Friday, April 11, 2008

How Accurate are Bloggers III

In my first article for the new Zignals blog I will follow up on two earlier pieces I wrote in May 2007 and September 2007 on the TickerSense Blogger poll. The TickerSense Blogger Sentiment Poll has had its fair share of criticism in the past, but it is still the best, single source gauge of blogger opinions on the market. It's somewhat unfortunate to see participation drop to about 20 bloggers, but it does at least capture some of the bullish to bearish opinion out in the blogosphere.

The Poll is based on a 30-day outlook for the S&P; bullish, bearish, or neutral. I assume a 'neutral' call to be less than a 1% gain or loss, with breaks on either side representing a bullish or bearish return.

I used the Monday open following the Poll mailing (usually over the weekend) as the starting price, matched with the closing price 1-month later. Where the difference between 1-month open and closing price was less than 1% the reading was considered 'Neutral'.

For the test period October 1st to the end of February the S&P recorded 16 periods of losses, 2 neutral periods and only 4 periods when the market closed higher a month later. This was in sharp contrast to the April to August analysis which had 5 bearish, 5 neutral and 10 bullish periods, and the prior December to April analysis which had 4 bearish, 4 neutral and 12 bullish periods. Will this sharp change in market behavior have changed any of the blogger outlook's during this period?

I excluded Blogger who returned fewer than five polls, so this meant saying goodbye to In the Money, and Daily Blog Watch. Incidentally, both of these bloggers got their single call correct.

Of the remaining bloggers, four were steadfast in their opinions and returned one unwavering call throughout the polling period. These were Carl Futia (bullish), Learning Curve (bullish), Millionaire Now (neutral), and Random Roger (bearish). Carl and Roger have been consistent over the past year in been 100% bullish or bearish.

The most accurate blogger from October to March was Slope of Hope at 80% correct, closely followed by Random Roger at 70%. The majority of bloggers had 40% or fewer of their responses correct.

Sixteen bloggers returned polls across my three test periods. The law of averages kicked in so the spread was relatively low between the least accurate blogger (Millionaire Now) at 14% correct responses, and Daily Dose of Optimism the most at 46%. When variance is considered, Quant Investor does well with 42% accuracy on variance of 2.7%. My responses returned a 37% accuracy on 2.0% variance, with Peridot Capitalist and Information Arbitrage the most consistent with 0.2% and 0.6% variance respectively.

Did bloggers change their views over time?

The last chart assigns bearish (max -100) to bullish (+100) values according to the sum of their returns. For example, a blogger returning a bearish call every time will score -100. A blogger returning half bullish and half bearish calls will have a score of 0, and all bullish calls will score 100. Highlighted boxed represent a bullish (>50) or bearish (<50). Red or green text reflect a bullish or bearish change from the previous call. At the very bottom is the performance for the S&P during this period. It is clear in the most recent analysis that the S&P has switched strongly bearish (but had been in decline throughout 2007).

The biggest about turn came from Learning Curve who went from strongly bearish to strongly bullish (unfortunately, at the wrong time!). Others, like Daily Dose of Optimism stayed bullish throughout, but within the optimism was a gradual bearish drift. Quant Investor and I jumped sharply on to the bullish bandwagon after two periods of neutral outlook. Millionaire Now and HedgeFolios turned bullish, but did so from bearish levels to leave their current outlook as neutral. Both Information Arbitrage and Elliot Wave Lives On turned more bearish as early bullish outlooks dropped to neutral.

If you look at the period April to September 2007 there were an equal number of strong bulls and bears (prior to the sell off). Currently strong bulls outnumber strong bears by over three to one.

From the three test periods there are bloggers who are more consistently more accurate than others? High accuracy by Carl Futia and Daily Dose of Optimism is offset by the variance of their calls. Quant Investor leads the pack with a low variance and decent accuracy during a bull and bear market. Will it continue?

Dr. Declan Fallon, Senior Market Technician, the free stock alerts, market alerts, and stock charts website


muckdog said...

Yes, I do horrible in that survey. LOL. Thanks for doing the research and crunching the numbers. I do enjoy reading about how folks are doing.

And like all fortune tellers, I'll try to do better with my 30-day window guessing...


The Learning Curve

Declan ( said...

I find the 30-day (well 22-day really) outlook such a tough one. Often enough nothing changes!

Yeah, there are the big up days, but these are usually followed by a number of big down days. So by the time the month is out we are left exhausted and relatively unchanged!

I think it is interesting to see how our perceptions change. I hadn't realised I was so bullish in the most recent analysis having flipped around in neutral territory for much of the year.

Kudos to all who contribute - it's an interesting study.