Lumpy but Robust [ed note: this article has originally appeared at the Evil Speculator and was written by trader and ES contributor Scott. We provide a link to Scott’s past articles below this post for readers who want to get more familiar with his ideas and/or any unusual terminology used in this article] One continual theme in my trading is that every time I think I have it figured out, I get punched in the face by an unexpected problem. The tendency is to go more complicated, but often the solution is a degree of acceptance with respect to the nature of the game. Sometimes my edges work, sometimes they don’t. Sometimes they stop working for long periods of six months or more. Financial markets – multi-layered like onions That’s actually fine for me, but it isn’t for many
Evil Speculator considers the following as important: The Stock Market
This could be interesting, too:
MN Gordon writes Credit Contraction Episodes
Pater Tenebrarum writes Strange Moves in Gold, Federal Reserve Policy and Fundamentals
Dimitri Speck writes Seasonal Trends in Individual Stocks – More Than Just Random Patterns
Dimitri Speck writes Boosting Stock Market Returns With A Simple Trick
Lumpy but Robust
[ed note: this article has originally appeared at the Evil Speculator and was written by trader and ES contributor Scott. We provide a link to Scott’s past articles below this post for readers who want to get more familiar with his ideas and/or any unusual terminology used in this article]
One continual theme in my trading is that every time I think I have it figured out, I get punched in the face by an unexpected problem. The tendency is to go more complicated, but often the solution is a degree of acceptance with respect to the nature of the game. Sometimes my edges work, sometimes they don’t. Sometimes they stop working for long periods of six months or more.
Financial markets – multi-layered like onions
That’s actually fine for me, but it isn’t for many other people. The systems one ultimate chooses have to suit one’s personality. If one cannot handle extended periods of working hard without making money (I can) one has to retool one’s trading systems to avoid such situations.
My opinion is that the best edges are robust. Robust edges tend not to disappear, but don’t post objectively high results either (in terms of SQN, Sharpe or expectancy). I’m investing in a fund that generates the returns (after fees) shown below, by using a very simple and standard approach to trend following, not much different from very similar methods employed by numerous other firms.
The fund is clearly an amazing investment, but its returns are lumpy; e.g. it generated a 108% return in 2008, but had a few negative years as well. Its Sharpe ratio is only around 0.7, not exactly institutional grade for most people. But its edge is robust, demonstrably provable, the system is simple and I understand it.
It is almost inconceivable that 20 years from now (the time frame I intend to keep this investment) I will be looking at an empty account saying “I wonder why trend following stopped working”. If the dreams of the bears come true, the odds are very strong that this fund will post another triple digit year. This is the gold standard of a robust edge for me.
Looking for an Edge
My opinion is that simplicity is sophistication, and complexity is laziness. The first thing one needs to figure out before going any further is: Does a system have an edge? Use a scatter plot to figure that out. If you don’t want to do that, pick the classic edges used by other professionals and you won’t go too far wrong.
At Evil Speculator, user Francis (Mulv) has gone out of his way to provide some excellent statistics. What he discovered is a classic property of mean reversion systems, which segues into a classic mistake often made in back-testing. That problem is selection bias; if e.g. a random entry is selected and tested on a market one knows has gone up, it will probably show an edge.
One can avoid this tendency to deceive oneself by testing random markets (and keeping the results hidden so one cannot trick oneself). Professionals refer to this as “out of sample data”. Another method is to sequester part of the available data and keeping it aside, and then test one’s data against it to see if one’s hypothesis matches clean data.
One thing we know for sure is that mean reversion works dramatically better in the direction of the higher time frame trend. So if this works as a short setup in the current stock market, it should work even better as a long setup (or something is very wrong). Let’s delve deeper into this and take a look at the bounty Francis has provided.
I haven’t had the time to personally verify the statistics, but at a glance they look right. Take it for what it is, an interesting exercise you can apply to your own potential setups, rather than an authoritative data set. Also, there is a significant chance that I have misinterpreted the data, since it isn’t my own spreadsheet I’m working with. If so, mea culpa, but it remains a valuable exercise to learn from.
Detailed instructions on how to do this can be found by clicking here (read the bit about adding regression lines). I will break down the results below. First off, Francis has tested with and without the condition “break of the low of the setup bar”, so we can take a look at how much, if at all, it changes things.
The reason we want to test first without the break of the low/high and then with the break, is that we can test everything one at a time this way. If the setup works with a break of the low, the effect may possibly only be due to the break. That is the kind of thing we want to find out.
At this stage I will only look at long setups. In a bull market the short side is categorically not an edge (not surprisingly, shorting the e-mini has been a good way to get one’s face ripped off over the last 7 years). The bottom line is that one has to take all these results with a grain of salt, since it is a strong market.
First let us look at what the market does on the day after making a double bottom (without breaking the high). You can see below that there is a decent positive correlation, the points are reasonably clustered (i.e., the R squared value is OK). This tells us what we would usually expect, namely that the market is trying to go up after making a double bottom.
Things that would be worth testing in this context would be buying the close and selling the following close, or alternatively buying the old spike low on a limit and selling the following close.
Day 2, however, is showing that the results of the first day are just a blip – the market is now negatively correlated. That is not what one would want to see, but perhaps there is a retest of the low (that’s drawing a very long bow).
On Day 3 we have a weak positive correlation, so the market is trying to rise again.
So what do we have here? We have a market with a very weak tendency to rise on days one and three after making a double bottom. This is not at all what we would expect from a strong edge. Aside from the result obtained on day one it actually isn’t an edge at all.
When we move on we will look at how it behaves with a break of the lows/highs as part of the conditions. Next we will consider it as a component of mean reversion, at Bollinger bands, Keltner bands and 50-bar highs/lows. The logic behind this is that a reversal setup is likely to work better at extremes.
For now we have to conclude that there is nothing inherently bullish about a short term double bottom in ES beyond producing an intraday bounce, which is actually great to know.
To be continued.
Here is a link to Scott’s previous articles at Evil Speculator for readers interested in learning more about the systematic approach discussed above and other strategies Scott has written about.
Charts by Mulvaney Capital Management, Scott/ Evil Speculator
Chart and image captions by PT, as well as light editing of the original article