Monday, October 31, 2016

Which are the Top 3 Argentine tango orchestras?

TL;DR answer for three ways of understanding the question, based on the statistics of "Danceable Tangos of the Year," a weekly programming of :

(A) Whose Argentine tango orchestras recorded the largest number of danceable tangos?
1. Juan D'Arienzo
2. Francisco Canaro
3. dead heat between Aníbal Troilo, Carlos Di Sarli, and Rodolfo Biagi.

(B) Whose Argentine tango orchestras recorded the largest share of danceable tangos (compared to all of the given orchestra's own recordings)?
1. Rodolfo Biagi
2. Pedro Laurenz
3. Carlos Di Sarli

(C) Whose Argentine tango orchestras recorded the largest number of "best" danceable tangos?
1. Juan D'Arienzo
2. dead heat between Aníbal Troilo, Carlos Di Sarli, and Osvaldo Pugliese
3. Francisco Canaro

These answers are, as with all questions of taste, ultimately subjective. What may render them somewhat interesting is that they are based on a statistics whose collection was prima facie independent of the desire to produce such "top 3" lists. Below I explain the details and some of the significant qualifications.

Background info (skip this section to go directly to the stats). The ranking is based on recordings selected for "Danceable Tangos of the Year", a weakly programming which airs every Monday and Wednesday on Argentine Tango Radio (listen: ; podcast version: Danceable Tangos of the Year Podcasts ). The first episode, covering the years 1927-1928, aired in the beginning of January 2016 and the series will conclude with its 52nd episode, covering the years 1975-2016, at the end of this year. Each episode runs for roughly 55 minutes and hence could only feature up to 17-20 songs. This limitation on length forced selectivity, that is, only a portion of danceable tangos recorded in a given year were included. Since not all years spawned the same number of danceable tangos, I grouped some years together in a single episode and I also spent several episodes covering a single year, with the overall aim of keeping the portion of selected tangos to all danceable tangos roughly the same for each year. The selectivity and pre-assigned grouping are major potential sources of bias; looking back to how the process worked out, I'm mostly comfortable with how I pre-grouped the years together in episodes, with some exceptions.

Here is a distribution chart for how it all worked out in the end:

On the selection procedure: for each episode I took the time (!) to listen to most of the tangos, valses, and milongas I have in my musical collection for the relevant years. (Note: I have near-complete discographies for most of the major orchestras and fairly extensive collections for other orchestras that are played in milongas, festivals, and marathons). I then narrowed down the list in several successive stages. During this process I focused on one primary and three secondary qualities. The primary quality was danceability in a regular milonga environment, at least how I understand that after many years of dancing and DJing. The secondary qualities were the quality of the recordings; variety; and familiarity. Thus I decided to skip some great songs that I only have in quality that does not bode well with radio listening; I sometimes included songs to feature an orchestra or a tendency which would otherwise not be included; and I skipped great danceable songs that I know to be unfamiliar to most dancers (which in turn alters danceability for them). Thus these secondary qualities induce deviation from the primer goal of danceability, however I feel that, with the possible exception of quality of the available recordings, they didn't alter the results more than 20%-25%. In the end I needed to make a painful judgement call anyway, since for most years there were many more danceable tangos that I'd wanted to have included than I could due to length limitations. Finally, for each episode I tried to choose the "best" - most loved, recognized, innovative, or unique - song to start the episode with. This latter is probably the most subjective choice of all.

It should be clear from the selection procedure but let me emphasize it explicitly: the numbers and ratios below do not reflect solely on the numbers and ratios of danceable tangos of a given orchestra, but also on the competition with other tango orchestras that recorded in the same period. Hence they can at best be taken as lower estimates of the absolute numbers and ratios of danceable tangos of a given orchestra (with respect to my subjective criterion of danceability).

Detailed stats. Overall 921 songs got selected for "Danceable Tangos of the Year", 32% of which are instrumental and 68% features a singer. By genre: 78% tango, 14% vals, 7% milonga, and less than 1% other genre; I collectively (and admittedly loosely) refer to all of these as "tangos" for the stats. Note that I write "whose" orchestras as opposed to "which" orchestras since I'm lumping together different orchestras which had the same orchestra leader (notably the orchestra(s) of Francisco Canaro with Quinteto Don Pancho and Quinteto Pirincho, as well as the sexteto and the orchestra of Carlos Di Sarli; OTV and Adolfo Carabelli were separately counted). With this remark 68 different orchestras were covered overall, of which only 18 orchestras had at least 10 works that made it to the mix. I'll display the detailed all stats for these 18 orchestras only (omitted but close runner-ups include Ángel D’Agostino, Ricardo Malerba, and Adolfo Carabelli).

With these in mind, let's get to the details.

(A) Whose Argentine tango orchestras recorded the largest number of danceable tangos?

Well, here are the numbers, clustered into six groups:

GroupOrchestra##/921Total works#/Total works
Group 1:Juan D'Arienzo15517%106015%
Group 2:Francisco Canaro10511%37993%
Group 3:Rodolfo Biagi718%20235%
 Carlos Di Sarli697%41817%
 Aníbal Troilo687%53113%
Group 4:Edgardo Donato505%36514%
 Osvaldo Pugliese425%30514%
 Enrique Rodríguez404%35311%
 Alfredo De Angelis374%33811%
 Miguel Caló323%3848%
Group 5:Osvaldo Fresedo283%3847%
 Ricardo Tanturi253%18014%
 Francisco Lomuto253%6044%
 Pedro Laurenz222%8028%
Group 6:Roberto Firpo182%28701%
 Lucio Demare121%8814%
 Federico Domingo101%1626%

Juan D'Arienzo is king (and not just that of the rhythm), Francisco Canaro is second, and Carlos Di Sarli, Aníbal Troilo, and Rodolfo Biagi comes third.

Notes: the total number of works is gathered mostly from discographies by Bernhard Gehberger, Christoph Lanner, and Gabriel Valiente. If I needed to pick two orchestras that I felt were most unfairly underrepresented (given the selection criteria that I set to myself above), one would be that of Francisco Canaro (!) and the other would be that of Carlos Di Sarli. For Canaro, the number of good danceable tangos were simply too much compared with the frequency that his works are played in milongas; also, many of the omitted works are good, but not great. For Di Sarli I mostly wanted to avoid over-repetition of his instrumental works, many of which he recorded over and over again, in at least three major waves.

(B) Whose Argentine tango orchestras recorded the largest share of danceable tangos (compared to all of the given orchestra's own recordings)?

The numbers are already displayed in the previous table, but let me cluster them here separately:

GroupOrchestra#/Total works
Group 1Rodolfo Biagi35%
 Pedro Laurenz28%
Group 2Carlos Di Sarli17%
 Juan D'Arienzo15%
 Osvaldo Pugliese14%
 Ricardo Tanturi14%
 Edgardo Donato14%
 Lucio Demare14%
 Aníbal Troilo13%
Group 3Enrique Rodríguez11%
 Alfredo De Angelis11%
Group 4Miguel Caló8%
 Osvaldo Fresedo7%
 Federico Domingo6%
Group 5OTV5%
 Francisco Lomuto4%
 Francisco Canaro3%
Group 6Roberto Firpo1%

Rodolfo Biagi finishes on the top: more than one third of his total number of works got selected - and remember, these numbers can be taken at best as lower estimates! Pedro Laurenz comes as second - this came as a bit of a surprise, but I was not fully aware of the fact that he only recorded 80-or-so tangos overall. It may be pure coincidence, but I found it fascinating that the 1-in-7 ratio holds for most of the other great tango orchestras: Carlos Di Sarli, Juan D'Arienzo, Osvaldo Pugliese, Aníbal Troilo, Ricardo Tanturi, Edgardo Donato, and Lucio Demare. As if all of their Sunday recordings turned out to be amazing! Carlos Di Sarli is bit ahead of this pack, especially given the underrepresentation that I remarked above.

Note that while Francisco Canaro came out second in terms of the absolute number of selected danceable tangos, he is only second-to-last in terms of the ratio of selected danceable tangos to his total number of works, at a mere 3% - no big surprise if we remind ourselves that he made almost 3800 recordings overall! I'd also want to note that I have less than fourth of Federico Domingo's 162 recordings; among those that I know of, he came in pretty strongly, and so I wouldn't be surprised of him moving up significantly on this list if I had the chance to listen to all of his recordings.

(C) Whose Argentine tango orchestras recorded the largest number of "best" danceable tangos?

As explained above, this is clearly the most subjective of all categories, based on "first songs" that I chose for starting the episodes. Here we go:

Group 1Juan D'Arienzo815%
Group 2Aníbal Troilo510%
 Carlos Di Sarli510%
 Osvaldo Pugliese510%
 Francisco Canaro48%
Group 3Miguel Caló36%
 Edgardo Donato36%
Group 4Osvaldo Fresedo24%
 Alfredo De Angelis24%
 Roberto Firpo24%
Group 5Ricardo Tanturi12%
 Lucio Demare12%
Group 6Rodolfo Biagi00%
 Enrique Rodríguez00%
 Francisco Lomuto00%
 Pedro Laurenz00%
 Federico Domingo00%
Other orchestras 815%

Juan D'Arienzo is the first again, followed by a dead heat between Aníbal Troilo, Carlos Di Sarli, and Osvaldo Pugliese. So, unintentionally, the great four tango orchestras, compulsorily mentioned in the preface of every book on tango, came out on top. Francisco Canaro is a close third, and Miguel Caló, Orquesta Típica Victor, and Edgardo Donato are also close runner-ups.

Note: no songs from Rodolfo Biagi made it to this list, despite him finishing strongly in the previous two categories. This may be by pure chance, but I indeed tend to feel that while he has the largest share of "A" recordings, in terms of "A+", he is lagging behind other orchestras. I do play Biagi in most of my milongas, but I don't think of him as being as unavoidable as D'Arienzo, Troilo, or Di Sarli. I might change my mind when Japanese releases of Biagi finally surface; I refuse to believe that the thin-sounding available recordings are best representation of how his orchestra could have sounded.

Again, this is just a play with numbers and tastes - hope that I was not the only one having fun with it. I'd be curious of similar rankings by other DJs and dancers, I'd love to hear back from you! Alternatively, let the shit-storm begin - Caló and Fresedo fans, nail me to the cross!


  1. I am a tanguero and a statistician (Full Professor of Statistics at the University of Palermo - Italy). So, you can imagine how happy I am to see the two things I love in life linked together! Thanks for the very interesting work !!!

    Gianfranco Lovison

    1. Thanks!

      In connection with a discussion you had: as I wrote above, the episodes covered 68 different orchestras. This included that of Ángel D'Agostino: 8 recordings with Vargas, mostly from 1941, plus the later Café Dominguez, which is one of my favorites - in fact it is the only tango song to which I paid a separate long blogpost here on this blog. But I also explicitly emphasized, "the numbers and ratios below do not reflect solely on the numbers and ratios of danceable tangos of a given orchestra, but also on the competition with other tango orchestras that recorded in the same period." 1941-1942 saw many-many great orchestras and pieces, and in competition with these I felt that other D'Agostino-Vargas recordings are not to be preferred. In other words, had I included more D'Agostino-Vargas, it would have needed to be done in the expense of other songs. Tastes of course differ, but my challenge to you is to say which song would you have rather excluded from our coverage of, say, 1941, in the expense of which additional D'Agostino-Vargas song that I omitted.

      I should add that the ratios above also don't reflect how frequently these songs could and/or should be played in milongas. It may be that one orchestra has fewer number of great danceable pieces, but these may be more frequently played than pieces from another orchestra that has a larger number of great danceable pieces (i.e. that of Francisco Canaro).

  2. "so what joke will Balázs next crack about Canaro" :) an avid listener of the series