Page 1 of 1

Question on valuation using MLE to fill in data

Posted: February 9th, 2018, 1:55 pm
by Aquila
I struggle with MLE. The 3 yr weighted average is fine for players who are veterans but when applying MLE it seems that younger guys get a major penalty. Take Jimmy Sherfy as an example. His projection calls for virtually full season's innings in the ARI pen. They also say 5.68/1.46 ratios. It feels odd to ding him so drastically for 1st time at level hiccups. Especially since there has been 1 bad streak of homers (1st time at Reno, PCL) and that the bb% seems to be solved.

Do you ever look at numbers, see they feel wrong, then do deep dive and adjust?

Thanks

Re: Question on valuation using MLE to fill in data

Posted: February 9th, 2018, 2:06 pm
by Todd Zola
I wouldn't call it a major penalty, but there are some cases where MLEs could under-appreciate the talent level. So yeah, as I discuss HERE, I'll massage some projections.

Re: Question on valuation using MLE to fill in data

Posted: February 13th, 2018, 4:47 pm
by Aquila
Well,

Here's the deal. I have been running my own numbers for over 10 years....Based on a conversation you and I had on the HQ forum way back then. You sent me a copy of a first generation manuscript which showed hit points and pit points. I have massaged the information over the years, basically heading in the direction you are now. Alas, I am probably 5 generations behind. At this point there is no need for me to catch up and learn the techniques when you provide the data.

Question. Why do you think your results and those of Rotowire and HQ have a decent amount of variance? Maybe not so much HQ, I find their numbers skewed due to they have a number of people who do projections and playing time evaluations. Maybe Jeff applies more gut opinions and less numbers (and maybe too much Liss??? LOL).

Re: Question on valuation using MLE to fill in data

Posted: February 13th, 2018, 5:06 pm
by Todd Zola
Sorry, I can only speak to how I do things.