Nate Silver Knows “Nothing”

  • Posted by
  • on November 10th, 2012

“The best model of a cat is another cat, or preferably the same cat.” Wiener/Rosenbleuth  via Nate Silver

I’ve mentioned Silver a few times in recent months, as he has a great ability to share facts and bring them to life. I’m a huge fan of his book The Signal and the Noise, a fun discussion of why some predictions in sports, weather, economics, and medicine have failed and others succeeded. He was the subject of both love and hate as the election approached, with(no surprise) Obama voters touting his forecast and Romney voters(and media pundits) bashing it.

As most know by now, he nailed the outcomes of all 50 states, focusing on his data and ignoring the headlines of the day. What’s not as well known is that he was not alone. As described in this BBC article, Silver was the best known of those aggregating polls, but others did the same with equal success. Money quote from Emory assistant professor and fellow aggregator Drew Linzer, “There’s no need to go on gut instincts or intuition or whatever else pundits are doing, when we have real information.” (HT to one of my favorite traders @debrink for the article)

Markets are full of information…is there a useful application of this approach? To an extent, I say Yes but with huge caveats. Let’s consider a few trading and investing approaches and the @fivethirtyeight equivalent:

1) Fundamentals- researching what the company is selling, how customers are responding to the product, measuring the needs and wants of the potential market. I forget, am I talking about companies or politicians? While this method can work, it often fails because the researcher cannot possibly know every factor that goes into a company’s(or politician’s) success, and in viewing it through their own lens can fail to see a competitive threat emerging from what seems to be a minor issue. Those in the trenches will always have an edge here, the market equivalent of knowing how a ballot initiative may impact turnout.

2) Sentiment- this is essentially what pollsters are measuring, the breadth(for whom are you likely to vote?) and depth(how likely are you to vote?) of voter intent. But it’s a huge error to translate this over to their opinion of $SPX; not only do people’s intentions change quite often, but each person polled has a different “vote”. In politics, a Wyoming conservative has the same impact as a New York liberal, getting 1 vote each. Sentiment polls give each person a single vote but take no consideration of portfolio size or trading frequency. If voter #2 runs a $10 billion fund, we’d certainly want to give his vote some extra weight, right? And back to the “intent” thing, voter intent and action might stick together but the mood and willpower of an investor can change on a dime. Unless the question is “Do you expect to buy/hold/sell this morning?”, and we weight the results based on assets to be deployed, we’re better off simply expecting that rising prices make investors confident and falling prices make them scared.

 3) Technicals- you knew I’d go here, but I’ll mention some obvious flaws. No matter its history, any one method can have success or failure. It may place too much emphasis on a bellwether that has served it well in the past. It may get caught up in its own sample and fail to include newer entrants. It may fail to include a large enough margin of error in its judgments. These are all flaws from established pollsters that can easily translate to market technicians.

This is where I stand alongside the aggregators. By a) not imposing their own biases, b) accepting new data as information, not a threat, and c) letting the consensus of outcomes dictate their judgment, they simultaneously let go of a need to judge every anecdote(noise) and end up with a more reliable forecast(signal).

This mix of humility towards what we can really know, and confidence that we can lean on the footprints of others, is much closer to the right recipe than anything being cooked up by media pundits. The aggregators have real information in the form of dozens of polls. We have real information in the form of price, why not use it?

And yes, I believe that observing the evolving footprints of active stocks is the closest we can get to measuring the true intent of the crowd…and that Nate Silver and his fellow aggregators would approve of this message.

The information in this blog post represents my own opinions and does not contain a recommendation for any particular security or investment. I or my affiliates may hold positions or other interests in securities mentioned in the Blog, please see my Disclaimer page for my full disclaimer.

blog comments powered by Disqus
Derek Hernquest Blog