This note was originally published
at 8am on March 15, 2012.
INVESTOR and RISK MANAGER SUBSCRIBERS have access to the EARLY LOOK
(published by 8am every trading day)
and PORTFOLIO IDEAS in real-time.
“Why are experts inferior to algorithms?”
That’s just another great risk management question from the guru of behavioral finance on page 224 of “Thinking, Fast and Slow.” If you don’t have time to read the entire book, definitely take the time to study and consider the implications of Chapter 21, Intuitions vs Formulas. In the last 4.5 years of my time away from a hedge fund desk, I’ve thought a lot about Experts vs Algos.
Understandably, algorithms scare people; particularly people who have zero analytical competence in modern math (i.e. 99% of the politicians and central planners running America). This contrasts sharply with the Chinese political leadership where 8 of China’s top 9 political dudes are mathematicians and/or scientists.
I started building my own predictive tracking algorithms so that I could attempt to remove the emotion when I hit buy and sell buttons. It helped me so much that we started plugging these predictors into our fundamental Growth and Inflation models for countries. That’s why our intermediate-term forecasts on things like GDP, PPI, etc. are so variant from the Old Wall’s consensus.
I’m not saying that what we have built is perfect. However, I am saying it’s better than what I used to use – and a lot of people on Old Wall Street still use what I was taught to use at A) the Keynesian School of Economics and B) the Sell-Side brokerage firms that perpetuate its dogmatic principles.
Kahneman’s answer to the aforementioned question is pretty simple. “One reason, which Meehl suspected, is that experts try to be clever, think outside the box, and consider complex combinations of features making their predictions. Complexity may work in the odd case, but more often than not it reduces validity. Simple combinations of features are better.” (page 224)
Kahneman goes on to remind us that “humans are incorrigibly inconsistent in making summary judgment of complex information.” And if there is one thing that any of you know about your own team’s investment meetings since late 2007, that’s God’s honest truth.
Back to the Global Macro Grind…
Rather than attempt to handicap who has to chase S&P 1400 into options expiration tomorrow (there’s a massively skewed position in the 1400 strike calls vs puts), I’ll just rattle off what my Algos think on risk ranges, prices, and probabilities vs Experts:
- SP500 could easily trade to 1401 inasmuch as it could fall to 1369 – that’s my immediate-term TRADE range
- US Equity Volatility (VIX) holds its long-term TAIL line of 14.21 support like a champ; upside to $17.34
- US Equity Volume/SKEW signals are at least as bearish as the 1987 signals that started developing in Q1 of 1987
- The first 2 of 9 S&P Sectors that have snapped their immediate-term TRADE lines of support (XLE and XLB) did last yr too
- Size (as in the risk management factor to describe cap) flashed another bearish signal yesterday (Russell 2000 = -0.82%)
- US Basic Materials (XLB) and Small Cap (IWM) stocks have been making lower-highs since peaking YTD on Feb 3rd
- US Dollar Index has moved back into a Bullish Formation (bullish on all 3 risk mgt durations – TRADE, TREND, and TAIL)
- US Treasury Yields are ripping above their intermediate-term TREND lines of 0.26% (2yr) and 2.03% (10yr), respectively
- US Treasury Yield Spread has widened 20bps as the Financials (XLF) have moved to immediate-term TRADE overbought
- US Technology (XLK) Sector Study is flashing a grossly immediate-term TRADE overbought signal at $29.97
While it’s tidy to tell ourselves that everything in America is fine, what all of this is really saying is that if Apple (17% of the XLK) and the Financials (up in a straight line in the last 2 days in response to the rallying cry of “success” to a made-up test) stop going up, the SP500 will probably stop going up too, in the immediate-term (3 weeks or less), at 1401.
What are the rest of the world’s signals telling us?
- Japanese Yen is crashing (yes when a Top 3 world currency drops 10% in a straight line, that’s a crash)
- Japanese Equities (like European Equities did around this time last year – pre Sov Debt Crisis) like a crashing currency
- Chinese stocks, down for 2 consecutive days (-3.3%) post the US “stress test”, still see Growth Slowing
- Indian stocks, down -1.6% overnight, failing at immediate-term TRADE resistance of 18,023 again, don’t like oil up here either
- Germany’s DAX melts up to +20% YTD as German bond yield spreads versus US Treasuries widen (bullish for Germany)
- Spain’s stock market (IBEX) is flashing a very bearish negative divergence vs Global Equities (down -1.4% YTD)
- Spanish bonds, currency (euro), and stocks are now all falling at the same time – clean cut sovereign debt alarm bell ringing
- Greece’s stock market would need to close > 771 on the Athex to signal any accomplishment of quantitative support
- Israel’s Equity market (TelAviv25) up for the 3rdconsecutive day, holding 1081 support (post Gaza “truce”)
- Dr Copper agrees with China on Growth Slowing, failing to close above its long-term TAIL line of $3.98/lb, again
Obviously weaving throughout this Storytelling of “growth is back” (as US GDP Growth gets cut in ½ sequentially vs Q4) are US stock market centric people trying to tell you that Gold falling is a “bullish sign for US stocks” (like they did in FEB and SEP of 2011). My Algos vs Experts on that say God Speed. Bond Yields spiked, momentarily, as US Stocks topped in February of 2011 too.
I’ve tried to not get mad at my Algos since 2008. They don’t give me any lip, and they don’t make excuses when they fail. They may have not always made my risk management views popular either. But at the big turns, before big draw-downs in asset prices, they’ve also gotten me out of the way.
My immediate-term support and resistance ranges for Gold, Oil (WTIC), US Dollar Index, Japanese Yen, and the SP500 are now $1644-1691, $105.02-106.49, $79.79-80.61, $82.71-83.98, and 1369-1401, respectively.
Best of luck out there today,
Keith R. McCullough
Chief Executive Officer