Resulting Stats (based on a neutral, modern Major League environment)

[Tony M]

How are these figures calculated? I'm trying to figure it out.

Take Dice-K for example.

His actual stats are 331AB (I think), 85 H, 9 HR, 27 BB, 85 K, 0.257 OAVG

I can get close to this by setting

Stuff = 136, Movement = 126, Control = 113 and Groundball % = 95! and setting the AB field to 331 I replicate everything except ERA (which is calculated at 3.74 and not his over 4.5). His OAVG in OOTP is 0.001 lower.

To replicate Schilling's stats so far

Stuff = 101, Movement = 111, Control = 160 and Groundball % = 95 again, his OAVG in OOTP is 0.001 lower than in RL

To replicate Beckett's stats

Stuff = 120, Movement = 154, Control = 145 and Groundball % = 23. His OAVG in OOTP is the same as in RL.

Is the formula for calculating Stuff, Movement, Control and GB% from H,HR,BB,K readily available?

[injury log]

I'm looking into this, it's sort of fun. I've left AB at 550, kept vs L and vs R values identical, and I've figured out how Control and BBs are related in projected stats- it's a three-part linear function:

Control > 210 --> BB = 5/550 AB
Control from 100-210: BB = -0.4*Control + 90 per 550 AB
Control from 0-100: BB = -1*Control + 150 per 550 AB



Now for the other ratings...

[injury log]

The output appears to be all linear; the only potential difficulty is the interaction between Movement and Stuff, but I don't think that's a big issue. Anyway, leaving Movement at 100, Stuff changes produce linear changes in predicted stats. For K values from 1-200, per 550 ABs:

Ks = 0.95*Stuff + 10 [had 9.5, had shifted decimals everywhere!]

For Stuff values over 201, nothing changes; the slope is 0.

Predicted IP = 0.1*Stuff +128

Predicted Hits allowed = -0.27*Stuff + 175

All of these values are affected by changes in Movement, however. I'll try to get a formula that combines them.

[Tony M]

Originally Posted by injury log

I'm looking into this, it's sort of fun. I've left AB at 550, kept vs L and vs R values identical, and I've figured out how Control and BBs are related in projected stats- it's a three-part linear function:

Control > 210 --> BB = 5/550 AB
Control from 100-210: BB = -0.4*Control + 90 per 550 AB
Control from 0-100: BB = -1*Control + 150 per 550 AB



Now for the other ratings...

I've started that as well, but I stuck AB up at 99999 because it gives a wider range of values.

Control is the only one to effect BB, Movement the only one to effect HR.

[StyxNCa]

Wouldn't it be a lot easier, not to mention more reasonable, to allow us to set up the era we are putting stats in for and just let us put stats in and let the ratings be based on that? This is the only game I've ever seen where we have to try to juggle the ratings to get the stats we want. Seems stupid to me.

[injury log]

So Movement is linear in HR as well. At 550 AB (it seems the line is normalized around 550; I get slopes which are perfect with one decimal place, in almost all cases), HR = 38 - 0.18*M [edited; I had 1.8 before, which is obviously ridiculous]. Above 200, Movement doesn't affect projected HR rate.

As you say, you're constrained from the outset to choose appropriate Movement and Control values to match BB and HR rate, so the interesting question is how GB%, Stuff and Movement affect AVG. That's the last question...

[injury log]

For Movement values above 100, HA - HR is constant for a given Stuff rating. This makes it possible to work out the HA/550 AB (this for a 50% GB) by finding the HA-HR equation for a given Stuff rating, then adding in the HR predicted by the Movement value. The equation is:

HA = 193 - 0.28*Stuff - 0.18*Movement

I haven't tried this for Movement values lower than 100, but it appears that extra hits (non HRs) are added in for low values of Movement, which will probably mess with the equation.

[injury log]

My first triola,

GB% appears linear in hits allowed on balls in play, predictably. Assuming GB% affects BABIP linearly, then the game produces a 0.273 BABIP for a 0% GB%, increasing by 0.0047 per 10% increase in GB%. So, baselining at 550 AB, and using the above:

HA = BIP*[0.273 + 0.0047*[GB%/10]] + HR
HA = [550 - K - HR][0.273 + 0.0047*[GB%/10]] + HR
HA = [550 - 0.95*S - 10 - 38 + 0.18*M]*[0.273 + 0.0047*[GB%/10]] + 38 - 0.18*M
HA = [502 - 0.95*S + 0.18*M]*[0.273 + 0.00047*GB%] + 38 - 0.18*M

and the final formula:
HA = 175 - 0.26*S - 0.13*M + G/100*(23.6 - .045*S + 0.0085*M)

I've tried with two sets of values (Movement and Stuff > 100 in both cases), and the formula is dead on- doing this all by hand, so it isn't quick to batch test numbers.

[RchW]

Originally Posted by redsoxford

How are these figures calculated? I'm trying to figure it out.

Take Dice-K for example.

His actual stats are 331AB (I think), 85 H, 9 HR, 27 BB, 85 K, 0.257 OAVG

I can get close to this by setting

Stuff = 136, Movement = 126, Control = 113 and Groundball % = 95! and setting the AB field to 331 I replicate everything except ERA (which is calculated at 3.74 and not his over 4.5). His OAVG in OOTP is 0.001 lower.

To replicate Schilling's stats so far

Stuff = 101, Movement = 111, Control = 160 and Groundball % = 95 again, his OAVG in OOTP is 0.001 lower than in RL

To replicate Beckett's stats

Stuff = 120, Movement = 154, Control = 145 and Groundball % = 23. His OAVG in OOTP is the same as in RL.

Is the formula for calculating Stuff, Movement, Control and GB% from H,HR,BB,K readily available?

Are these the ratings in the edit player tab? I hope so 'cause in my leagues a pitcher in the 70's is a stud. Sorry to be dense.

[RonCo]

Originally Posted by injury log

My first triola,

GB% appears linear in hits allowed on balls in play, predictably. Assuming GB% affects BABIP linearly, then the game produces a 0.273 BABIP for a 0% GB%, increasing by 0.0047 per 10% increase in GB%. So, baselining at 550 AB, and using the above:

HA = BIP*[0.273 + 0.0047*[GB%/10]] + HR
HA = [550 - K - HR][0.273 + 0.0047*[GB%/10]] + HR
HA = [550 - 0.95*S - 10 - 38 + 0.18*M]*[0.273 + 0.0047*[GB%/10]] + 38 - 0.18*M
HA = [502 - 0.95*S + 0.18*M]*[0.273 + 0.00047*GB%] + 38 - 0.18*M

and the final formula:
HA = 175 - 0.26*S - 0.13*M + G/100*(23.6 - .045*S + 0.0085*M)

I've tried with two sets of values (Movement and Stuff > 100 in both cases), and the formula is dead on- doing this all by hand, so it isn't quick to batch test numbers.

I hope you realize that the estimator is not reliable to any high degree of precision.

For example, testing actual results vs. ratings, one finds that an OOTP pitcher's GB% does not appear to have a statistically significant influence on his BABIP. There are calculation-order reasons for this that have been discussed in other areas.

[injury log]

Originally Posted by RonCo

I hope you realize that the estimator is not reliable to any high degree of precision.

For example, testing actual results vs. ratings, one finds that an OOTP pitcher's GB% does not appear to have a statistically significant influence on his BABIP. There are calculation-order reasons for this that have been discussed in other areas.

Yes, I don't put too much stock in the projected stats- I was more having fun with the math problem than anything else. Do you know if it's correct, however, to assume that a historical import determines ratings using these stat projections?

[RonCo]

Garlon or someone into historicals would be better to answer how ratings are imported. It's a shade complicated, and based on what parameters you select. But obviously, the base stats in the Lahman db are used as the fundamental building block and converted to ratings.

[jar2574]

So, injury_log, I've got my RC formula up and running.

Now I'm moving to a spreadsheet for XERA.

Thanks to your previous work on pitchers, I think this is doable, since all we need for XERA is expected Ks, BBs, HA, and HRs, and expected IP. I think you've done all that.

I only have one question. Am I right in dropping the forumlas from the portion of your thread where you discuss GB%. If that has no affect on HA, I'm going to use the (193 - .28 x Stuff - .18 Movement) for hits allowed.

Thanks for all your hard work on that stuff!

[injury log]

Originally Posted by jar2574

I only have one question. Am I right in dropping the forumlas from the portion of your thread where you discuss GB%. If that has no affect on HA, I'm going to use the (193 - .28 x Stuff - .18 Movement) for hits allowed.

If what RonCo says is true, that GB% doesn't affect HA, then you could omit it from the calculation. The only question is what GB% to assume to give a good result. In real life, the avg GB% for pitchers is slightly higher than 50% (around 54%), so the formula above should give an HA which is just slightly too low (higher GB% leads to more HA in OOTP). That's assuming OOTP produces a realistic number of groundouts, but I'd assume it does, considering how well it models most stats. Still, the difference probably isn't worth worrying about.

Originally Posted by jar2574

Thanks for all your hard work on that stuff!

It was fun, and do post an update about the results of your spreadsheet work. I imagine it might be very useful when making roster sets from real life players.

[jar2574]

Originally Posted by injury log

If what RonCo says is true, that GB% doesn't affect HA, then you could omit it from the calculation. The only question is what GB% to assume to give a good result. In real life, the avg GB% for pitchers is slightly higher than 50% (around 54%), so the formula above should give an HA which is just slightly too low (higher GB% leads to more HA in OOTP). That's assuming OOTP produces a realistic number of groundouts, but I'd assume it does, considering how well it models most stats. Still, the difference probably isn't worth worrying about.



It was fun, and do post an update about the results of your spreadsheet work. I imagine it might be very useful when making roster sets from real life players.

From what I've seen, the % of groundballs reflects the % in the GB rating for each pitcher. It's incredibly close over the course of a seasons.

So as long as the game models GB rating realistically, the % of groundballs should be realistic.

That said, I think I'll just use your basic HA formula, just to avoid the extra work of adding the GB part.

I'll try to post my results in the other thread when I get a minute. I wish I could post a spreadsheet, but I don't know how, and I don't know if it's safe to do so.

I just took the numbers we came up with and plugged them into Bill James's (very) basic Runs Created formula. Off the top of my head, I think it is ((H + BB) * TB) / (H + BB).

Now I know that RC formula doesn't account for GIDP, SB, CS, HP, SF, SH, etc.... But James's says it's 90% accurate, and that's close enough for me. I have been building teams in the Billy Beane manner (i.e. few basestealers, lots of walks & HR). So SB isn't that important to me. I don't use SH or SB much. And I figure GIDP evens out over time (except for really slow guys.)

So that basic RC formula is what I plugged into excel after I found the calculations for H, BB, and TB.

You showed me that Contact was related to Hits. I then took H - doubles - triples - HR to get singles. From there, total bases was a cinch.

[MrHaroldG]

I understand a pitcher's projected ERA is based on ratings on stuff, movement, control, ground ball % and even hit batsmen. Is there any type of formula that allows you to calculate his projected ERA?