In last week’s article, I estimated a dollar value of adding Roy Halladay for each of the contending teams mentioned in trade rumors related to Doc as of last week. Using an approximation of the effect on playoff odds and subsequent success in the playoffs, I found that Roy Halladay’s contract was worth nearly $15 million more to a contender than he is to the Blue Jays. However, there are many other factors that play a role in revenue that vary by team, and one factor is that Blue Jays fans love Roy Halladay. Many of those fans cannot stand the thought of losing him, and the Jays certainly may wonder if subtracting him would hurt their attendance enough to deem it unwise. The concept of a franchise player having a high sentimental value to a city is one worth exploring, so let’s discuss the effect of Roy Halladay on the Blue Jays’ attendance.
For a comparison, back in the late 1990s Curt Schilling played for a series of bad Philadelphia teams. However, Schilling was very good. As a result, the Phillies won 55 percent of the games that he pitched from 1996 until he was traded in 2000, against only 42 percent of the games that he did not pitch. As a young Phillies fan who hated going to games that his team lost, I developed a strategy of picking the games that I attended based on the pitching matchups, and as a result I went to about every fifth game-and the Phillies would win about 55 percent of the games that I went to. Throughout 2002-2009, the Blue Jays have won 64 percent of the games that Roy Halladay has pitched and only 46 percent of the games that he has not pitched. This led me to ask the question: Is it possible that there are a lot of Blue Jays fans doing exactly what I did in my pointedly aiming my attendance at Schilling’s starts? Are Blue Jays fans going to Halladay’s starts more often?
The reason that this is useful to study is that looking at attendance by starting pitcher is a way to gain some insight into the true value of a losing team’s franchise player. It would be difficult to study the cost of what would have happened if the Orioles had traded Cal Ripken to a contender during his career, because he played in every game. The question of how many Orioles fans came out only because they would get to see Ripken is a very difficult one to answer. However, as starting pitchers’ schedules are well known, Blue Jays fans may be going to games only to see Halladay.
Initially, I tried to simply compare the attendance in Halladay’s starts to the rest of the Blue Jays’ games. During the average Halladay start between 2002 and July 21, 2009, the attendance at the Skydome/Rogers Centre was 27,918; when Halladay did not start, it was 26,426, a difference of 1,492 seats sold. If we stopped the analysis there and concluded that these fans would simply stop coming if Halladay did not pitch, then with about 17 home starts per year and about $30 per ticket, then this would cost them about $761,000 per season. Given last week’s adjustment to Nate Silver‘s approximation of the value of a win excluding the impact on playoff odds would have put Halladay’s value at $6.75 million, then an additional $761,000 in value of people coming to see Halladay’s games is what I’d call ‘nontrivial.’ While some of that attendance factor may already be included in the value of a win, that still seems like a pretty large effect to ignore.
However, if we just focus on how the Blue Jays sell tickets in games that Halladay pitches and in games that he does not, then we ignore a lot of other significant factors in team attendance. For instance, Halladay pitches more often on Sundays and the Blue Jays get more attendance on Sundays than other days of the week. Halladay pitches more often against the Yankees too, and those are often hot tickets in Toronto. So I decided to run a regression analysis to consider the effect of Halladay pitching as well as other effects on attendance for the Blue Jays. I included dummy variables to account for year and month, as well as several other factors: whether Halladay pitched, whether the game was on a Sunday, whether the game was on a weekend, whether the game was against the Yankees, whether the game was against the Red Sox, and-another huge factor-whether the game happened on Opening Day. The inclusion of the home opener was important; the Blue Jays’ attendance in home openers over the last eight years is 49,342, while their attendance in other games is 26,551. As Halladay pitched in five of the last eight Opening Days, not including the impact of Opening Day will overstate Halladay’s effect on attendance.
Here are the effects on attendance of each of those variables (excluding the years and months) without including the home opener variable:
Halladay: 1,873 Sunday: 2,397 Other weekend: 1,918 Vs. Yankees: 10,231 Vs. Red Sox: 5,472
Here are the effects on attendance of each of those variables (excluding the years and months) with the regression properly specified:
Halladay: 758 Sunday: 2,901 Other weekend: 2,146 Vs. Yankees: 10,278 Vs. Red Sox: 5,359 Home Opener: 28,466
Each of the non-Halladay variables was statistically significant, as were all of the year and most of the month dummy variables. The one exception was whether Halladay pitched. Although there seems to be a slight positive effect, it is not even statistically significant and is less than half of the original estimate that did not account for these factors. Clearly ignoring the home opener effect is wrong, and doing so will lead us to overstate the effect of Halladay pitching by 150 percent. It turns out that instead of $761,000, the dollar value of the extra attendance in Halladay’s starts is around $386,000. What this means is that Halladay is probably not doing all that much to increase the Blue Jays’ attendance beyond winning ballgames.
This reinforces the idea that the goal of player acquisition should be to win baseball games and not build around hometown favorites. Even a Toronto legend like Roy Halladay is not helping their attendance very much, and it becomes even more imperative that Riccardi trade him. The dollar value of the offers being reported by various teams are very close to the value of Halladay to a contender, which is around $25-28 million. The dollar value of Halladay to the Blue Jays is clearly about half that much, even after adjusting the attendance effects as specified in this article.
Of course, this is just one pitcher’s effect on one team, and it may be that in other circumstances, the value of a franchise player is higher than this. There is also the much larger issue of price’s effect on attendance which frequently gets ignored in these studies, which I did not get into in this article. Nonetheless, the qualitative lesson that we can learn from this analysis is that franchise players do not have anywhere near the effect on revenue as the standings do. This reinforces what I said last week-if Riccardi does not move Halladay now, he’s costing the Blue Jays a lot of money and a lot of wins.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now
I think a great follow up article could be to analyze how attendance changed for teams in the past after they traded their best player/face of the franchise and "gave up" on the season. I realize the definition of such player is fairly subjective but I'm sure you could come up with a few comparables.
The follow up article combined with your original article might paint a more complete picture. Thank you.
A few comparable stature players from recent years that might merit comparative consideration: CC Sabathia ('08), Mark McGwire ('97), Randy Johnson ('98).
Keep up the good work.
Drew points out the Phillies' trading Abreu, which is interesting but would confuse the analysis because the Phillies saw Hamels and Howard click at once right after they traded Abreu, and so their attendance shot up in August and September. Including Abreu would make it look like the white flag was a ticket seller!
As a displaced Torontonian, I was a tad disappointed that attendance was only 24k according to the boxscore (I thought the TV broadcast had said 21k). Jays fans could only muster barely above average attendance in what could be Halladay's final home appearance as a Jay.
What would be useful, however, is seeing how (and if) attendance declines for non-competitive teams -- particularly those that make dump/rebuilding trades -- after the trade deadline.
Imperative? I like the strong language Matt. A little part of me thinks that you are hoping Riccardi reads your article and trades Halladay to the Phillies.
Maybe if the Jays are in a pennant race in september all the bandwagon fans will come.
It also might be interesting to see how Holliday's attendance impact compares to a bobblehead or fireworks night.
Extra Yankees and Red Sox attendance, especially on summer weekends, includes lots of tourists who support the other team and don't really care who starts. The real fans enjoy watching Halladay and you hear lots of "wish he was on our team" comments, but it's not why they showed up. I'm not sure this matters to the analysis, but it's interesting that on Yanks/RedSox game days in downtown Toronto, you might wonder what city you're in.
A couple of things you might also try: Stub hub prices on pitchers day. Thats good for marginal value fans add on those days, tough to extrapolate towards total value, but could give some evidence(or more death notice) to the franchise player.
Also, can you publish your methodology of valuation for making the playoffs and an individual win? If we're basing all of this on the revenue implications of marginal wins, it might be trickier to extrapolate additional revenue on a day by day basis beyond that. After all, in a macro sense, the doc's marginal revenue value to the jays already includes the revenue effects of those days when the team looks more likely to win, on what is probably only a evenly prorated basis. This stems from the multiplication of his warp total and the price per win.
Is there any conceivable way to find out how teams revenue changes on a year to year basis with performance? Regress revenue with win percentage, post season gear and look for collinearity? Is the implication of all these value per win figures that with enough 95 win seasons the mets would take over as the revenue champions of new york? If that is not so is there a seamy underbelly to all this analysis? If adding a win to go from 88 on the season to 89 can't statistically significantly gain you x million dollars, what is all this advanced analysis worth?
The methodology of valuation for making the playoffs is in last week's article which is linked above. If you need more details, please let me know and I'm happy to explain. In general, I checked the PECOTA playoff odds which projected an expected winning percentage and odds of making the playoffs, and I figured out about how much more likely the team would be to make the playoff if the (binomial) distribution if wins shifted up by 2.6. I also adjusted the playoff series effect of him replacing a replacement level pitcher (which isn't far off for most teams). As for the dollar value of each of those things, I approximated based on Nate Silver's numbers from his 2005 article but with some inflation in the values to make it match up with the commonly accepted price per win. I also cooked up some reasonable approximations for the value of winning each series-- changing these values didn't alter the results much so I just picked numbers that worked.
The way to determine revenue changes with respect to performance is very, very tricky. Check out the comments section on last week's article and you'll see a really interesting exchange between dpowell & ben solow about this very topic. There are a whole host of issues involved. Obviously, it's worth looking into but for these two articles, I just approximated what Nate Silver had used with some logic and some inflation.
Great article, btw.
In reality, pretty much all teams probably do variable pricing. They just don't call it that. Most teams send out two-for-one deals by email, and throw in giveaways sometimes too, all to effectively change the price of games.
He looked at a number of pitchers but only for 2009 data. He found that the effect of Halladay was much larger-- about 3,000, as compared with my 758. This was because I looked at 2002-2009 total.
Both methods have their drawbacks-- clearly more recent data is more relevant but the effects for one or two seasons are not going to be statistically significant (and so the 3000 and 758 are not statistically different from each other).
Cameron also found that Halladay had the largest effect of any pitcher in the big leagues. This brings home the point-- it's about the standings!
Just looking at 2009, Halladay's numbers don't seem to stand out except for that one game against Burnett and a May 17th game against the White Sox... there are some 20k and 15k attendance games in there that Halladay threw in.
Year Avg Attendance W/Ryan
1989 25,234 29,287
1990 25,096 31,586
1991 28,367 31,752
Not sure if season ticket numbers are readily available, however...