Sunday, March 8, 2009

Putting A Bullish Signal In Context

Figure 1 shows the "Dumb Money" indicator. The "dumb money" looks for extremes in the data from 4 different groups of investors who historically have been wrong on the market: 1) Investor Intelligence; 2) Market Vane; 3) American Association of Individual Investors; and 4) the put call ratio. The indicator has moved to an extreme reading suggesting that retail investors are bearish on the equity markets, and this is a bullish signal.

Figure 1. "Dumb Money"

So what does a bullish signal mean for equity prices? To assess the significance of a bullish signal, I will construct a strategy to buy and hold the S&P500 only during those times when investor sentiment, as measured by the "dumb money" indicator, is bearish (i.e., bull signal). So we "buy" the S&P500 when the "dumb money" indicator turns bearish (i.e., bull signal) and sell the position when the indicator moves to neutral. All signals are based upon weekly closing prices, and slippage and commissions are not considered in the results.

Since August, 1990, there have been 41 unique signals (or trades) when the "dumb money" went bearish. 81% of the trades were winners, and the average trade lasted 6.1 weeks. Such a strategy generated 549 S&P500 points; over the same time period, a buy and hold strategy yielded 348 S&P500 points. With the strategy, your total time in the market would have been 21%. What is noteworthy about this strategy is that for only 1/5th market exposure you beat a buy and hold strategy by a little more than 1.5 times (549 S&P500 points/ 348 S&P500 points). So when this strategy is active and in the market, you are earning a return at a 35% annualized rate. I would call that an accelerated gain.

Applying the same strategy to the NASDAQ 100, yields the following results: 1) there were 41 trades; 2) 78% were profitable; 3) this yielded 1883 NASDAQ 100 points; 4) over the same time period, buy and hold yielded 858 NASDAQ 100 points; 5) with the strategy, your time in the market was 21%. With the strategy and for only 1/5th market exposure, you beat buy and hold by over 2 times. Or when the strategy is active and in the market, you will see your money grow at an 80% annualized rate. Now that is definitely accelerated.

So this is just one of the reasons why I believe that a contrarian approach is worthwhile to pursue. Now remember, we are not crafting a real strategy here; we are only looking at this period when investors (i.e., the "dumb money") are bearish.

But as we all know, there is no free lunch when it comes to the markets, and any discussion of a strategy would not be complete without some mention of the risks involved. This is my way of saying that betting against the consensus is no "holy grail" strategy.

To assess risk and to put a bullish signal into some context, we will look at the MAE graph for the above strategy involving the S&P500. MAE stands for maximum adverse excursion, and each caret in the graph is representative of 1 trade from the strategy. MAE measures in percentage terms how much each trade moves adverse to or against its entry price before it is closed out for a profit or loss.

So let's refer to figure 2 and the individual trade inside the blue box. This trade was put on and lost 7% (x axis) before being closed out for an 8.5% profit (y axis). We know that it is a profitable trade because the caret is green. Now focus to the right of the blue vertical line. There were 6 trades (out of 41) that had excessive MAE's that resulted in significant losses. There is real risk even when betting against the consensus. Of note, the signal labeled #1 was from May to August, 2002; the signal labeled #2 was September to November, 2008.

Figure 2. MAE Graph/ SP500

The MAE graph for the NASDAQ 100 trade is shown in figure 3. Of the 41 trades, 9 had MAE's in excess of 10%; this is to the right of the blue line.

Figure 3. MAE Graph/ NASDAQ 100

So let's summarize to this point. Betting against the consensus can yield significant gains with minimal market exposure. As the MAE graphs show, risks still remain.

Now the question becomes: how do we continue to maximize our gains while minimizing our risks? To look at this question, we will construct another strategy where we buy the S&P500 after two consecutive weeks of the "dumb money" being bearish; we will sell our position, as before, when the "dumb money" indicator turns neutral. With this strategy, there are 37 trades. 78% are profitable yielding 626 S&P500 points. Our time in the market is now reduced to 17%. So just by waiting a week to "buy", we increase our points gained (by ~15%) and decrease our time in the market.

Furthermore, the MAE graph (see figure 4, below) now clearly shows that any trade that had a draw down greater than 6% from its entry price ended up being a loser or closed out for a small gain. These are the trades to the right of the blue line. This suggests to me that any trade in the S&P500 that incurs a loss greater than 6% will likely not recover and lead to further losses. In other words, this is a failed signal.

Figure 4. MAE Graph/ SP500

Applying the same strategy to the NASDAQ 100, yields the following results: 1) there were 37 trades; 2) 73% were profitable; 3) the strategy now yields 2206 NASDAQ 100 points; 4) your market exposure is reduced to 17%. More points with less exposure equates to more efficient market timing. The MAE graph is shown in figure 5.

Figure 5. MAE Graph/ NASDAQ 100

So what does it all mean?
1) To beat the market, you cannot be the market. Therefore, bet against the consensus.

2) Like all strategies, there are failures.

3) When the "dumb money" indicator turns bearish, the optimal -not the best - time to act is after two weeks. Gains can be improved and risks reduced (but not eliminated). Using stop losses, hedging strategies, or entries based upon other technical markers (i.e, a close greater than the previous week's high) may improve the efficiency of this strategy.

4) I hope I have demonstrated the importance of paying attention to extreme investor sentiment.


Tim said...

How do you tell from your chart that the indicator has reached an extreme reading? From what I see, there are many times which the indicator went even lower? Wouldn't a bull signal only be trigger when the indicator starts reversing rather than just going below 0?

Guy M. Lerner said...

Below that level is extreme....what I have found is that waiting for confirmation leads to underperformance....which should tell you something about accepting risk

Below that level there is a high likelihood of reversal and when it will come is difficult to predict but somewhere in here

Anonymous said...

Where does the "smart money" figure into these trades? Is the Smart Money bullish in these scenarios, or is the Smart Money not considered?

Guy M. Lerner said...

Smart Money is not considered in these strategies as laid out in the article; in the January 20, 2009 sentiment post I wrote about the importance of the smart money providing confirmation:

"In 18 years of data and in over 45 signals, the "smart money" has confirmed the "dumb money"in over 95% of the cases. By this I mean, we need to see the "smart money" turn bullish as a confirmation of the “dumb money” turning bearish. Typically, lower prices will make the “smart money” turn bullish."

Anonymous said...

FANTASTIC timing on your bullish signal posts, Dr. Lerner. Today, Tuesday, had the profile of a classic 'bull market blastoff.' Granted, there can be fakeouts. But we have a good candidate here.

I've been reading your essays posted at SafeHaven for awhile. My trading experience has shown that by fading extremes in sentiment, it is sometimes possible to nail tops and bottoms to the exact day. Of course, like Babe Ruth, one invariably strikes out sometimes too, when 98 percent bullish sentiment goes on up to 99.99 percent bullish ... or the inverse, in the case of bearish sentiment. I thought the Nov. 20th low probably was THE low ... but your unyielding bearishness over the past weeks proved correct, in pointing to a lower low.

As you probably know, sentiment data is difficult to incorporate into mechanical trading models because it isn't well behaved. The absolute peaks of VIX, for instance, are much higher in some bear market bottoms and crashes than in others. However, your analytical approach seems to have made some progress toward normalizing sentiment data, so as to filter out secular trends such as the extremely low VIX in the early Nineties, versus the extremely elevated level at present. This has the makings of an excellent mechanical trading system, if sentiment is used to bias a trend-trading model -- i.e., making it easier to buy and harder to sell when sentiment is bearish, and vice versa.

I've had good results using this approach with interest rate trends -- biasing the model to buy more easily when rates are falling, or to sell more easily when rates are rising sharply. Combining the monetary environment with the sentiment environment as biasing parameters, with the stock index trend serving as the ultimate arbiter, could make a killer trading model with a long-term compound return in the 15 to 20 percent range.

-- Jim Haygood

Guy M. Lerner said...

Jim Haygood: Thanks for the compliments, but let me point out 3 things: 1) I didn't "call" anything as I am just interpretating the sentiment data as in based upon past instances what is likely to happen going forward; 2) even though the sentiment indicators gave bullish signals my use of them suggests that 2 consecutive weeks of the "dumb money" being bearish is the best time to buy---so I am still in cash with regards to equities(save for a hefty position in TBT -betting on higher yields); 3) as I stated in this article, it is my belief that initiating strategies at this juncture will easily beat the market; obviously I have combined these with other indicators and studies yet for the most part it really doesn't matter; I am a little leary of using a measure to guage how accommodative the Fed/ interest rates are as this has become a very poor metric since 2000; as far as real strategies that I trade, the CAGR are about 2.5x the major indices with very tolerable drawdowns (since the 1990's)

One more last point about sentiment in general: while I believe you cannot beat the market by being the consensus, it should be stated that sentiment was a very poor indicator from 1969 to 1982; looking the Investors Intelligence data from this period suggests that this is the case; it was a fair predictor of prices; I throw this out there to suggest that I do takes some skill, a bit of luck, and a bit of humility to realize that the only thing that we can control is when we enter and exit the market and how much we bet; the rest is up to the market.

Tyler said...


In your last comment you write:
"while I believe you cannot beat the market by being the consensus, it should be stated that sentiment was a very poor indicator from 1969 to 1982; looking the Investors Intelligence data from this period suggests that this is the case; it was a fair predictor of prices;"

Could you expand on this a bit in terms of what you mean by it was a poor indicator but a fair predictor of prices? That seems a bit contradictory.

Some excellent posts of late Guy, thank you,


Guy M. Lerner said...


Sorry just poor choice of words....sentiment did not work all that well in the 1970's...just looking at the Investors Intelligence data (which is the only data we have for this period), shows that it was just a fair tool for timing the market

If time permits, maybe I will put the data up next week