Last week I promised to speak a little about how to win more points by taking a dynamic approach to forecasting, making frequent and multiple forecasts on the same question. The principle is that you can book a profits on a forecast without waiting for a question to settle in your favour. I’ve written about the underlying processes here and here, and now I’m going to focus more on strategy.
Below is the one-month forecasting history of our Trump for President forecast market – one of our more popular questions. Trump is without question a divisive figure in society and you can see from the chart below that there have been a number of big forecasts away from the central trend.
One way to make points is to play the “reversion to mean” game. This is to assume that the central trend is probably accurate and unlikely to move far over the short-term. If you see the market move sharply away from the mean*, normally driven by a forecaster who had particularly strong views one way or the other, simply nudge it back towards trend. When (if!) the market falls back into line, or even moves past it, take a profit.
This is fairly low risk stuff, as long as you don’t keep your eye completely off the ball. The upshot is that you can bank points in a matter of hours or days, rather than wait months for a question to settle. You can then re-invest these points on other questions in a more traditional manner.
Another tactic is to be quick with the news. If, say, you were to see a big poll had just been released showing Trump ahead you could adjust your forecast to reflect all of this new information. Or you could move the market a little in that direction and wait for others to get hold of the news and move it further. That would again leave you in a position to take profit.
If you want to be first with the news, Google Alerts is a handy tool. Twitter can also be helpful if you can find a way to plough through the swathes of rubbish in search of valuable information.
If you want to be even smarter, you can anticipate the news and make points by guessing where the market might get pushed next. It is, for instance, fairly well understood that US presidential candidates tend to get big poll boosts after their respective national conventions. This proved true again this time around. Looking at the chart below, do you reckon you could have made a profit timing your forecasts to predict movements in the crowd forecast?