WAIBI DISSECTION

ERA & WHIP: The peak of the Iceberg

This article refers among others:

• Introduction to basic principles of WAIBI SABERMETRIC SYSTEM’s formulas.
• Algorithmic and conceptual analysis of ERA. Updating formula step by step to PTRA.
• Many inconsistencies at Rules Scoring greatly distort the veracity of ERA.
• Ghost Runs & Circumstantial Runs, the best allies of ineffective pitchers.
• Algorithmic and conceptual analysis of WHIP. Updating formula step by step to PSIP.
• Semintentional Bases for Ball & Strike Out with Wild Pitch, "ghost" and "circumstantial" data respectively, in the calculation of WHIP.
• ERA & WHIP don’t supplemented. Comparative tables of the top 20 pitchers in MLB 2014 according to ERA, WHIP and PSIP.

One of biggest problems in Baseball stats, is that some traditional parameters (widely used yet) were designed to evaluate something, that its algorithmic structure fails to convey. Even though sometimes these leave out key variables, making the result confused and incomplete, its "practical simple use" and easily understood  by managers, coaches and fans, "legitimize" them as highly reliable. However, its conceptual weakness, VERY OFTEN shows many capable players that are not actually so capable, and others very efficients as if they weren’t.

This is the case, among others, of ERA and WHIP, two very useful  tools for decades, but now archaic to measure the pitchers effectiveness. Formerly there were more differences between  teams quality, and these inaccuracies didn’t take much weight. However, after the internationalization of strategies, tactics, systems and technical skills training of the best competitive level schools (mainly US and Japan), the quality’s contrast between players around the world has decreased greatly. Therefore, it is increasingly important to make a very careful  selection to aspire to victory. 

 ERA

Although many important sabermetric statheads have developed more functional variations of this formula, looking as close as possible to the truth, I consider that  the implementation of  ERA has more value as a general historical data, than as a selection’s tool. Its invention dates back to the late nineteenth century and is attributed to the English writer and historian Henry Chadwick, the founding father of Baseball statistics, who also created, among others, the first scoresheet, the scorebox, and AVG formula. In the times that we mean, there was not a statistical background in this sport, and Chadwick was inspired by the Cricket’s game stats, in which certain events are far away from Baseball complexities.

Without Henry Chadwick would not exist Statistics, in the same way that without Alexander Cartwright, the Rules of Baseball (1845). If these two ballgame’s columns (rules and statistics) had been developed conceptually linked, there would be more thorough methods for selecting the best players. However, most of the rules were added at the same time that the game itself was showing its diversity, while Stats was most focused on data processing, not questioning the rules, which could help to deepen them each other, by vital feedback. It is possible that because of its traditional use, ERA variations have been developed by analyzing the basic equation, and not for reason it was created. In doing so, we found a substantial discrepancy between the theoretical approach and its practical application. So, let’s  analyze the root of ERA, the original formula:

First, Chadwick had the great idea of eliminating unearned runs, to evaluate only those that are a pitcher’s responsibility.

Where ER (Earned Runs) are the Runs allowed JUST BY THE PITCHER !!!!

So, careers in relation to the pitcher, are classified in two types: EARNED and NOT EARNED, where every earned run IS a pitcher’s responsibility, every unearned run IS NOT a pitcher responsibility .

According to the Rules of Baseball Book:

• (10.12)The official scorer shall charge an error against any fielder:  a) whose misplay (fumble, muff or wild throw) prolongs the time at bat of a batter,prolongs the presence on the bases of a runner or permits a runner to advance one or more bases, unless, in the judgment of the official scorer, such fielder deliberately permits a foul fly to fall safe with a runner on third base before twoare out in order that the runner on third shall not score after the catch.
• (10.16) An earned run is a run for which a pitcher is held accountable. In determining earned runs, the official scorer shall reconstruct the inning without the errors (which exclude catcher’s interference) and passed balls, giving the benefit of the doubt always to the pitcher in determining which bases would have been reached by runners had there been errorless play. For the purpose of determining earned runs, an intentional base on balls, regardless of the circumstances, shall be construed in exactly the same manner as any other base on balls.

However, let’s analyze  the following cases, which show overwhelming and interesting inconsistencies: 

       IN THE SCORING RULES

• In a Perfect Game-ending play, a pitcher is winning one by zero. To tie the game with a homerun, the opposing team chooses a dangerous substitute hitter that had acchieved 4 HR, 1 triple and 1 double in 6 at bats this season, facing this pitcher. The manager (NOT the pitcher) determines to give him an Intentional Base on Balls (IBB), even if this action will break down the Perfect Game, to attempt to finish the game with the next batter, who has been struck out three times in the game. However, other two runners reach bases by fielder's choice, having had both real out possibilities. So, now bases are loaded, and the pitcher is achieving a No Hit No Run game (even though  IBB wasn’t his personal tactical decision). Then he makes a strange movement that injures him, so he is replaced. The pitcher who relieves him, allows a homerun and the opposing team scores 4 runs as Home Club. The starter loses the Perfect Game, No Hit No Run , and the game, and further, earns 3 runs without responsibility. So, his ERA grow up from 0.00 to 3.00, even though he had technically performance  a 28 outs perfect game. The pitcher is highly penalized having the most positive action.
• If as a variation of the same situation above, we assume that the game is tied at zero, and the offensive team is Home Club , the homerun is enough to win the game one by zero, I mean, the result would’ve been the same regardless the number of men on bases (even without someone). However, the run that gives victory to rival, "belongs" to the pitcher who threw the technical Perfect Game, and the truly guilty (who allows the homerun) even lost the game. Again the pitcher is penalized without commiting faults.
• In a possible inning-ending play, a runner scores by error. According to the scoring system, all runs scored after the error, until the end of the inning, will be considered “unearned”. The situation could be partially understandable, if always there were at least one runner on bases, because the pitcher will have to adjust his tactics to avoid the runner’s advance. But, when he faces a batter without runners on bases, it is technically the same situation  he would’ve faced starting the  following inning, but having the additional advantage that there are two outs. Thus, the situation is even more favorable to the pitcher, because he will throw for just one out, not three. The defensive formation also will be the same. Therefore, every runs after the empty bases,  are in practic just responsibility of the pitcher, until another error occurs. In this case the pitcher is rewarded even if he commits negative actions.  Interestingly he’s  only  exonerated in the ERA record, and the other parameters (hits, BB, etc) are normally counted. This behavior is typical of the "Ghost Runs", I’ll iexplain it later.
• When runners advance by wild pitch or passed ball, there are not  errors, so, those actions produce earned runs. However, the passed ball is just a catcher responsibility, as well as a great part of the wild pitches by low throws (The catcher knows the throw in advance, so, he should be positioned correctly to block the ball, not to try to capture it ) . The passed ball is counted like error, only if the runner scores because of it, but it isn’t if only he advances to the second or third base. The reason for this is that these two positions (pitcher and catcher) have a lot more shares than the rest. The problem is when the rules release the catcher of responsibility in advancing runners, we charge it to the pitcher by default, who made a perfectly catchable throw. Thus, pitcher is penalized having made a positive action.

      

        GHOST RUNS

• In a possible inning-ending play, the hitter-runner is safe by error. The next hitter produces a homerun. According to the scoring rules, both are unearned runs. It is understandable when we consider the error, but the homerun is a direct allowed run, because the pitcher and the batter are the only ones involved. This is what Waibi Sabermetrics System have called "Ghost Run", because statistically, this occur "without responsability of any player", and simultaneously these are neither earned nor unearded. Let’s recall that, according to the Baseball Rules Book, each run must be in force, one of those classifications. In the scoring rules there are some peculiarities that, although they’re not mathematically correct at all, are applied fairly and ingeniously,  because every action must be JUSTIFIED, and be responsibility of someone. (like when a pitcher who hasn’t a finished pitch to a batter is relieved). But in practice, there are several situations such as the example, in which the variable in analysis literally DISAPPEAR in the system. To be understood, the homerun counts as a RBI, as hit, and modifies all offensive parameters, but it never will exist in ERA count,  as if it hadn’t been hit against someone. Returning to the first example, let’s suppose a pitcher with a low concentration level, who becomes angry because of an error of the defense, at Perfect Game-ending play, BEING THE HOME CLUB team. Under the current scoring rules, the thrower could allow homeruns of each one of the next batters, (10, 20, 100 ... 1000 or more) but his ERA will remain in 0.00. Here the pitcher isn’t penalized even though commits SUCCESSIVE negative actions. 

 

        CIRCUMSTANTIAL RUNS

Pitchers with best teams defense receive fewer earned runs.

• In sacry fly plays, when the outfielders throw the ball slightly open, or the catcher cannot keep the ball because of a wild bown, it’s not considered an error because logically, the runner will not advance to another base, because home is the last one. However is not responsibility of the pitcher. Therefore, teams with the best outfilders (rifle throwers & ball hawks ) and he best defensive catchers avoid more runs. So, this is another case in which ERA will be modified independently of the pitcher’s performance. These are valued by Waibi Sabermetrics System as "Circumstantial Runs".
• Other "Circumstantial Runs" occurs depending on the strategy and technical-tactical discipline of the defense. For example, the fielder’s choice is a situation where the batter has been dominated, but kept safe by a tactical defense error, which isn’t registered as such.    Meanwhile, is contradictorily  analized, by default, as a pitcher’s responsability (earned run) if the runner scores. Therefore,  the teams with best tactical discipline prevent more runs. The pitcher is penalized again having made a positive action.
• When the runner reaches the next base on catcher responsibility, without an error, such as : a) when the catcher throwing arrives in time  to put out the stealer runner, but the shot is open, b)  the pitcher infielder looses the ball when he is trying to touch the runner; c) worse yet, when pitcher surprise a stealer, and he arrives safely to the next base after a rundown play, without a defense error. Those are another technical unregistered errors, because it will be scored as an error only if it causes that runners advance to the next base. It will not be a scored error even though it has been on defense’s responsabily. The runner on third, scores by sacry flay, and this counts as an earned run. Therefore, teams with players with greater technical-tactical virtues prevent more runs.   Here the pitcher is penalized without commiting faults, or worst yet,  having made a positive action.
• When there is a hit caused by defense’s mental errors, as when there is a collision, or on the contrary, the players leave to each other the responsibility on capturing the ball, and it falls in fair territory. If those runners score, the runs are counted like earned. Here the pitcher is penalized without being guilty.
• Another interesting example occurs in “squeeze”, “hit and run” and others small ball plays. All of these ones, are produced by disarticulating the defense. Defense and HeadQuarter (NOT the pitcher) most decode his oponente strategy, and then the catcher will quarterback the defense. These are plays designed  in  laboratory,  and impeding their success doesn’t depend on the pitcher, but on the team’s "intelligence". The thrower must be focused on maintaining the runners near the base, and dominate the hitter. So, teams with a better "intelligence" prevent more runs. Here the pitcher is penalized without being responsable.

Pitchers from teams with better "team work",  and players with better technical skills, receive fewer "earned runs".

• When with runners on first and third, and one out, the Manager decides, after evaluating the score: a) close the infield to avoid the run, b) put the infield to "halfway" to search for the double play at second base (if it fails, the opponent will score at least a run) c) play with the infield in standard position when there is much advantage, ignoring the runner to third, where the batter just need to put the ball in play to score. This positioning of the defense is a Manager tactical decision that affects drastically the rival scoring possibilities. However, in all these situations, the absence of error, those runs are charge them to the pitcher register, even if he mastered the batter, and the defense has been surprised. Here the pitcher is penalized without being responsible .

 

GHOST CIRCUMSTANTIAL RUNS

This is a special (and alarming) case, where runs could be classified as earned, or "ghost", in the same practical situations:

• Situation 1: In a possible inning-ending play, a player is still safe by error. The next ten batters connect hits, producing eight runs.
• Situation 2: In a possible inning-ending play, a player is still safe by error. (as in Situation 1). The pitcher is replaced. The next ten batters connect producing hits eight runs (as in Scenario 1).

As clearly seen, the ONLY DIFFERENCE between both situation is the pitcher replacement. Beyond that, it's exactly the same. However, the rule says that in Situation 1 ALL runs are unearned, while in the Situation 2,  error’s run is the ONLY unearned. The same data, in almost  similar situations, are processed with completely contrary rules. The problem  is when corrent stats, evaluate both for selections, use them as if they had been processed equally, increasing the margin  of error in direct proportion to "ghost runs" difference between them. Thus, we have in those examples, a pitcher with a 7.00 ERA, that actually had allowed the same runs than other with 0.00, because of the conceptual weakness of this scoring rule.

There is a considerable higher quantity of plays where the concept of earned runs, freak-outs or helps pitcher records. So, using these to measure throwers quality isn’t  precise at all, but very confusing. Also, another problem that increases the margin of error, is the subjectivity of the oficially scorer, which neither isn’t the same in each game.

 Then, what’s expresses the so overrated ERA?

As we’ve seen clearly, defense’s behavior has almost as much impact as pitcher’s performance. Thus, ERA estimates the number of earned runs (taking the Book Rules current concept) allowed by a SPECIFIC team when the analized pitcher plays. Anyway, it’s still vague because of the same problems we remarked above. ERA depends not only on pitcher’s quality, but also on his team players quality, and on their team-work, being less reliable, when bigger are the differences  between the teams. Usually,  as AMATEUR, A and AA leagues, chemistry between players (on a personal level), becomes an important variable that affects this parameter. Besides, if ERA definitely relates team & pitcher performances, it wouldn’t make sense to remove the unearned runs. It's definitely just a global defense stat (that includes the pitcher), and it has rather an illustrative than selective value, to evaluate the yield OF A GROUP under specific circumstances (when the pitcher in question is acting) and NOT of a particular player (the pitcher). I mean, it’s a COLLECTIVE performance stat.

Being PTRA ( Pitcher & Team Run Average ) the allowed runs average for an specific team when a particular pitcher throws;  R are all runs, earned or not.

Waibi Sabermetric System fully discards this important traditional tool, and measures differently the actions related with runs on pitchers responsibility (no earnead & unearned concept). Although it obviates some very specific variables yet, Waibi achieves to reduce greatly, and even eliminates, several of these defects. However, we need to add a simple data to scoresheets to implement it on the third level, which is its current development phase. This one greatly improves the accuracy of the data interpretation  respectly to level two, which it’s what we’ll use becouse of the current MLB database.

A formula that spills light on this point, is the FIP (Fielding Independent Pitching), created by Tom Tango after "Vörös" McCracken’s DIPS (Defense Independent Pitching Statics). The FIP test  pitchers, taking into account only those actions that depend exclusively on them: HR, BB-IBB, HBP and K. This formula will be analyzed later. I’ll just say for now, that these kind of actions are generally a very small group of doings on mostly pitchers performances, and the proportion in which they appear respectly to the balls in play, is sometimes significantly less, and the outcome may depends more of pitcher physical characteristics than his actual effectiveness. However, FIP is revolutionary, and has, at the same time, a solid conceptual reasoning. 

WHIP

This is another widely used stat to determine the quality of a pitcher, and even some analysts give it more value than to ERA. Usually, these stats are analized together because of the false perception that they supplement each other. WHIP was created in 1979 by the writer Daniel Okrent, to calculate how many on based players the pitchers allowed per inning:

  

Something really incomprehensible regarding the history and the development of this formula, is that Okrent decided to exclude the Hit By Pitch parameter (HBP) ONLY because at that time, this data wasn’t published regularly. However, several years after HBP was formally and officially booked in stats, it hasn’t been added to the formula, which is still used without this important variable, increasing their margin of error. On the other hand, I don’t know if  thirty-five years ago, when Okrent created WHIP, Intentional Base on Balls (IBB) was a registered data. Anyway, this one must be extracted for quality stats processing. Otherwise it's just a general performance stat.

I think is very important to emphasize on IBB analysing, because rarely it is a pitcher decision. Anyway, in a case like that, he’ll  receive mandatorily the manager confirmation, so, automatically, this action will not longer be on his responsability. Furthermore, it is the only special case on Baseball, in which the play’s success, is almost one hundred percent guaranteed, and besides,  having the peculiarity that there isn’t, with very few exceptions, a real confrontation hurler-hitter that could influence the outcome. The IBB is clearly a manager decision, similar in practice (conceptually) to position the infielders for a double play, but  executed by the pitcher.

However, the walks have always an ignored detail, that Waibi Sabermetric System called  "Circumstantial Ghost Data": the Semi-intentional Base on Balls (SBB). Why? Although it is also a manager decision, THERE IS a real confrontation between pitcher and batter, where the hitter doesn’t know what the pitcher will beat him to try to put him out. This play is comanded when the situation demands to walk a batter, but he doesn´t have good discipline at home plate, or is very aggresive with low contact percentage. If ,after some throws, the pitcher deception fails and has an unfavorable count, he will throw as many free balls as needed to reach four, converting the base in “intentional”. As there is a confrontation (although partial), it should be counted to evaluate the pitcher, because depending on his quality, he’ll achieve to cheat the batter, or not. In addition, many at-bats are decided in the first pitches, which is the objective here. As  in both situations these are scored as equal intentional walks, it will include a bigger margin of error in data processing. This is another rules scoring problem in quality and selection stats.

Another detail is when a Strike Out with Wild Pitch (SOWP) occurs. However, here the scorer must consider if to charge the responsability on pitcher or catcher, for the reasons we’ve explained when we spoke about  ERA , and depending on that, add it to the formula.

Thus, the WHIP reviewed formula would be more complete if measures all Direct Negative Actions (DNA) per innings pitched:

Although Waibi Sabermetric System used this WHIP variant in its initial analysis, it must be discarded to search for more accurated methods. The reason is that even with this reviewed algorithm, there is still false perception because: it doesn’t reveal WHEN a pitcher has quality, but WHEN he HASN’T¡¡¡

Let me explain: A pitcher with a very high WHIP, will never be a good hurler, because he allows many opponents arrive to bases. If he has a low ERA, usually responds to an excellent global defense of his team, or the absence of powerfull hitters in the opposing team, or even the luck, among others already discussed in ERA chapter. In other words, his low ERA is due to an OPPORTUNISTIC PERFORMANCE, not to his actual effectiveness. Otherwise, a low WHIP doesn’t give us information about the pitcher capacity for spacing the rivals positive actions, the detail which defines his capacity to avoid runs: A pitcher could maintain a low WHIP, even if each allowed hit was a home run.

Therefore, in a second development phase, I tried to differentiate the pitchers according to the specific hits type allowed, linking DNAIP with the general concept of Sluging. Thus, pitchers who allow fewer extra-bases hits would have a lower average. Taking that:

Being TB Total Bases traveled.  I used 1B to refer “one base hits”, to differentiate it from H, which we use to identify “hits”  in general.  If WHIP means Walk and Hits per Innings Pitched, I’ll call this variation PSIP: Pitcher Sluging per Innings Pitched:

This setting allows us to evaluate a pitcher effectiveness much more better than the traditional WHIP, because it tells us not only how many men have been on base on pitcher responsability, but, most important, how many bases his opposites stroke per inning in direct offensive actions (those that have been executed only from the batters box). A pitcher that allows many extra-bases, have a much higher PSIP, which greatly clarifies the choice, because PSIP error is notably lower than using ERA and traditional WHIP together or separately.

However, Waibi Sabermetric System, in its third level (fourth level is already under analysis) developed a several times more complex version, because PSIP still leaves some variables that add error to the structure: a pitcher with equal PSIP than another, can allow many more runs.

Let me explain: Two pitchers faced each other, allowing nine hits in 9 innings. All hits were doubles, but the difference is that the hurler A allowed 1 hit in each inning, while the hurler B, allowed them all in the same entry. Thus, the pitcher A is scoreless, while the B, earned  8 runs. As noted, there is still a margin of error too large to compare crearly two players, which according to stats, had equal performances. Even so, PSIP is still several times more reliable than ERA & WHIP combination, as we have discussed up to this point.

These problems are solved by Waibi Sabermetric System’s algorithms.

 

Why ERA & WHIP don’t complement?

TO COMPLEMENT means going together, to help to define which pitcher is better than another. However, after analyzing the roots of their algorithms, we find that WHIP is more illustrative: a good pitcher can have either a high or low ERA, but he’ll never have a high WHIP. In ERA analysis, we saw numerous situations where inconsistencies were interchangeably scored in favor or against the pitcher record, and many "false" data are officially registered, which prevent this supplementation. Therefore, I consider that to use them to select players is a demonstrable mistake. Its use, as I said, has an historical value, as most of the PERFORMANCE statistics, which we’ll analyze.

Let's look at some data from 2014 Major League Baseball season:

Using ERA, we obtain the following season top 20 board. This players would be called to a preselection team for an hypothetical short tournament. Let’s imagine a Super World Leagues Cup, which faced MLB’s best players team, with its Japanese Professional League counterpart:

No.	LANZADOR	TEAM	ERA
1	Clayton Kershaw	LAD	1.77
2	Felix Hernandez	SEA	2.14
3	Chris Sale	CHW	2.17
4	Johnny Cueto	CIN	2.25
5	Adam Wainwright	STL	2.38
6	Doug Fister	WSH	2.41
7	Corey Kluber	CLE	2.44
8	Jon Lester	BOS/OAK	2.46
9	Cole Hamels	PHI	2.46
10	Garrett Richards	LAA	2.61
11	Henderson Alvarez	MIA	2.65
12	Jordan Zimmermann	WSH	2.66
13	Zack Greinke	LAD	2.71
14	Lance Lynn	STL	2.74
15	Alex Wood	ATL	2.78
16	Tyson Ross	SD	2.81
17	Tanner Roark	WSH	2.85
18	Alex Cobb	TB	2.87
19	Julio Teheran	ATL	2.89
20	Dallas Keuchel	HOU	2.93

In WHIP assessing, we see that many of our ERA top 20, not only change position in the board, but even disappear from our list. Worse yet, some really great effective pitchers, according to WHIP, are not even in  ERA top 40:

No.	Rk ERA	LANZADOR	TEAM	WHIP
1	1	Clayton Kershaw	LAD	0.86
2	2	Felix Hernandez	SEA	0.92
3	4	Johnny Cueto	CIN	0.96
4	3	Chris Sale	CHW	0.97
5	5	Adam Wainwright	STL	1.03
6	10	Garrett Richards	LAA	1.04
7	41	Hisashi Iwakuma	SEA	1.05
8	12	Jordan Zimmermann	WSH	1.07
9	22	Jeff Samardzija	OAK/CHC	1.07
10	6	Doug Fister	WSH	1.08
11	19	Julio Teheran	ATL	1.08
12	29	David Price	DET/TB	1.08
13	7	Corey Kluber	CLE	1.09
14	17	Tanner Roark	WSH	1.09
15	21	Madison Bumgarner	SF	1.09
16	8	Jon Lester	BOS/OAK	1.10
17	25	Stephen Strasburg	WSH	1.12
18	39	Jason Hammel	OAK/CHC	1.12
19	38	Josh Collmenter	ARI	1.13
20	42	Phil Hughes	MIN	1.13

 

As I explained in this article, the PSIP formula, even with its flaws, is highly fuller than ERA & WHIP, (together or not), and it shows marked  inconsistencies in  traditional analysis:

No.	Rk ERA	Rk WHIP	LANZADOR	TEAM	PSIP
1	1	1	Clayton Kershaw	LAD	1,20
2	10	6	Garrett Richards	LAA	1,30
3	2	2	Felix Hernandez	SEA	1,31
4	5	5	Adam Wainwright	STL	1,38
5	3	4	Chris Sale	CHW	1,39
6	4	3	Johnny Cueto	CIN	1,45
7	12	8	Jordan Zimmermann	WSH	1,52
8	7	13	Corey Kluber	CLE	1,54
9	18	22	Alex Cobb	TB	1,55
10	17	14	Tanner Roark	WSH	1,55
11	41	7	Hisashi Iwakuma	SEA	1,56
12	8	16	Jon Lester	BOS/OAK	1,57
13	19	11	Julio Teheran	ATL	1,57
14	24	33	Sonny Gray	OAK	1,57
15	20	28	Dallas Keuchel	HOU	1,59
16	9	24	Cole Hamels	PHI	1,59
17	22	9	Jeff Samardzija	OAK/CHC	1,61
18	6	10	Doug Fister	WSH	1,61
19	21	15	Madison Bumgarner	SF	1,61
20	16	34	Tyson Ross	SD	1,61

I've marked the players according to the following legend: 

	LEYENDA
	Coinciden los datos
	Lanzador SOBREVALORADO
	Lanzador SUBVALORADO
	Datos No Determinantes

For these reasons, as I’ve explained, Waibi Sabermetric System, developed a very more accurate variant, with a SYSTEMIC structure, particularly affecting, EACH REGISTERED player shares.

Waibi Sabermetrics System website, coming soon.