan	11	6	7	7	7	38	62
983	Encarnacion	Edwin	9	3	5	8	6	31	50

Any idea what I am doing wrong?

[Riverdome]

I don't see your error but I will plug the book available on this site: http://www.baseballhq.com/content/ho...serie-baseball

If you can handle a little Excel it's an easy to implement method and has worked VERY well for me in my AL-only 5x5. Great thing is once you get your spreadsheet setup you can just paste in the new projections every year and VOILA! you have your values.

[jbluestone]

I have the book, have found it not very helpful in calculating PVM, which is what I prefer for HTH.

[RAY@HQ]

Single-category exercises like this are great for understanding the relative value of players, but they won't ever total out in the way you're asking. The reason for this is that the values are built from the bottom (the $1 player) up. So in SBs, for instance, if you're doing a one-category exercise, the $1 player might have, say, 6 SB. All of the other dollar values are built upward from there, based on how much better than 6 SB the player is.

But once you get into multi-category calcs, the baseline level for SB changes. In your 5-category model, the $1 player across all of those categories might have only 3 SB, instead of the 6 they had in the single-category valuation. So now the value everyone gets from SBs is based on how much better they are than a 3-SB guy instead of a 6-SB guy. That has a big impact on the values, when all five baseline numbers are different in the 5-category calc from the single-category calc.

The more tedious, but perhaps better, approach might be to do a series of four-category valuations, leaving off one of the 5 categories every time. That might back you in to the single-category value based on the gap between the four-category value and the five-category value.

[jbluestone]

Single-category exercises like this are great for understanding the relative value of players, but they won't ever total out in the way you're asking. The reason for this is that the values are built from the bottom (the $1 player) up. So in SBs, for instance, if you're doing a one-category exercise, the $1 player might have, say, 6 SB. All of the other dollar values are built upward from there, based on how much better than 6 SB the player is.

But once you get into multi-category calcs, the baseline level for SB changes. In your 5-category model, the $1 player across all of those categories might have only 3 SB, instead of the 6 they had in the single-category valuation. So now the value everyone gets from SBs is based on how much better they are than a 3-SB guy instead of a 6-SB guy. That has a big impact on the values, when all five baseline numbers are different in the 5-category calc from the single-category calc.

The more tedious, but perhaps better, approach might be to do a series of four-category valuations, leaving off one of the 5 categories every time. That might back you in to the single-category value based on the gap between the four-category value and the five-category value.

Hmm, good point, i was thinking it might have something with the changing player pool...

[jbluestone]

Basically, i have done some research, where i regressed end of year values with the number of HTH wins, and have gotten some pretty strong correlations and prediction equations. Basically in my $380 league I get the regression equation of HTHWins = .1(Value)+78, or in other words a replacement team would win about 78 games (we play 220... 10 per week for 22 weeks), and for each $10 of value a player earns your team would gain another win above 78.

However, since a win is essentially a category victory in a given week, I wanted to look at this more granular. So I am trying to figure out the best way to do that. I know what our league average is for each category for each week, I know what the replacement level is for each category, but I am having trouble figuring out the best way to convert this information into "wins", or essentially "category wins above replacement".

I tried looking at what it would take to win a typical week in each category, so for example 9 hr would win most weeks (though obviously not every week). Miguel Cabrera hit 44 hr in 26 weeks or about 1.7 per week. A replacement player would hit about 13.666 hr or .65 per week. This is where I am not sure of the best methodology. I am taking muggy and subtracting the replacement to get his above replacement value... which is about 1.05. I then divided 1.05/9 to get the proportion of miggys value added vs. what is needed (though I am not sure if that makes sense). to get a WAR value for miggy in HR of .116 per week or 2.548 over the 22 week regular season. I repeated this for each of his categories (except OPS which was more convoluted as a rate stat). Overall i got (for 26 week value since that what end of year values are calculated at) about 6 WAR which would equal about $60 in value... and the HQ custom valuator has miggy as worth $62 in our league.

Obviously I would have to run this for a bunch more players to see if the numbers are working out relatively similar, but I wanted to ask the smart people that work and post on the HQ boards if they had any thoughts about the methodology and ways to improve or even alter it.

My main goal, is I would like to be able to create a spreadsheet that allows me to plug in projections and figure out each players win value added per category.

I know this is fairly standardized using SGPs for roto, but since we are a HTH where each category counts as a win per week, I am not sure the clearest way of calculating this. I also know that even the best estimates will be muddy because of week to week inconsistency and matchup situations etc. but thats OK with me.

Thanks