clock menu more-arrow no yes mobile

Filed under:

Identifying Buy-Low Pitchers

Astros' Sabermetrics: As the Astros consider trading for a starting pitcher, can advanced stats identify potential targets?

Jeff Hanisch-USA TODAY Sports

It's no secret that the Astros are looking into the trade market with the July 31 deadline looming on the horizon.  And a starting pitcher is rumored as the position most likely to be targeted.  But this article isn't about  Hamels, Cueto, or Kazmir.  Instead, I'll start from the premise that the best strategy is to buy-low in the trade market.  Buying low means targeting a pitcher whose performance this year appears to be sub-par or below his normal level.  The idea is to reduce the trading cost, given the risk that the player's sub-par performance will continue after he is traded.

Assuredly, another element of the buy-low strategy is attempting to reduce the risk that the pitcher will continue to pitch with the same results which made him a buy-low option. Therefore, executing the strategy requires identification of pitchers who are most likely to improve their performance in the future.

The benefit of the buy-low strategy is two-fold.  First, it produces the most cost-effective value in the trade market.  Second, the number of appealing prospects who have to be included in trades is minimized.  The Astros are in a delicate position of improving this year's team without damaging the farm system and future teams too much.

I will use advanced statistics, available on Fangraphs, to show how a group of buy-low starting pitchers can be identified.

Frequently ERA is an unreliable indicator of pitchers' skill level or how well pitchers are performing.  The actual rate of runs allowed over less than half a season tends to fluctuate enough from year-to-year that it may not be predictive of the remainder of the season.   The ERA may not be consistent with the pitcher's underlying peripheral stats or skill level.  In that case, we expect the pitcher's future performance to regress in the direction revealed by the advanced stats.

Identifying Potential Buy-Low Pitching Trade Targets

A reasonable assumption is that buy-low candidates will have a current year ERA higher than their Fielding Independent Pitching (FIP) stats.  My initial screening of pitchers on Fangraphs' 2015 leaderboard is limited to starting pitchers with the biggest gap between ERA and a lower FIP.  I also removed teams which are less likely to be selling pitchers at the trade deadline because the teams appear to be in contention for a Wild Card slot.  At this stage of the season, the list of eligible teams is reduced considerably because so many teams believe they can compete for a wild card.  I also made some subjective calls which may be debatable.  For instance, I included pitchers from the Red Sox and Padres, although both teams are built to compete this year and may be reluctant to become sellers.  I have included both rental pitchers (i.e., pitchers who will be free agents after this season) and pitchers who will still be under team control in future years.

The pool of trade candidates, below, includes all starting pitchers with an ERA at least 0.38 runs higher than FIP (excluding pitchers on teams which are potential contenders).

Listed In Order of ERA Minus FIP

Rick Porcello Red Sox 4.61 1.47 4.02 3.91
Kyle Lohse Brewers 4.94 1.3 4.22 4.22
Jeremy Hellickson D Backs 4.36 1.02 4.03 4.02
Jeff Samardzija White Sox 3.64 0.91 3.83 3.77
Bartolo Colon Mets 3.69 0.86 3.63 3.62
Clay Buchholz Red Sox 2.66 0.82 3.18 3.21
Tyson Ross Padres 2.93 0.65 3.18 3.39
Matt Garza Brewers 4.99 0.53 4.24 4.42
Rubby de la Rosa D Backs 4.11 0.5 3.26 3.38
John Danks White Sox 4.88 0.5 4.53 4.48
Mike Fiers Brewers 3.68 0.46 3.8 3.65
David Phelps Marlins 3.81 0.38 4.15 4.27

The table, above, includes several defense independent pitching stats which may suggest whether the pitcher is likely to benefit from regression in the future.  FIP is based on the peripheral rates for strike outs, walks, and home runs; this metric shows the results which the pitcher had the most control over, and excludes much of the "noise" which influences ERA.   For most pitchers, home runs per fly ball will regress toward league average.  x-FIP is normalized to show FIP with a league average rate of HRs per fly ball.  SIERA is like x-FIP, but the pitching stat is a more complex formula for recognizing the pitcher's underlying skill during the period.  SIERA considers types of batted balls (e.g., ground balls) and recognizes pitching characteristics which tend to lead to weak batter contact or double plays.  The pitcher's current SIERA may give us a glimpse of the potential improveme nt in ERA over the remainder of the year.

Luck and Defense

We can also add advanced stats to the table which provide more detail regarding factors such as luck and team defense that may explain the variance between pitchers' ERA and FIP.  The table below show Fielding Dependent Pitcher-Wins (FDP-Wins), which explains the difference (in the units, "wins") between WAR based on actual runs allowed (RA9-Wins) and fangraphs WAR (f-WAR).  FDP-Wins is the sum of ball-in-play wins (BIP-Win) and left on base wins (LOB-Win).

BIP-Win is based on BABIP luck, which includes the effect of good or bad defense. A negative BIP-Win is a strong indicator of potential beneficial regression in BABIP.  LOB-Win reflects the impact due to the sequence of events.  For example, if a pitcher allows three walks and 1 HR in a game, the results will be quite different if the sequence of those events occurs in a single inning instead of spread across all innings.  A negative LOB-Win also suggests possible future regression toward a less harmful sequence of events.  This isn't to say that the pitcher has no control over BIP-Win or LOB-Win, but these are metrics which are more likely to regress.  If you want to examine how much negative FDP-Wins for a pitcher is derived from BIP-Win vs. LOB-Win, fangraphs has the breakdown here, or you can go to the value section of the player page.

The table below also provides a comparison of the FDP-Wins to Defensive Runs Saved (DRS) team defense rankings.  These rankings include DRS shift runs saved, and were tweeted by ESPN's Mark Simon.

Above Avg. Negative DRS Team

Soft Pct. FDP Wins Rank
Rick Porcello Red Sox
-0.8 22
Kyle Lohse Brewers
-1.2 23
Jeremy Hellickson D Backs
-0.4 4
Jeff Samardzija White Sox 18.6% -1.1 29
Bartolo Colon Mets
-1.4 16
Clay Buchholz Red Sox
-0.9 22
Tyson Ross Padres 19.8% -0.9 25
Matt Garza Brewers
-0.8 23
Rubby de la Rosa D Backs
-0.4 4
John Danks White Sox -1 29
Mike Fiers Brewers 19.6% -0.9 23
David Phelps Marlins
-0.3 10

Most of the selected buy-low pitchers played in front of sub-par team defense.  David Phelps (Marlins) and Jeremy Hellickson (D-Backs) were the only pitchers on the list who played for teams which were not below average on defense.  The White Sox and Padres pitchers may have been particularly affected by weak defense, since those two teams are among the worst overall defenders in baseball.  This suggests that previous pitching under performance may be due to the quality of defense.  More importantly, the Astros were ranked No. 1 on the same DRS defensive rankings; this suggests that the listed pitchers can improve simply by changing teams.

The Astros also like pitchers who induce soft contact.  The Fangraphs' batted ball section includes contact percentages separated into soft, medium, and hard contact.  Three of the listed pitchers induced a proportion of soft contact above the MLB average of 18.5%: Fiers, Ross, and Samardzija.  These three pitchers also experienced very high BABIP (in excess of .328).  Combined with the proportion of soft contact, this points to poor luck or poor defense affecting the high BABIP rate.


We don't know the trading cost that would be demanded for the players on this list.  Without that information, I can't be definitive in identifying the best targets among the players on this list.  The trading cost of long-term trades (trades for players under team control after this season) is likely to be significantly higher than for rental trades (trades for players who will be free agents after this season).  However, if the Astros really like a long term pitcher and believe he can contribute to future teams, they may be willing to pay the higher trading cost.


It's hard to predict whether the Red Sox will hit the reset button on this season and begin trading pitchers.  Porcello looks like a pitcher who can pitch well in the future.  He has the largest gap between ERA and FIP on this list, which makes him a good rebound candidate.  However, the Red Sox signed him to a pricey extension, and the contract cost is not favorable for a pitcher who is likely a No. 3 type pitcher.  Clay Buchholz could be great future piece for the Astros' rotation.  But I have my doubts that the required trade return would be commensurate with a buy-low situation.  Buchholz has a good ERA and he is underperforming only relative to the Red Sox expectation that he is a top of the rotation pitcher.

Among the long term trade possibilities, Tyson Ross would be my favorite (depending on the cost, of course).  He induces soft contact, has been hurt by terrible defense, and carries an elite ground ball rate.  He would be the right hand version of Keuchel.  He is relatively young, which makes him an attractive addition.  However, his ERA has already begun to regress in a positive fashion, reducing the potential for a buy-low discount.  That said, I include the Padres on this list because their front office is difficult to predict.  They have shown the willingness to take risks in the trade market.  Undoubtedly the trade return might be painful, but it might be worth it.

Mike Fiers, with the Brewers, might be a less costly candidate.  He is a pitcher in his late 20's who hasn't even reached arb eligibility yet.  He has an impressive K rate and induces soft contact.  He had been hit hard by the BABIP bug, which makes him an excellent candidate for BABIP regression.  The Brewers likely will be sellers, but it's unclear how aggressively they plan to rebuild.


Presumably the rental pitchers would produce less pain for the farm system.  Jeff Samardzija probably is the best rental candidate.   He has been hurt by his BABIP and a bad defensive team.  But SIERA says he has a pitched like a 3.77 ERA pitcher.  The question is whether he would be too expensive in trade return for a rental.

Another possibility is for the Astros to look in the bargain bin, which presumably would require less trade return.  I like Kyle Lohse as a candidate to produce in the same manner we expected from Feldman.  And Lohse has experience as a premium pitcher in the past, which should help down the stretch.  Given regression possibilities, it's possible that the Astros catch lightning in a bottle.

The same thing could be said for Bartolo Colon, but maybe even more favorably.  SIERA gives him a nice 3.62 skill level ERA.  I included Colon because the Mets seemingly have an oversupply of pitching talent, and probably would be willing to move him.  The Astros can hope that Colon would provide the same beneficial impact that he provided the A's a couple of years ago.

What are your thoughts for trade candidates based on the list?