Betting on elections in the Sixteenth Century
It started with neighborhood gossip in Rome. The most common form of betting was called maschio et femina (masculine or feminine) - betting on the sex of unborn children of a specific pregnant woman in the neighborhood. Unfortunately, when the game proved popular, the brokers themselves got in the act. They either faked a nonhuman birth or declared that a dead hermaphrodite was born.
Then, the wagers moved to the papal election (scommesse a fare il papa). Speculation usually began even before the pope died (normally, a pope has to die before another one gets elected). When princes and cardinals prepared for a possible conclave, the city began to speculate and place bets on possible successors to the papal throne. Once the pope had died, speculation grew as the cardinals prepared for the conclave by coming to Rome. The punters got information about the cardinals who were papabile - usually from the servants at the Vatican. The punters knew about each cardinal’s career and political alignments.
Dateline: Banchi, Rome
The brokers who took the wagers were wealthy merchants, typically cloth merchants, spice dealers, apothecaries, and bankers that had their offices in the financial district called the Banchi in Rome.
Why Banchi?
(1) It was strategically located: From the quarter of Ponte stretched from the Florentine national church of San Giovanni dei Fiorentini to the Old Mint (Zecchia Vecchia) near the bridge leading to the Vatican.
(2) It was adjacent to the quarter of Parione where most of the newsletter writers and the printers had their offices and shops.
(3) This was where the Piazza di Pasquino is located - Romans congregated around Pasquino to hear the latest news, read invectives mocking the cardinals and gossip.
(4) Most importantly, the Banchi was close to the Ghetto where many Jewish pawnbrokers and moneylenders lived and plied their trade. These Jewish merchants and dealers actively served as brokers.
Habemus Papam - not Papadam (We have a Pope)
In his bull of 21 March 1591, Cogit nos, the Pope Gregory XIV forbade (under pain of excommunication) all betting concerning the election of a Pope, the duration of a pontificate, or the creation of new cardinals. There must have been enough betting in each category for the Pope to specifically name them.
There were some arrests of guilty parties but it was clear that some of the cardinals themselves (probably through their employees) were surreptitiously placing bets on papability. Since the cardinals themselves were meting out the law, they did not go too strict to enforce it.
Executive summary: Cogit nos eventually ended formal gambling on papal elections. [It did little to illegal gambling on it.]
Even today, the UK government allows gambling on political outcomes. But, it takes a dim view about trading on insider information. On June 20, 2024, one of British Prime Minister Rishi Sunak's police bodyguards was arrested over alleged bets on the date of Britain's national election made before it was announced. A fuller treatment of the elections in the UK, see
https://tapen.substack.com/p/betting-on-not-so-great-britain
To understand the importance of papal electoral gambling, we need to realize that the Popes had a lot of political and economic power in Catholic countries like France, Spain and Italy (and England, Scotland and Ireland). Catholic churches owned vast tracts of land, ran businesses and funded armies.
Fast Forward to the Twenty First Century India
In May 2024, the (informal) speculative markets of betting on Indian Parliamentary elections became news. They disappear and reappear every election cycle. The markets are illegal - they work on the trust of the traders in each other. In the past centuries, the contracts were agreed upon in the narrow alleys of big cities in India - in particular in Bombay, the Gateway of India. Today, they have gone online. The big traders operate using encrypted WhatsApp messages. The offers appear and disappear in real time. For a descriptive analysis of that market, see
https://tapen.substack.com/p/election-betting-in-india-satta-bazar
The markets: There were ten active markets for this election cycle. Phalodi is reputedly the biggest one running into trades of hundreds of millions of dollars. Since all of this is illegal trade, such figures have to be taken with a sack of salt.
The biggest trading days for these markets was the last week of May.
Table 1
Several facts emerge from these results:
(1) Not all betting markets produce the same results. There is considerable regional variation. They bear no relation to the voting tendencies of their geographic locations. [If you torture the data enough, you might find the markets in the BJP dominated states have overestimated the strength of the BJP.]
(2) If we examine the errors of the betting markets and compare them with the exit poll results, the size of the errors are much smaller for the betting markets: minus six percent and minus ten percent compared with minus forty percent and twenty eight percent for the exit polls.
(3) Almost all the betting markets overestimated the results for the Congress party and underestimated the results for the BJP.
Table 2: Betting market errors
(4) The results here are exactly the opposite of what happened in the exit poll results where all the results overestimated the strength of the BJP and allies and underestimated the strength of the Congress party and allies.
Table 3: Errors in Exit Polls
While there were independent candidates (and some independent candidates like Engineer Rashid won against big name Omar Abdullah), seats must have a nearly one for one trade-off between NDA and INDIA. Suppose, hypothetically, there were no independent candidate, then NDA versus INDIA would tradeoff would look like this:
If there were no errors, the relation should be linear with R = -1, R^2 = 1. A loss for BJP+ is almost surely a win for Congress+. NDA and INDIA contested in almost all the seats. So, unless “the others” is a significant force (it never is), the correlation of the errors between NDA and INDIA results should be very high.
How do the real life results look?
For the exit poll markets, R^2 turns out to be 0.2227. Statistically it is much closer to zero than to -1.
For the betting markets, things look very different.
Executive result: The relation is much tighter - the correlation is now statistically not different from -1. However, the errors do not run through (0,0).
Suppose now, for a moment, all the errors are uncorrelated in Table 2.
We should then get an equal number of positive and negative numbers. Out of 40 numbers, only nine are negative. What are the chances if the odds are even to get such a result? We are looking at a Binomial distribution with n=40 and p=0.5. The probability of nine or less negatives is one in a thousand - a very unlikely chance event.
However, the numbers in the four columns are not independent because of political ties. It shows up in the correlation between Congress and the BJP with R=-0.89. If predictions are good, a high negative R is expected. But the correlation between INDIA-Congress and NDA-BJP shows R=+0.46 - the wrong sign.
Executive conclusion: If all these markets were frictionless and information was free flowing, there was a possibility of a cross market arbitrage (assuming each market was deep enough). However, in real life, with problems of the legality of these markets themselves, such sophisticated arbitrage would be too much to expect.
Executive thought: Unlike the Exit Poll forecasts, the punters in the betting market have skin in the game - wrong bets lead to direct financial losses. So, their incentives are better aligned to the outcomes of the elections.
[Most punters do not bet on a single outcome. They hedge. For example, a saddle-maker, Gaspar Romano, arrested for betting on the papal election in 1590, admitted, “I bought two tickets for Cardinal Santi Quattro for four scudi and one for Cardinal Castagna for three scudi, which I then sold for twenty-two scudi.” Cardinal Castagna was elected as Pope Urban VII.]
Executive postscript: Which politician won the biggest prize in the parliamentary election in India in 2024?
The biggest winner is the (returned) Chief Minister of Andhra Pradesh, Mr N Chandrababu Naidu. His family owned (dairy) business has doubled in value right after the election far outstripping the general stock market rise netting his family a cool billion dollars.
Executive outcome: More Ferraris for his son’s collection of cars.