2018 OSFM Hybrid Roster Project

[patsen29]

Re: 2018 OSFM Hybrid Roster Project

If you are referring to OSFM, they do not change attributes, so the speed and steal attributes are what sds set them at just a fyi.

Sent from my LG-D850 using Tapatalk

Right. We use statcast sprint speed to determine speed for major leaguers, and Bill James speed scores for minor leaguers. Those include steals, but also other signals, such as triples and runs.

If you have suggestions to improve minor league speed, we'd be happy to listen.

[forme95]

Re: 2018 OSFM Hybrid Roster Project

Right. We use statcast sprint speed to determine speed for major leaguers, and Bill James speed scores for minor leaguers. Those include steals, but also other signals, such as triples and runs.

If you have suggestions to improve minor league speed, we'd be happy to listen.

I was just responding to the poster that osfm (ridin) doesn't change attributes, if that was the roster he was referring to. To his comment of speeds being to low. It was nothing to you guys about how you edit attributes.

Sent from my LG-D850 using Tapatalk

[KingJHip]

Re: 2018 OSFM Hybrid Roster Project

Any word on when the rosters will be ready?

[Bunselpower32]

Re: 2018 OSFM Hybrid Roster Project

Any word on when the rosters will be ready?

Two things. First, despite there being answers in the thread, it's not on the OP so I'll forgive you for not searching through that massive pile of 11 pages of posts to find it.

Second, again, benefit of the doubt because your join date is kinda new, asking when these rosters will be ready is like talking about a perfect game in the 8th inning. Don't do it.

The question just doesn't need to be asked. There is no beneficial reason to anyone, including yourself, other than making them answer it and not work on the roster.

[WaitTilNextYear]

Re: 2018 OSFM Hybrid Roster Project

Right. We wanted to handle more ourselves, so we don't get burnt when sources drop out.

We're pulling statcast pitch data, which is the same data Brooks draws from. It's ball and swinging strike based, but with a few adjustments to smooth out the edges.

Results look really promising. I can't wait until we can start releasing.

Just two cents from a guy who was in the neighborhood, but I wouldn't consider data from Statcast and Brooks Baseball to be analogous even if they are both pulling from MLBAM. I find the Statcast pitch data to be pretty underwhelming and less accurate than what's at Brooks (I also vaguely remember reading a FiveThirtyEight dot com article on Statcast pitch inaccuracies).

When comparing both websites last year--I wanted to use Statcast because it is easier to download and has fancy additions like spin rate--I noticed that lots of guys would have both sinker and 2-seamer (they shouldn't and it shows the arbitrariness of their pitch detection methods) and there were also significant discrepancies between slider and curveball along with no data at all on pitch grips.

I think Brooks' experience with PITCHf/x and their method of manually classifying pitches ultimately makes them superior and your data set will have fewer oddities in it. I'd also recommend checking out Baseball Prospectus' website because they host Brooks Baseball data there and you can sort/copy/paste by pitch type without grinding through hundreds of individual player cards.

tl:dr - Statcast is phenomenal for things like sprint speeds, pop times, launch angles, and all manner of sorting/collecting data, but I wouldn't use them for pitch edits.

[tapout8302]

Re: 2018 OSFM Hybrid Roster Project

Can’t wait for the roster. Thanks for doing it again. I just have 1 question. It seems in years past that most pitchers with a sinker, get switched to a 2sfb in the hybrid roster. I’m just curious why that is. I can barely tell the difference in the game, but I’m curious why it’s done

[bobloblah1980]

Re: 2018 OSFM Hybrid Roster Project

Can’t wait for the roster. Thanks for doing it again. I just have 1 question. It seems in years past that most pitchers with a sinker, get switched to a 2sfb in the hybrid roster. I’m just curious why that is. I can barely tell the difference in the game, but I’m curious why it’s done

It's usually about the pitchers ball Movement.

[Qb]

Re: 2018 OSFM Hybrid Roster Project

It's usually about the pitchers ball Movement.

[Cavicchi]

Re: 2018 OSFM Hybrid Roster Project

Just two cents from a guy who was in the neighborhood, but I wouldn't consider data from Statcast and Brooks Baseball to be analogous even if they are both pulling from MLBAM. I find the Statcast pitch data to be pretty underwhelming and less accurate than what's at Brooks (I also vaguely remember reading a FiveThirtyEight dot com article on Statcast pitch inaccuracies).

When comparing both websites last year--I wanted to use Statcast because it is easier to download and has fancy additions like spin rate--I noticed that lots of guys would have both sinker and 2-seamer (they shouldn't and it shows the arbitrariness of their pitch detection methods) and there were also significant discrepancies between slider and curveball along with no data at all on pitch grips.

I think Brooks' experience with PITCHf/x and their method of manually classifying pitches ultimately makes them superior and your data set will have fewer oddities in it. I'd also recommend checking out Baseball Prospectus' website because they host Brooks Baseball data there and you can sort/copy/paste by pitch type without grinding through hundreds of individual player cards.

tl:dr - Statcast is phenomenal for things like sprint speeds, pop times, launch angles, and all manner of sorting/collecting data, but I wouldn't use them for pitch edits.

Yup, totally agree. And that's why I liked your pitch edits last year, most accurate of all.

[WaitTilNextYear]

Re: 2018 OSFM Hybrid Roster Project

I'm suddenly having the urge to get some nachos, befriend a guy named Todd (who is totally cool), and watch some Gwar music videos. Because Gwar rules.

[patsen29]

Re: 2018 OSFM Hybrid Roster Project

Just two cents from a guy who was in the neighborhood, but I wouldn't consider data from Statcast and Brooks Baseball to be analogous even if they are both pulling from MLBAM. I find the Statcast pitch data to be pretty underwhelming and less accurate than what's at Brooks (I also vaguely remember reading a FiveThirtyEight dot com article on Statcast pitch inaccuracies).

When comparing both websites last year--I wanted to use Statcast because it is easier to download and has fancy additions like spin rate--I noticed that lots of guys would have both sinker and 2-seamer (they shouldn't and it shows the arbitrariness of their pitch detection methods) and there were also significant discrepancies between slider and curveball along with no data at all on pitch grips.

I think Brooks' experience with PITCHf/x and their method of manually classifying pitches ultimately makes them superior and your data set will have fewer oddities in it. I'd also recommend checking out Baseball Prospectus' website because they host Brooks Baseball data there and you can sort/copy/paste by pitch type without grinding through hundreds of individual player cards.

tl:dr - Statcast is phenomenal for things like sprint speeds, pop times, launch angles, and all manner of sorting/collecting data, but I wouldn't use them for pitch edits.

You make good points. The Statcast data seemed good enough for what we needed from it, and saved us a lot of data mining.

I wish we had spoke earlier, because we finished pitch edits weeks ago, and it's a bit late in the process to change things now. It's still better than what was previously available to us, so we'll roll with it now, but look at improving it going forward.

If you want to talk more on the matter, I'd be happy to hear from you. My goal is to publish the methodology and results once the roster is up, so we can build on it going forward.

[WaitTilNextYear]

Re: 2018 OSFM Hybrid Roster Project

You make good points. The Statcast data seemed good enough for what we needed from it, and saved us a lot of data mining.

I wish we had spoke earlier, because we finished pitch edits weeks ago, and it's a bit late in the process to change things now. It's still better than what was previously available to us, so we'll roll with it now, but look at improving it going forward.

If you want to talk more on the matter, I'd be happy to hear from you. My goal is to publish the methodology and results once the roster is up, so we can build on it going forward.

No worries. It's not worth re-doing that's for sure. Statcast is better than what the game ships with, which is basically "take an educated stab at it." With Statcast, your pitch velocities will be fine even if some of the pitch types can be wonky (2SM vs SNK, CB vs KCV vs SL vs SLRV, 4SM vs CUT). It's also possible that Statcast is more accurate this year than in years past. Sorry I didn't make my post earlier.

[KingJHip]

Re: 2018 OSFM Hybrid Roster Project

Two things. First, despite there being answers in the thread, it's not on the OP so I'll forgive you for not searching through that massive pile of 11 pages of posts to find it.

Second, again, benefit of the doubt because your join date is kinda new, asking when these rosters will be ready is like talking about a perfect game in the 8th inning. Don't do it.

The question just doesn't need to be asked. There is no beneficial reason to anyone, including yourself, other than making them answer it and not work on the roster.

It's not like I put a gun to their heads and demanded an answer. I was just curious because I haven't been able to do jack **** without the rosters. I've had the game since it came out and frankly i'm getting bored with it because I can't do franchise.


All I want is a roster with all the correct names, i'd use OSFM but they insist on including the top prospects who might still be in rookie league with ZERO chance of making High A by years end! I don't need all that nerdy BABIP, spin rate, Statcast bull**** I don't care about stupid numbers, i'm a traditionalist. I just want names.

[xNobleEaglex]

Re: 2018 OSFM Hybrid Roster Project

It's not like I put a gun to their heads and demanded an answer. I was just curious because I haven't been able to do jack **** without the rosters. I've had the game since it came out and frankly i'm getting bored with it because I can't do franchise.


All I want is a roster with all the correct names, i'd use OSFM but they insist on including the top prospects who might still be in rookie league with ZERO chance of making High A by years end! I don't need all that nerdy BABIP, spin rate, Statcast bull**** I don't care about stupid numbers, i'm a traditionalist. I just want names.

You're probably looking at the wrong roster then, if names are all you care about. This roster is annually one of the best out there because it re-rates every players' ratings to make players play like their real life counterparts, and it fixes the pitch edits and equipment to make them more realistic.

Obviously that is a huge task to accomplish, and it takes a lot of time to do so.

[JDRAKE349]

Re: 2018 OSFM Hybrid Roster Project

Thank you for making this! I asked around on Twitter a month ago if anyone created such a roster and got flamed big time for not being happy with OSFM (great roster but correct MiLB without correct MLB doesn’t do much for me). Can’t believe no one could just point me to this.

Again, thank you!! I know all the ridiculous hours it takes to do these things and get them right!