Evidently, our auRthur’s statistical department utilized a model that had some flaws – “past performance is not necessarily a predictor of future returns”. This is because the harvesting of the turkey is a “rare event.” Rare (unprecedented) events are difficult to predict. The story is not terribly amusing to turkeys to begin with – but becomes less amusing to humans when understood as a metaphor of the financial meltdown and statistical modeling in use by banking institutions. Essentially, banking institutions assumed a huge amount of risk because a catastrophic meltdown was simply outside the realm of consideration. It was not represented in most of the models in use.
A great and vivid illustration. See Nassim Nicholas Taleb’s essay where this chart and illustration originally appeared at edge.org. This article discusses the limits of statistical thinking and is a good springboard to other writings by Taleb – who was a practitioner of risk as he ran a hedge fund for a number of years and saw many of the practices in the financial industry up close and personal.
The chart above was created using R and ggplot2. The data frame named DF was populated with data related to the Turkey Welfare Index.
TWI Day color
1 14 1 black
2 15 2 black
3 16 3 black
4 17 4 black
5 18 5 black
6 19 6 black
7 20 7 black
8 -100 8 red
UPADTE: This can be entered in a few different ways. One is through a grid (which requires that you specify the Day as a factor).
Plotted using ggplot2:
ggplot(data=DF, aes(x=Day, y=TWI, fill=color)) +
scale_fill_manual(value= c(“black”, “red”)) +
theme_bw() + scale_x_discrete(breaks = NA) +
title=’Turkey Welfare Index’)