Page 1 of 2

Math help- what is my next step

PostPosted: Sat Feb 22, 2014 8:34 pm
by AEMRICH
Hi Everyone,

I've spent about 10 hours today getting ready for a points league that I'm in and I think my brain is fried now, so I need help with my reasoning. I've calculated my projected points for each player and I have the current avg. auction values on each player. What I am trying to do now is figure out how to best max out my $ for the max amount of points. I've put together some MOCK targets to see what combo seems to produce the most number of points.

What I'm trying to do now is go position by position and figure out who I should grab. For example what is better Wilson Ramos at $1 with 260 points or Matt Wieters at $3 and 310 points? Where I am stuck, currently, is how do I go about comparing price and points for each player.

Not sure if that makes any sense at all, but any help would be appreciated.

Re: Math help- what is my next step

PostPosted: Sat Feb 22, 2014 9:27 pm
by Skin Blues
I've never played in a points league, but this should work in theory. As with anything, theory can be impractical, so if something looks way off, it probably could use some tweaking.

First, if it's roto with Games/IP limits, you might want to convert projected fantasy points into points/game for hitters, and points/IP for pitchers. If it's H2H, you can probably skip that step. I will give the rest of the guide assuming you're using total points for H2H but it's easy enough to apply the theory to points/game points/IP.

Rank all players by projected points (or points/game, points/IP). Calculate total active rostered players (ie: 12 teams x 13 players/team = 156 players). Check if there's enough players at each position to fill all rosters within those 156. If not, add enough points to each player at those positions (same amount to every player at a given position, ie: add 100 points to all catchers and 50 points to all SS) until the top 156 have enough players to fill all active roster spots. Once this is done, subtract X amount of points from every player, where X is the amount of points scored by the 156th ranked player. The remaining value is the PAA (points above replacement; it will be negative for players ranked worse than 156). Now, simply find out how much available money there is for each PAA. If there are 20,000 total PAA among the top 156 players (don't include negative values or else theoretically there could be a negative amount of total points) and total money allocated to hitters of $2184 (12 teams x 70% of a $260 budget in a normal 70/30 hitter/pitcher split), then each PAA is worth $0.1092. So if Miggy is worth 550 PAA, then he is valued at $60.

Do the same for pitchers, except the budget changes to $936 in this example.

Now, you should have values to be able to determine if Player X at $3 is a better buy than Player Y at $1. This probably seems confusing, and it's possible I screwed something up, but if you have any questions just ask. As I said earlier it will also likely need some tweaking. For instance, valuation of closers almost never follows theoretical values, and since supply is so low, it's a very volatile market that depends on owners' tendencies.

Re: Math help- what is my next step

PostPosted: Sat Feb 22, 2014 10:22 pm
by AEMRICH
Skin Blues wrote:I've never played in a points league, but this should work in theory. As with anything, theory can be impractical, so if something looks way off, it probably could use some tweaking.

First, if it's roto with Games/IP limits, you might want to convert projected fantasy points into points/game for hitters, and points/IP for pitchers. If it's H2H, you can probably skip that step. I will give the rest of the guide assuming you're using total points for H2H but it's easy enough to apply the theory to points/game points/IP.

Rank all players by projected points (or points/game, points/IP). Calculate total active rostered players (ie: 12 teams x 13 players/team = 156 players). Check if there's enough players at each position to fill all rosters within those 156. If not, add enough points to each player at those positions (same amount to every player at a given position, ie: add 100 points to all catchers and 50 points to all SS) until the top 156 have enough players to fill all active roster spots. Once this is done, subtract X amount of points from every player, where X is the amount of points scored by the 156th ranked player. The remaining value is the PAA (points above replacement; it will be negative for players ranked worse than 156). Now, simply find out how much available money there is for each PAA. If there are 20,000 total PAA among the top 156 players (don't include negative values or else theoretically there could be a negative amount of total points) and total money allocated to hitters of $2184 (12 teams x 70% of a $260 budget in a normal 70/30 hitter/pitcher split), then each PAA is worth $0.1092. So if Miggy is worth 550 PAA, then he is valued at $60.

Do the same for pitchers, except the budget changes to $936 in this example.

Now, you should have values to be able to determine if Player X at $3 is a better buy than Player Y at $1. This probably seems confusing, and it's possible I screwed something up, but if you have any questions just ask. As I said earlier it will also likely need some tweaking. For instance, valuation of closers almost never follows theoretical values, and since supply is so low, it's a very volatile market that depends on owners' tendencies.


thanks! i will check this out and let you know how it goes. :-) ;-D

Re: Math help- what is my next step

PostPosted: Sat Feb 22, 2014 10:43 pm
by AEMRICH
the league has a limit of 1350IP. Do I divide each pitcher's points by 1350? Or do I divide by my projected IP for each player? I've tried with the 1350 and the Points Per IP is much lower than the Points per game for the hitters. Thus when I sort them the pitchers are all much lower than the hitters. Should I rank the hitters and pitchers separately?

Re: Math help- what is my next step

PostPosted: Sun Feb 23, 2014 1:39 am
by Skin Blues
(First of all, I accidentally used PAA as an acronym for Points Above Replacement in the first post. Should be PAR)

No, don't divide by 1350. Divide by each pitcher's IP. So, if Kershaw has 450 PAR in 225 IP, that's 2 points/IP. If Kimbrel has 210 PAR in 70 IP, that's 3 points/IP. Tweaking might be needed here to make sure that the total amount of IP among the X best pitchers equals 1350 IP x 12 teams (otherwise your ideal team would be full of relievers and despite having a great points/IP ratio, you would miss out on points by not hitting the IP limit). So, SP might need to be have points added to their totals before calculating PAR, in order to get enough of them in the positive value player pool (like I explained for SS and C earlier). This is admittedly complicated, but if you can figure it out it's much better than just using raw point totals.

And yes, you should do hitters and pitchers separately. Give ~70% of the budget to hitters, and ~30% of the budget to pitchers. This part can be tweaked, depending on settings. Could also be 60/40 or 50/50 depending on how points are given. You could look at points above replacement (PAR) totals for the top 156 hitters and 132 pitchers (or whatever they are) and use that as the ratio, so if its 125 PAR per hitter and 85 PAR per pitcher (important to use PAR and not raw points totals) would be a 60/40 split. But no guarantee this is optimal; this is where tweaking comes into play.

It makes sense in my head but I might not transfer it into words properly, so if you have any more questions feel free to ask.

Re: Math help- what is my next step

PostPosted: Sun Feb 23, 2014 4:34 am
by TheTrith
In a points league, why would you split it 70/30 (or even split at all?)
Shouldnt it depend entirely on # of starting lineup positions for hitters/pitchers? It doesnt matter where the points come from does it?

If it turns out that the most points/$ you can fit onto your roster is by spending 50% of your budget on 5 SP, then thats what you should do, no?

I could be missing something totally obvious. Now my head hurts.

Re: Math help- what is my next step

PostPosted: Sun Feb 23, 2014 12:10 pm
by Skin Blues
Not straight points per dollar, but points above replacement per unit (GP/IP) per dollar. Because 100 points in 50 IP is much more valuable than 100 points in 200 IP. So yeah, you probably don't want to just arbitrarily pick a 70/30 split. But there's also a bit of game theory that depends on playing style. It's not as simple as it seems at first, since there are a lot of variables.

Re: Math help- what is my next step

PostPosted: Sun Feb 23, 2014 5:16 pm
by TheTrith
This is really complicated, and not nearly as simple as I thought it would be at first glance.
Might not even be worthwhile to find the "optimal" player per position, since a single player going for over projected price could skew your entire auction strategy (assuming you didnt build a nearly infinite set of backup players for each position).

Just taking the point/$ ratio into your auction along with a plan for how many players you want at each price tier, a good idea of how many SP + RP you need to hit the limit, and a pretty good idea of which players are going to be undervalued is probably almost as effective and not nearly as time consuming.
And I imagine would still you give a pretty big edge over the majority of owners in the league.

Re: Math help- what is my next step

PostPosted: Sun Feb 23, 2014 9:58 pm
by Skin Blues
TheTrith wrote:probably almost as effective and not nearly as time consuming.

Sir, how dare you suggest that I stop spending waaaaay too much time analyzing data for barely any added benefit. That's what spring training is for.

Re: Math help- what is my next step

PostPosted: Mon Feb 24, 2014 12:51 am
by TheTrith
Lol, spring just wouldnt be the same, would it :-D