predict the next USA president (vote) - trump/clinton

[Philip_Mitterrand]

Hillary is more crooked than a dog's tail and her reign will only be marginally diff than the current feckless president. Trump is a much higher threat to wall street gangs than hillary under whom the inquities will only rise.

From the selfish working in finance point of view, Clinton will be a god send. Wall Street is going to flourish to new heights.

 

[diegosanaz]

Let's remember that the question here is who we think will be the next US president. Not who we want it to be...

http://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo

 

[vertigo]

Trump won. Well done, Donald. Rather, Mr President.

The people who predicted Clinton to win - don't even try to explain yourself. Your opinion can be thrown in the trash can - and you'll also find the Democratic party there given that the Clintons have now been purged from politics.

This is a quantitative finance forum. So let us rephrase this prediction question in terms of a mathematical finance question....

We must price a digital option (1 if Trump wins, 0 if Clinton wins) if the volatility itself is not constant and if the underlying path (Trump electoral votes, for example) is not continuous? Let V be the option price on a short dated digital option on the underlying variable S. We assume that S follows a Heston model with a jump component, say a compound Poisson. How do you estimate the jump size? The jump size should be approximately to the bias in the Democrat / Republican polling numbers (approx 6 p). Now price V. This (V) will tell you the probability of Trump winning.

The key feature to such a model is convexity. Vega convexity (how quickly your Vega moves as the vol moves) is captured and that is the key - suddenly a small change in the votes implies Trump's chances are much higher. With a high drift estimation, the Vega convexity will be skewed upwards. Nate Silver's model or any of these other econometric garbage models don't capture that. They never will, either. Pollsters don't understand statistics and econometrics don't understand stochastic. Academics don't understand either but write a lot about either.

If you did all of that ^^^ you will find Trump had a approx 70% of winning 1 week before the election, approx 90% 1 day before the election, effectively converging to 100%.

Remember: Never respect a forecasters's decision AFTER the outcome. If he is wrong, he is wrong and his opinion can be thrown in the trash can.

 

[pingu]

Trump won. Well done, Donald. Rather, Mr President.

The people who predicted Clinton to win - don't even try to explain yourself. Your opinion can be thrown in the trash can - and you'll also find the Democratic party there given that the Clintons have now been purged from politics.

This is a quantitative finance forum. So let us rephrase this prediction question in terms of a mathematical finance question....

We must price a digital option (1 if Trump wins, 0 if Clinton wins) if the volatility itself is not constant and if the underlying path (Trump electoral votes, for example) is not continuous? Let V be the option price on a short dated digital option on the underlying variable S. We assume that S follows a Heston model with a jump component, say a compound Poisson. How do you estimate the jump size? The jump size should be approximately to the bias in the Democrat / Republican polling numbers (approx 6 p). Now price V. This (V) will tell you the probability of Trump winning.

The key feature to such a model is convexity. Vega convexity (how quickly your Vega moves as the vol moves) is captured and that is the key - suddenly a small change in the votes implies Trump's chances are much higher. With a high drift estimation, the Vega convexity will be skewed upwards. Nate Silver's model or any of these other econometric garbage models don't capture that. They never will, either. Pollsters don't understand statistics and econometrics don't understand stochastic. Academics don't understand either but write a lot about either.

If you did all of that ^^^ you will find Trump had a approx 70% of winning 1 week before the election, approx 90% 1 day before the election, effectively converging to 100%.

Remember: Never respect a forecasters's decision AFTER the outcome. If he is wrong, he is wrong and his opinion can be thrown in the trash can.

This is all MMQB.