Friday, May 22, 2009

3 Lower Closes - A Largely Misunderstood Edge

The S&P 500 closed lower for the 3rd day in a row yesterday. Three lower closes is often cited as having an upside edge. And it does – kind of. That edge is often misunderstood, though. The first place I saw 3 lower closes quantified was in Larry Connors book “How Markets Really Work”. One of the chapters in the book looked at consecutive days higher and lower. It basically found that after the market has moved in 1 direction for several days, there is a tendency for it to revert.

He measured 3 lower closes in that book from 1989 – 2003. (All tests were run over that period. It wasn’t specific to this particular setup.) When Larry measured 3 lower closes he looked at any time the market had pulled back for at least 3 days in a row and then showed performance statistics for the following days. What many traders fail to realize when they review his research is that there is a large historical difference between “at least” 3 days in a row and “exactly 3” days in a row. I decided to examine this in some detail tonight.

First let’s look at a chart of buying the S&P 500 any time is has closed lower for at LEAST 3 days in a row and then selling the next day. Keep in mind, if it is down again day 4 it will be bought again. Same with day 5, 6, 7 etc. until there is finally an up day.

There are a few things to note here. First, trying to buy all 3+ day pullbacks prior to 1987 was a losing strategy. After that it the market showed less tendency to trend and an increased tendency reverse. Buying 3+ day drops became profitable. The period covered by the blue arrow shows the period covered in “How Markets Really Work”.

Now let’s look at what happened if you bought the market after 3 and only 3 lower closes. In other words, the third day lower was bought. The position was exited the next day. If the market continued to head south it was ignored. There was no further buying on day 4, 5, or 6.

There is a striking difference between the two graphs. There does appear to be a recent upside edge, but most of it is concentrated on some outlier trades that occurred in the last year. Until very recently there was no advantage to buying the third close lower in a row.

Why the stark differences? I believe much of the reason is due to the strength of the eventual snapback. The more stretched the market gets to the downside, the greater the snapback typically is. The “3 or more” study guarantees a winning trade in every sequence. In other words, it will continue to buy each day until it has a winning day. Only after that will the count reset. The further the market drops, the more vicious the snapback is likely to be.

Below is a table that illustrates this concept covering the time period of 1989 – present. The left hand column is the number of days the market moved lower. The right hand column is the average up day the following day. (Down/losing days are not looked at here.)

As I stated above, the longer the pullback, the stronger the snapback.

Also consider when snapback is most vicious…during bear markets. Note the two time periods on the 2nd equity chart where buying three lower closes has actually provided an edge on the 4th day. Those 2 time periods were during the current and prior bear markets.

So is there really an edge to buying (exactly) three lower closes? Recently during bear market periods – yes. Historically, no. Of course if you understand the mean reversion will likely eventually take place, you can still take advantage of the pullback. You will need some patience, though. Below is a chart of buying day 3 and holding for 3 days (rather than 1).

This chart looks more like the 3 OR MORE chart shown above. In this case it is due to the longer holding period. The longer you hold the more likely you are to participate in the eventual snapback.

The bottom line is that 3 lower closes may indicate the market is getting stretched. The market will likely bounce some time soon. It doesn’t normally offer much of a day 4 edge on its own, though.


Daniel said...

> there’s a “historical difference between AT LEAST 3 days in a row and EXACTLY 3 days in a row”...

That kind of intelligent exactitude and careful precision in analysis is why we pay you the big bucks, Rob H.

Basically the ‘question’ in investments is always an eternal one, if properly phrased. (And if it is NOT an eternal one it is not properly phrased...)

?? When does one celebrate WEAKNESS and buy it, because it affords a more favorable entry, and when does one eschew weakness because it will lead to LOSSES within one’s target timeframe??

There can never be a simple or ultimately non-paradoxical answer to an eternal question-- which is why Rob’s methodology is so outstanding. It does not throw up “answers”, but only probabilistic anticipations based on past patterns of a similar nature.

This is why I cringe when new bloggers wander in and say NICE CALL Rob or TERRIBLE CALL Rob regarding a short term outcome. He does not make calls, he points out probabilities. An exam of his tables indicates that even the strongest seeming edges had non-successful outcomes in the past, some of the time.

It is entirely risk/benefit allocation. Very little prediction, as the term is commonly understood.

All that said, Rob, I would imagine that filtration with various nuance parameters such as ABOVE / BELOW some significant MA, MACD on BUY or SELL, New Highs/New Lows levels here or there, etc. etc. may well indicate or suggest different wrinkles, with an underlying concept as amorphous as “3 days in a row of lower close”...


Blue said...

Since I agree entirely with Daniel, the following question may seem to have missed the point. But the question is: have you done a similar study for multiple updays? Also why start with 3? What about buying/selling after 1+ and 2+ down/up days?

Rob Hanna said...

I ran the study based on 3 down days because that is what had just occured. Based on the comments received I think this post will need some follow-up. I'll be sure to look at consecutive runs a few different ways.

As Daniel suggested,if we use up/down days in conjunction with other filters or parameters they can be helpful in establishing edges. You can also adjust your exit to help establish an edge.

A recent example is the gap study I posted within the last few weeks that looks at large gaps up after the market has closed down 2 days in a row vs. up 2 days in a row.

Anyway, I'll be sure to do some follow-up on this concept and look at it a few different ways.

Also, Daniel, I too cringe at some of the nice call / terrible call comments. I've been thinking about doing a post that dicusses in more detail how the edges may be used to better establish a mindset rather as a mechanical entry.

Thanks both of you for the thoughtful comments.


Quentin said...

Thanks for this article . Really got me thinking. Then again, you ALWAYS get me thinking :)

AlphaSeeker said...

The 3 days runs do actually work pretty well, but as Rob says, one needs to have patience and nerves of steel.

It is relatively symmetrical trading strategy, i.e. 3 up and 3 down produces a pretty good effeciency after the E-mini contract was introduced. An addition would be to slap on a fixed ATR money management on top of it.

The strategy works best on US futures markets, and is a losing strategy on less liquid markets...