Trying to understand the value of pitching

[cummings2]

Hmm...interesting,

So what you guys are talking about is that for a notoriously groundball pitcher the IFxs would have a higher pct. (as it is now) but for Flyball pitchers the OFx would be challenged more often?

And altogether high strikeout pitchers would have an even lesser reliance on defense?

Hmmm...

Not quite sure about this...I think I agree with the fact that if a pitcher is a notorious Fly ball pitcher his card should challenge the OFers more often. GB pitchers with IF as well.

As far as high K pitchers not relying as much on defense...not quite sure if I agree because conversely this would mean that a SS would have a better FLD PCT behind Randy Johnson than say behind Paul Byrd just because it's RJ.

I agree with your ideas U-Shocks and Lucky so I find it hard to hold my stubborn line of disagreeing. There must be something I am not getting in my head...you mean to tell me that the [u:361760febc][b:361760febc]Fielding Percentage[/b:361760febc][/u:361760febc] is actually higher (Better) with high K pitchers?

The ammount of errors I understand and I agree with entirely but:

Lets say that after 100 balls the defense makes 10 errors
and after 50 Balls the defense makes 5 errors

Still the same Fielding PCT. More errors, true, but the same PCT.

I really think that better defenses make better pitchers, not the other way around.

But as I said, I think I'm not getting something of the argument or that I am getting it wrong.

For what is worth I agree entirely with the GB pitchers having a card that reflects so and the FB pitchers as well.

Food for thought: don't you think that the cards compensate for this by the usage of the gb(A) on the pitcher's card?

[UrbanShockers]

More GBA would reward the pitcher, but the reward would be the same whether he has a good IF D or a bad one behind him, which isn't realisitic.

I would assume that (real world) the fielder's Fldg. Pct. and range are the same behind any pitcher. Some believe that K pitchers depress one or both because fielder's lose focus, but that would apply only to the wild ones, I think, and isn't certain (and certainly isn't quantifiable) even as to them.

The thing the current setup doesn't capture is that the variables represented by FP and range would come into play more often with a pitcher who lets more balls into play than one who lets less into play. The same is true as to those factors, specific to IF and OF D, for groundball and flyball pitchers, respectively.

I think the cards show more GB for some P's and more FB for others. But, if I'm right, they do this only in the non-X results, where range and e-number are irrelevant. That prevents the Def.'s strengths and weaknesses from interacting with the P's tendencies in the way that I think they should: a way that would not only enhance realism, but also add another dimension to strategy).

So it seems to me . . .

[maligned]

Very provocative analysis on all of this, Cummings. I have two thoughts about your original Perez/Clemens analysis:

1. As was mentioned above, you didn't consider SLG. nearly enough. I've noticed that TSN values opponent OPS(or at least HR allowed) very highly in pricing pitchers. If you see two pitchers who are priced differently with similar WHIP and ERA, all you have to do is look at the cards quickly to see that the difference is power allowed. ERA seems to not be considered at all. TSN seems to be much more concerned about considering the frequency of raw results (K,BB,HR,H, etc.) in terms of how a pitcher's ERA SHOULD look. This is not dissimilar to valuing hitters based on OPS. A site I often look at is Jeff Sagarin's ratings site. He uses Markov Chain mathematical analysis to predict expected pitching and hitting values of players. Basically, Markov Chain analysis, as I understand it, allows you to take the frequencies of specific events in terms of how they cause other events and make predictions about how combined event frequencies will predict events in the future. Sagarin has included more than 50 years of baseball data to predict how the combinations of different events (BB, H, HR, SB, CS, etc.) result in one other specific event: a run being scored in baseball. The long and short is, he takes Roger Clemens' raw allowed stats from 2004 and predicts his ERA at 2.55. He predicts Perez's to be 2.89...a 0.34 difference. This is because of Perez's higher rate of allowed SLG.

2. I don't know if anyone else mentioned it, but Clemens will not end up pitching 7/6 the amount of innings per start that Perez pitches because of the endurance rules in Strat. Basically, once the two have reached their endurance innings, they become equals in endurance. As long as they both continue to pitch well, they stay in the game. Clemens would theoretically not come close to pitching 7/6 the amount of Perez per start because both would have so many quality starts that would go beyond 6 or 7 innings.

In summary, my guess is that TSN does not predict such a large difference in Innings Pitched with these two pitchers. Thus, Perez's comparative Dollars/Inning go down. In addition, I think TSN predicts Clemens' average outcomes to be a reasonable amount better than Perez's, which explains why Clemens would cost more per inning.

To figure out endurance values, I would recommend comparing two very closely related pitchers like Rich Harden and Tim Hudson. They are both right-handed, faced the same real competition playing for Oakland, and their allowed OPS and Sagarin predicted ERA for 2004 are virtually identical (.684/3.10 and .685/3.08, respectively). TSN values the S7 (Hudson) at 3.97, and the S6 (Harden) at 3.62. In this one case, TSN seems to give the S7 almost exactly a 10% bonus on top of the S6 value. In other words, don't take 10% of the S7 and subtract, but take 10% of the S6 and add. I haven't compared any other similar examples, but this seems to give us one glimpse into endurance value for starters. I'll let you know if I come up with more examples.

With all of this said, as I mentioned, all of your logic was very provocative. Much of what you say, I agree with. You mention numerous points that I haven't seen discussed anywhere else and seem to be right on.

[maligned]

Actually, I already found a trio of pitchers I'd like to analyze: Mussina, Vazquez and Lieber. Again, all right-handed and all from the same team. Thus, they all faced the same competition, which Strat has identified as an important component in determining results for its cards.

Lieber: .757 OPS, 3.63 Sagarin ERA, S7, $1.64
Vazquez: .756 OPS, 3.68 Sagarin ERA, S6*, $1.59
Mussina: .759 OPS, 3.70 Sagarin ERA, S6, $1.52

With these SPs of lower value, keeping them in the game longer and using them on three days' rest is not as advantageous. Thus, TSN seems to give Lieber only an 8% bonus for an extra inning of endurance and Vazquez only a 5% bonus for his *value.

[cummings2]

O.K. I am running thorugh the comparisons using the process I described above. I will post the results of the comparisons for us to feast on, hopefully you have found a major mistake in these computations so we can all learn something.

Just one point before I go on: You lost me a little in the Clemens/Perez example, especially since at the end of the arguments you say:

[quote:263af741a4]I think TSN predicts Clemens' average outcomes to be a reasonable amount better than Perez's, which explains why Clemens would cost more per inning.[/quote:263af741a4]

But according to my simplistic math Oliver perez is actually a bit more expensive than Clemens per inning :?:

But I'll go through the numbers and post you what I come up with.

In regards to considering SLG. This is where I am at in my current number-crunching. It seems to me that the key is finding the relation between OAVG and ERA, that gives you a rough estimate of the Opp. SLG. On the other hand, I believe that pitchers with a high WHIP but a low OAVG seem t have a higher value, the example that comes to mind is Al Leiter. I believe that the reason for this is that the pricing is somehow related to what the pitcher is responsible for, that is the outs, walks, hits and nature of hits given.

Also, keep in mind that the numbers I came up with were derived from their season stats as listed in their cards, quite simple actually but that simplicity might have overlooked some things you're catching.

So, to the naked eye:

Hudson has a WHIP of 1.26

He gave up 44 BB in 188.2 Ip, so he gave up (44/188.6=.[b:263af741a4]2333 BB P/Ip[/b:263af741a4])

He struck out 103 in 188.2 Ip so he K'd (103/188.6=.[b:263af741a4]5461 K P/Ip[/b:263af741a4])

He gave up 194 Hits in 188,2 IP, he gave up (194/188.6=[b:263af741a4]1.0286 H P/Ip[/b:263af741a4])

This would mean that his OAVG would be (1.0286 * 9) / (27 + (1.0286*9)
So His OAVG would be .[b:263af741a4]255[/b:263af741a4]

***The only reason why I posted the way I came up with the numbers is so that you guys can see them and probably spot any flaws in the numbers***

So similar thing with Harden :

BB/Ip= .427
K/Ip= .880
H/Ip= .9018
H/9ip= 8.1342
OAVG= .231


Now Pricing:

Harden 3.62M / 198 (innings for an S6)= $18,282.82
Hudson 3.97M / 231 (innings for an S7)= $17,186.14

This would seem to indicate that according to SOM Harden is a slightly better pitcher than Hudson given their parameters.

Now with this example it is fairly easier to illustrate a trend I've seen. Run some numbers on the K:BB Ratio and K:BF ratio on both pitchers and you'll see an interesting figure in front of you, tha relates to the point I talked about up above: High Strikeout pitchers seem to be slightly more expensive, my feeling is because they're slightly more capable of overcoming defensive mishaps.

Anyway, I'll run some of the other comparisons and post what I find, I am as interested in you elaborating on the topic as in whatever I come up with.

I should also mention that one of the ideas that's floating in my head right now is that the pricing [u:263af741a4][b:263af741a4]has[/b:263af741a4][/u:263af741a4] to be based on a universal set of criterea, that is I that I think that the pricing of players has to be derived from certain numbers, stats, etc that are applied to all cards, regardless of their balance, players they opposed, stadium they played in etc... the prices can not be derived from any subjective element because that would open a can of worms, so whenever I came up with trying to explain the value of a pitcher by saying "well, this one is more likely to..." I immediately dismissed that thought.

My logic behind all of this is (and the reason why I used the max total of innings for a pitcher given his endurance rating) that when you get a player you are paying for certain elements, some are priced at a higher premium than others, how those elements interact with your choice of stadium, lineups, manageral tendencies, etc that is up to you but the card is still the same. The context under which it's used changes but the card is still the same. That's why I tried to get to the numbers without any regards for stadium or defensive ratings behind the pitcher, or opposing lineups. I've been trying to stick to seeing numbers from the cards only.

This last point is where my I question my logic so much.

I really appreciate your input Maligned, I've been looking forward to your two cents since we talked about this thread several days ago.

[cummings2]

I forgot to mention,

In re: considering SLG.

That would be a factor modifying the pitcher's value per inning. I agree entirely with the fact that it has to be considered BUT as an element that plays a role in the pitcher's final value per inning.

So, the question would be (if we were to want to figure it out) how is a pitcher's salary calculated. For this, I'd think that the first step would be to bring all pitchers to a context where they can all be evaluated in equal terms, that was the logic that led me to going to the common denominator that I identified as the single inning.

As I said before in Disclaimer #3, I appreciate very much you guys telling me where my sense of logic has failed. :D

[maligned]

My point in my Clemens/Perez summary was that I agree that innings pitched are projected by TSN in determining value. However, I think that Clemens is projected to perform at a somewhat higher level per inning (because of the desparity between the two in allowed SLG). In addition, I think that Perez will be projected to pitch a greater ratio of innings compared to Clemens than the 6/7 per start that you project. For example, if you run a projection in which Perez pitches an estimated 220 innings, compared to the 287 for Clemens, Perez's price per inning comes down. In other words, I don't think your inning projections are correct because I don't think Perez is valued by TSN as highly as Clemens is in raw performance. You made the assumption they are valued based on having the same raw performance. Perez HAS to have a lower price per inning for their comparative values to make logical baseball sense. Stay with me...here are a lot of thoughts I have about how value is determined:

I think the base for determining value with pitchers is centered on their raw projected allowed results in terms of how they affect a game. All prominent baseball theory today suggests that you can very closely predict the number of runs a team will score by simply multiplying its OBP times its TB. Look at any team from any season in history and you will find this to be a fairly accurate quick predictor. Thus, the best quick predictor of value for any hitter is his OPS (OBP + SLG) and the best quick predictor of pitcher value is his allowed OPS. ERA, for example, is a very rudimentary way to assess a pitcher's true success from hitter to hitter. Look at Rick Bauer. He had a very poor ERA, but is nonetheless valued very highly by Strat. He got VERY unlucky in real life. His allowed OPS was very good, and his raw sesults, as they play out in a simulation, played against the same competition he faced during the 2004 season will make his ERA much, much lower than what he actually earned during the 2004 season. (As a sidenote, the K has been mathematically shown to have a minimally greater effect on a pitcher's performance compared to other types of outs.) The way that Strat has always said it has made its game cards reflect reality is by making its players produce raw individual results (BB, 2B, 3B, HR, K, etc) at the same rate that they did during the season the card is supposed to represent. If an AL pitcher allowed 5% walks during the season against AL opponents, he will walk 5% of his hitters when he faces the same opponents in a Strat replay. The ONLY way to truly assess any player's value, then, is to predict how he projects to affect an offense(hitters) or how many runs he projects to allow (hitters) based on these raw results.
I know that Strat projects the pitchers I mentioned before (Harden/Hudson, for example) to produce very similar results in the long run. They had virtually identical opponent OPS values and played against the same competition. Thus, we can use these pitchers to eliminate the questions of which stats contribute to monetary value, and focus solely on how endurance, *'s, significant strength against one side of hitters, and ballpark effects contribute to monetary value.
The problem is, you have to go to another source to find better stats like opponent OPS, then come back to the Strat site and run your calculations.

[maligned]

I forgot one important clarification: In summary, I think you were trying to figure value based too much on inning projections. I think you have to use raw performance per inning as the base, then factor in the other variables of endurance, *'s, etc.

[cummings2]

very cool stuff maligned, thanks for the added clarifications. I think I now know where you are coming from. I'll munch on your comments and run some figures, seems to me, right now that your points are very much on track and that you've "cracked" some problems I've been facing especially when comparing relievers to starters.

As far as endurance goes: I agree that my approach is quite simplistic and because of that it's likely to overlook an important element or two. I settled for using the starter's endurance and simply multiplied times the max ammount of starts even though I am quite aware that pitchers often exceed that ammount of innings, best example in starters is Randy Johnson. I felt, however that given how HAL uses the pitchers, that at the end of the season the number of projected innings was a fair estimate.

Now, food for thought: since SOMs fatigue rules don't take into account SLG. My initial thinking was that once the pitcher reaches the POW then the ones with the lowest OAVG will have a greater chance of dodging fatigue, so in that sense OAVG would have a greater value, at least in SOM, as far as giving the pitcher a great chance of exceeding his projected innings. By this I don't mean that OPS looses value, I just mean that lower OAVG would be priced higher than at first sight (Oliver Perez & Clemens again)

As far as the value of Strike outs, I came to assuming a greater value for Ks after running some comparisons between pitchers. In particular Jose Lima and David Wells I agree with the OPS argument in those cases but if they were to have the same OPS but a much higher K:BF rate, my feeling is that their cards would be much more pricy, I guess that I should find a high K pitcher with similar OPS to Lima/Wells and then compare the salaries. If you find a pitcher that matches that criteria let me know :)

One of the problems I am personally having in strat is that there are some of the figures that go counter to my so-called-knowledge of baseball (quite old school with bunts and sacs and H&Rs, etc.), for example, I believe that a pitcher that walks batters is always asking for trouble, that walks will always come back to haunt you, so I thought the pricing of pitchers would reflect that, however I haven't seen that, actually it's almost as if a pitcher that walks a batter as opposed to give up a SI** is valued much higher in strat, and I agree and understand why, but still it's a bit counter-intuitive for me.

I agree with your comments and your example of Rick Bauer is a a very good one. We're on the same page.

Your thoughts and comments are much appreciated and welcomed. I'll run some figures on a few more pitchers still holding on to the way I've been doing so far, not because of being stubborn and petulant but because I am interested in comparing those results with the ones that factor in your line of thinking.

Good stuff maligned!

[maligned]

Cool. I'll be interested to see your figures. In your calculating, if you want to continue to factor in baserunners allowed because of how it could cause fatigue more quickly, don't forget that the strat fatigue rule is based on BB+Hits, not just hits. In the board game, once your pitcher has reached his POW inning, he becomes fatigued when he gives up any combination of three walks or hits in any one inning. Thus, WHIP would be the stat to look at in determining a potential fatigue risk. Roger Clemens: 1.16, Oliver Perez: 1.15.

David Wells' allowed OPS: .694
Jose Lima's allowed OPS: .778

I didn't find examples I liked with these two, and I'm out of time, so I can't show examples I did look at. However, it does seem like there's some merit to TSN's choosing to add bonus value for strikeout pitchers. I looked at a couple closely related pitchers (in terms of OPS and opponents played and ballparks played in). Strikeout pitchers were priced a small percentage higher in a couple examples.

Keep punchin' that calculator!