OK here's an example of what I'm talking about. These are my OF rankings under one SB value. I then re-sort the list from highest to lowest and it comes out as the second list due to changes in the RP profile. Back and forth, back and forth. Ignore the actual dollar values as I have not yet sanitized the projections. Just pay attention to the valuation gaps.
I don't really understand what the dollar values are for... is there a typo with WHIP being $42 and HR being 42 cents? Maybe I'm just not understanding how you dealt with the rate stats. In any case, I'd probably weight K/BB more heavily since it's more predictable than ERA and WHIP and is the most telling stat about a pitcher which basically indicates how good the rest of his numbers will be.
Skin Blues wrote:I don't really understand what the dollar values are for... is there a typo with WHIP being $42 and HR being 42 cents? Maybe I'm just not understanding how you dealt with the rate stats. In any case, I'd probably weight K/BB more heavily since it's more predictable than ERA and WHIP and is the most telling stat about a pitcher which basically indicates how good the rest of his numbers will be.
$42 for WHIP means every 1.00 above replacement value is worth $42 for a player's auction price. So if replacement player is 1.30 WHIP, a player with a projected 1.20 WHIP would be worth $4.20 assuming equivalent IP.
I dialed back K/BB b/c i felt that keeping it high was double-counting K since I put so much weight on that category. It was also giving me a really high value on Halladay b/c he's such a stellar performer in that category.
...aaaaand i just found out i've been double-weighting rate stats. Excuse my brain fart, but should I be weighting pre-unit-valuation or post-? Or does it matter?
Rate stats should be converted into counting stats. Otherwise a 3.50 ERA in 210 IP counts the same as a 3.50 ERA in 70 IP, which is obviously not the case. There's no hard and fast way of doing this. From the top of my head, you could find the amount of innings your pitchers are expected to pitch (this is easy if you play with an innings limit) and choose a nominal ERA. Maybe the average ERA of players in the player pool, or some other value. Then you can see how many earned runs a team will give up. Say you have a 4.1 ERA over 1200 IP as your nominal values. You'd have 547 ER in a full season. A pitcher with 210 IP and a 3.50 ERA would give up 81.7 ER compared to the nominal value of 95.7 ER. He saved you 14 ER. The guy with 70 IP and the same ERA gave up 27.2 ER with the nominal value being 31.9. He saved you 4.7 ER. We can clearly see that the guy with three times the IP saved three times as many ER. Using "runs saved" as a counting stat makes things much easier to quantify. Same can be done for BA, WHIP, etc. It's a lot of work but it's worth it.
Skin Blues wrote:Rate stats should be converted into counting stats. Otherwise a 3.50 ERA in 210 IP counts the same as a 3.50 ERA in 70 IP, which is obviously not the case. There's no hard and fast way of doing this. From the top of my head, you could find the amount of innings your pitchers are expected to pitch (this is easy if you play with an innings limit) and choose a nominal ERA. Maybe the average ERA of players in the player pool, or some other value. Then you can see how many earned runs a team will give up. Say you have a 4.1 ERA over 1200 IP as your nominal values. You'd have 547 ER in a full season. A pitcher with 210 IP and a 3.50 ERA would give up 81.7 ER compared to the nominal value of 95.7 ER. He saved you 14 ER. The guy with 70 IP and the same ERA gave up 27.2 ER with the nominal value being 31.9. He saved you 4.7 ER. We can clearly see that the guy with three times the IP saved three times as many ER. Using "runs saved" as a counting stat makes things much easier to quantify. Same can be done for BA, WHIP, etc. It's a lot of work but it's worth it.
Converting rate stats to counting stats is unnecessary as long as your weighting them by IP/AB/PA, right?
Yeah it does essentially the same thing. I'm not a math whiz so I'd have to do some samples to see what method works the best, but as long as it's weighted by innings, etc it should accomplish the same goal. It might be a bit wonky and depend on order of operations to incorporate replacement level and all that. I just use "runs saved" and stats like that to keep it simple and consistent, but that doesn't mean it's necessary.
Cafe Home
Fantasy Football
Fantasy Basketball
Fantasy Hockey
by RugbyD » Wed Jan 12, 2011 10:53 am 
