Before the Oscars Polymarket had a bunch of markets on who would win what categories. I had not seen a single one of the nominated movies, so most of these were quite uninteresting, but one market caught my eye, namely one on whether any movie would win six or more awards. At the time ‘No’ shares were sold for $0.54, implying a 54% chance that no movie would win six Oscars. I thought it was under priced, so I attempted to calculate the probability of one of the movies winning six or more awards.
To calculate this, I needed some probabilities of each movie winning each category it was nominated in, so I decided to use the bookmaker odds, which I got from OddsChecker.com, a site listing odds for all the major betting sites. Since I didn’t have a scraper for the site handy and they don’t allow for easy download, I went through the tedious process of copy-pasting the odds into a spreadsheet for each category. Since only five movies where nominated for six or more awards, I only kept the odds for these titles. I then calculated the mean of the odds from the different bookmakers, hoping they in aggregate would be closer to the true probabilities. This gave me the following table with probabilities: XS
| The Power of the Dog | Dune | Belfast | West Side Story | King Richard | |
|---|---|---|---|---|---|
| Best Picture | 0.469 | 0.021 | 0.073 | 0.029 | 0.026 |
| Best Actor | 0.143 | 0.000 | 0.000 | 0.000 | 0.745 |
| Best actor supporting | 0.270 | 0.000 | 0.044 | 0.000 | 0.000 |
| Best Director | 0.824 | 0.000 | 0.055 | 0.049 | 0.000 |
| Best Adapted screenplay | 0.392 | 0.042 | 0.000 | 0.000 | 0.000 |
| Best original screenplay | 0.000 | 0.000 | 0.308 | 0.000 | 0.036 |
| Best editing | 0.181 | 0.439 | 0.000 | 0.000 | 0.260 |
| Best Sound | 0.050 | 0.730 | 0.025 | 0.125 | 0.000 |
| Best Cinematography | 0.189 | 0.650 | 0.000 | 0.047 | 0.000 |
| Best Song | 0.000 | 0.000 | 0.035 | 0.000 | 0.104 |
| Best supporting actress | 0.114 | 0.000 | 0.031 | 0.784 | 0.045 |
| Best costume | 0.000 | 0.181 | 0.000 | 0.068 | 0.000 |
| Best original score | 0.171 | 0.678 | 0.000 | 0.000 | 0.000 |
| Best production design | 0.060 | 0.607 | 0.000 | 0.058 | 0.000 |
| Best visual effects | 0.000 | 0.794 | 0.000 | 0.000 | 0.000 |
| Best makeup | 0.000 | 0.141 | 0.000 | 0.000 | 0.000 |
I wrote a bit of Python code to calculate the probability of each movie winning six awards or more. The outermost for loop loops through each movie and the second loop computes the probability that a movie wins six or more awards by making a list of all possible combinations of the categories
import pandas as pd
import itertools
from operator import mul
from functools import reduce
df = pd.read_csv('movie_probs.csv')
categories = list(df.columns[1:])
MIN_WINS = 6
six_wins_probs = []
for i, movie in df.iterrows():
prob_sum = 0
for i in range(MIN_WINS, len(categories) + 1):
combinations = itertools.combinations(categories, i)
for comb in combinations:
win_category_probs = [(movie[cat])for cat in comb]
diff = set(categories) - set(comb)
lose_category_prob = [1-(movie[cat])for cat in diff ]
probs = win_category_probs + lose_category_prob
prob_sum += reduce(mul,(probs),1)
six_wins_probs.append(prob_sum)
titles = df["movie"]
for i, title in enumerate(titles):
print(title, six_win_probs[i])
print(sum(six_wins_probs))
| Title | Probability |
|---|---|
| The Power of the Dog | 0.0229 |
| Dune | 0.1655 |
| Belfast | 0.0000 |
| West Side Story | 0.0000 |
| King Richard | 0.0000 |
| Total | 0.1885 |
According to my calculations, Dune is most likely to win six awards with, 16.55% chance. Belfast, West Side Story and King Richard on the other hand had basically zero chance of winning six awards. Adding up the probabilities gives 18.85% chance of a movie winning six or more, which in turn is 81.15% chance of no movie winning six or more awards. That is much higher than the 54% chance implied by the price on the market, so I bought a small amount of ‘No’-shares.
However, as is now known, Dune ended up winning six awards, so I didn’t win the bet.
What went wrong? Well, 81% chance still means that the outcome will be unfavorable in one out of five instances, so maybe it was just bad luck. It is also possible, that the bookmakers set the odds of Dunes winning chances too low on some of the categories. Last, but not the least, it is entirely possible, that my calculations are wrong. Please let me know, if that is the case.
