Flashcards in Trends and forecasting Deck (39):
What is the problem with logical positivism in regards to prediction a.t. Aligica?
That positivism seeks to explain, not predict, but by tying them too close together the legitimacy of the methods derived from the two is undermined.
What is the aim of the Aligica text?
To help create a firm epistemological ground for Futures Studies
What is the difference between an explanation and a prediction?
If the argument precedes P, then we have a prediction; if it succeeds P, we have an explanation.
Explaining P is simply “predicting” P after it has already happened, and predicting P is simply “explaining” P before it has already happened.
The Hempel Model / covering law
What does the Hempel model / covering law explain?
The symmetry between explanations and predictions
What is the main critique of the Hempel model?
That there exists explanations that lack “predictive power”, and that a more thorough study of the relationship would have shown it to be more complex
What affects predictions despite the past a.t. Aligica?
Background information - e.g bus ticket prices.
Local knowledge and tacit knowledge is therefore important
What is a retrodiction?
Establishing something based on a known general law (?)
What is the main point of Wolfers and Zitzewitz? 2004
They show that market-generated forecasts are typically fairly accurate, and that they outperform most moderately sophisticated benchmarks.
Calls these prediction markets
What is prediction markets?
Markets where participants trade in contracts whose payoff depends on unknown future events.
If it is efficient, the market price will be the best predictor of an event
What is the assumption behind a prediction market?
That the market is motivated by rational traders
Which three types of prediction markets does Wolfers and Zitzewitz talk about?
Winner takes all
How should one assess the accuracy of prediction markets?
By comparing them to individual analysts - not poll-based forecasting
What kind of events is the prediction market best to predict?
Middle-probability, since psycological studies suggest that people tend to overvalue small prob events and undervalue near certainties
What behavioral bias are there at a prediction market?
People over- and undervalue, and they trade according to their desires rather than objective probability assessments
What was the DARPA market?
The prediction market put up by the US Defense Ministry in 2003.
Pros and cons for DARPA?
Created fear that terrorists would profit from kidnappings and acts of terror. On the other hand some argued that the terrorists would then reveal their intentions in how they traded.
When will a prediction market work most effectively?
When the subject is discussed and has a high entertainment value.
There is not sufficient data to say whether it makes a difference to use real money or not.
What is prediction?
It is about foretelling something, it deals with indicating what the future will be (which scenario forecasting does not)
They can be both categorial (this happens) and conditional (it happens if that happens).
Can predictions be falsified or verified?
Temporally universal theses can never be decisively verified, whereas temporally particularized theses can never be falsified
What does it mean to ascribe a probability to a future event?
It does not answer a predictive question but instead characterizes the status of such a prediction
What is a forecast?
It is a definite prediction concerned with specific and concrete events
What is a pseudoprediction?
A prediction that is empty of any descriptive bearing, saying something we already know. E.g. It will be Christmas again next year
What is a metaprediction?
A prediction about a prediction e.g. whether a prediction is correct
What does Rescher mean when he talk about a feedback phenomena on predictions?
Predicting something can influence people to do or not do something, e.g. traffic jams
Which three preconditions does Rescher put up for rational predictions?
Data availability, pattern discernability and pattern stability
What is the Delphi method?
It is a way of aggregating expert predictions, while avoiding that they influence each other.
It is a judgmental method (as opposed to formal)
What have the delphi method been criticized for?
That experts are afraid of ruining their reputation, hence they answer conventionally, which then creates a 'false' consensus. It further assumes that there is only one future and the problem is to identify it
Which formal methods does Rescher talk about?
1. Trend projection: linear extrapolation (unrealistic)
2. Curve fitting: extrapolation does not need to be linear (is about finding a function 'close to')
3. Cyclical analysis: problem to say where we are on the cycle
4. Circumstancial analogy: often used when placing something in a statistical reference group. E.g. giving that behavior and health he most likely will die before 70
What can indicators be used for?
If you have two factors that are closely linked, a predictive indicator presupposes the operation of causal linkages
What is scientific prediction?
Law-based: Our most sophisticated predictive method is that of inference from formalized laws
What is the problem of choosing a predictive method a.t. Rescher?
Justifying the employment of a specific predictive method will always itself involve the familiar inductive move from a past pattern to a future expectation
Is it difficult to predict human actions?
To some extent yes, since they have freedom of choice. On the other hand they are controlled by their own preferences, which can to some degree be predicted
What does the stock market predict a.t. Rescher?
The present economic attitudes rather than the future economic reality
Do other social sciences predict?
Yes, however difficult to predict in sociology as it is about changes in patterns and difficult in political science as well (e.g. elections). Demographic prediction is very developed.
Can history predict a.t. Rescher?
Not really, since e.g. small interactions have no implications in the short run but do in the long run. However history provides us with an understanding of possible outcomes, and places us in a better position to act with intelligence
What is the punctuated chaos doctrine?
That there is period of local stability in history where it is possible to predict
What is a usual regression mistake?
To assume that if something extreme happens, it will happen again (e.g. on the stock market)
Furthermore when one talks about regression towards the mean, it is a problem that we don't know what the mean is or if it changes
What is required to make the substantial jump from probability to prediction?
Contextual knowledge, personal expertise etc. Not enough with statistics