An evaluation of preseason predictions

[FRENCHREDSOX]

Awesome, FRS!...who said that baseball fans were dumb?

It is basic Stats formula used since the late 18th century - nothing awesome my dear Rongar - now IF I had done it in my head,instead of "cheating" by using a calculator that WOULD have been AWESOME LOL

[boa]

Funny thing is, I was going to post on how bad Pecota was compared to previous years.

[HoustonGM]

Funny thing is, I was going to post on how bad Pecota was compared to previous years.

And that's funny because...it wasn't?

[boa]

I'm still waiting on the end of the season and the other prediction simulators. I might have been hasty though because it seems like this year was rather dynamic and everyone did poorly. Wasn't pecota around 5 wins +/- avg and slipped to 10?

[HoustonGM]

I'm still waiting on the end of the season and the other prediction simulators. I might have been hasty though because it seems like this year was rather dynamic and everyone did poorly. Wasn't pecota around 5 wins +/- avg and slipped to 10?

That's correct...the accuracy of EACH prediction was vastly worse this year, at least compared to last year. I don't know about years prior to that. Either last year was incredibly easy to predict, or this year was very hard to predict...or both. This year certainly had a good deal of surprises. PECOTA is the only system that saw Tampa contending...nearly nobody expected the White Sox and Twins to be at the top of the AL Central, as most people choice the Tigers and Indians to go 1-2 in some order...etc.

[FRENCHREDSOX]

That's correct...the accuracy of EACH prediction was vastly worse this year, at least compared to last year. I don't know about years prior to that. Either last year was incredibly easy to predict, or this year was very hard to predict...or both. This year certainly had a good deal of surprises. PECOTA is the only system that saw Tampa contending...nearly nobody expected the White Sox and Twins to be at the top of the AL Central, as most people choice the Tigers and Indians to go 1-2 in some order...etc.

apart from BM mog sims eg Yours N mine + default

[FRENCHREDSOX]

Funny thing is, I was going to post on how bad Pecota was compared to previous years.

It has its "standard" 10% deviation as ALWAYS because it overvalues "rookies" but THE WRONG rookies

[DarthJaker]

just found my preseason picks
NL East:Braves
NL Central:Brewers
NL West:Rockies
NL WC:Mets
NLCS:Braves over Rockies
AL East:Redsox
AL Central:Tigers
AL West:Mariners
AL WC:Indians
ALCS:Indians over Redsox
WS:Braves over Indians
wow i screwed up

[HoustonGM]

It has its "standard" 10% deviation as ALWAYS because it overvalues "rookies" but THE WRONG rookies

[RedsoxRockies]

My picks actually had the Rays contending and Cleveland struiggling, but also had Chicago and Minnesota doing bad and the Tigers doing well. And I had the Rockies in 1st, and the Cubs behind Milwaukee.

[gRYFYN1]

Not sure how I feel I scored an 11.2 based on my BP preseasons.

TB, Sea, Was, SD (22 off ) cost me the most.

But NYA, NYM, TOR, were all only 1 off or less

[justanewguy]

I never predicted wins, but I had the NL perfect (so far) in my preseason predictions. With the Dodgers, Cubs and Phillies winning the divisions, and Mets taking the wild card. I predicted the Diamondbacks, Brewers and Astros to compete but come up short. In the AL I was way off, except I had Anaheim on top of their division. But who didn't? I had Cleveland in the Central, New York in the East, and Detroit for the wild card with Boston coming up a game or 2 short.

[Rongar]

It is basic Stats formula used since the late 18th century - nothing awesome my dear Rongar - now IF I had done it in my head,instead of "cheating" by using a calculator that WOULD have been AWESOME LOL

Me, I wouldn't even have known where to look on a calculator for a mean root, sliced & diced, or whatever

[FRENCHREDSOX]

PECOTA (& their analysts) LOVE to overvalue heavily "rookies"/prospects & ipso facto their production & their effect on teams BUT at the same time UNDERVALUES veterans & what I would call AAAA "rookies" who play for more significantly a role than PECOTA's predictions IRL - eg Tatis or Hoover come to mind.

Also BA tends to do the same - if you look at their top 100 & PECOTA's predictions they intermesh BUT if you compare "rookies" performance a lot of 08 rookies where unlisted in either system & thus have a zero effect on PECOTA's predictions (eg SFG or BAL who called up guys like Burris & Waters who had significant effects on the parent teams' season & thus W/L total)

Me, I wouldn't even have known where to look on a calculator for a mean root, sliced & diced, or whatever

LOL - all it was is the DIFFERENCE between the predictions squared (which gets you always a POSITIVE number) divided by 30 (the number of teams in the MLB) which gave each team's value. Then you added up all 30 results (E1,E2....E30) & simply square rooted.

Example Below :
Boston :
HGM 95 wins
IRL :.583 W% x 162 games = 94.446

IRL - HGM = 94.446 - 95 = - .554

E0 = -.554 x -.554 = 0.306916
E1 = E0/30 = 0.0102305333

Tampa:
HGM 77
IRL : .600 x 162 = 97.2

IRL - HGM = -20.2
E02 = -20.2 x -20.2 = 408.04
E2 = E02/30 = 13.60133333

& SO on......

You do the same for all 30 - add them up (E1 + E2 + ..... E30) & Square root the total

[HoustonGM]

PECOTA (& their analysts) LOVE to overvalue heavily "rookies"/prospects & ipso facto their production & their effect on teams BUT at the same time UNDERVALUES veterans & what I would call AAAA "rookies" who play for more significantly a role than PECOTA's predictions IRL - eg Tatis or Hoover come to mind.

PECOTA can't forsee unexpected breakthroughs. Not sure what the criticism is. It has nothing to do with heavily overvaluing rookies or undervaluing veterans. The PECOTA projected standings use projected playing time of players, which DOES favor veterans. While PECOTA gives projections for a bunch of young prospects, most of those guys don't get taken into account when they run the projected standings.

Also BA tends to do the same - if you look at their top 100 & PECOTA's predictions they intermesh BUT if you compare "rookies" performance a lot of 08 rookies where unlisted in either system & thus have a zero effect on PECOTA's predictions (eg SFG or BAL who called up guys like Burris & Waters who had significant effects on the parent teams' season & thus W/L total)

This is because BA ranks prospects, and that takes into account both "current" value and "future" value, and guys like Chris Waters don't have very good ceilings. They're filler players.

Yeah, PECOTA can't account for unexpected call ups, but nothing can.