What are the best technical indicators?

This is something I don’t know a lot about. Since according to bchad we should try to up the level of conversation on the forums, I’ll give this a shot.

What are the best technical indicators and why are they useful?

I use the 50 and 200 DMA as a general guide because I’m always interested if a stock is under distribution or accumulation but I wouldn’t rely on that metric alone. Actually, my favorite stocks are in equilibrium because that tells me the market isn’t on to it yet. You can see in microcaps when the market starts to get onto some story because they will have huge buy volume and “come to life” after being illiquid for a long time.

What else? Do you use RSIs? MAC D? Etc.

My overall view of the market in general is that individual schools of thought are incomplete so it’s useful to try to understand a variety of methods and integrate them. I’m really only interested in patterns (fundamental or technical) that can be explained by some sort of behavior. This is the problem I’ve had historically with technicals. Cup and handle? What does that even mean and why should I care?

Someone please educate me. Thanks.

I’m not particularly well versed in TA or Stats for that matter but here’s my take. Most TA is a less sophisticated method of quantitative analysis, likely invented before better methods were known/common or used because of their simplicity. For example, moving averages are known as non-causal finite impulse response filters in the time domain in the signal processing world, an EWMA is a form of a GARCH model, and a cup and handle is…well, a cup and handle.

TA can be used to describe and quantify the different moments of a distribution in the same way stats can, just less gooderer.

There are obviously some less reasonable TA tools that border on the absurd, but things like moving averages and range based studies are staples and useful to many practitioners trading almost anything.

At the end of the day you’re either the kind of person who buys a stock after an up move or sells it after an up move - and everything else is a description of how you do that.

Can you share an example of what you mean by a fundamental pattern that can be explained by a behavior? Are you talking about things like P/B ratios, things like joining an index, or macro-economic thangs?

Thanks for the reply LPoulin. I’ve often thought that moving averages do not inherently have any meaning (as compared to say, cash flows, which are a tangible thing) but that because so many people look at moving averages, they are highly meaningful.

Pattern is a broad word but here’s one example:

Let’s say that a particular company has earned a very consistent range of earnings over a long time period, say perhaps $1.00 to 1.50. Then one day for some reason one of their facilities burns to the ground. Now everyone knows they are going to lose money. The stock pummets. I think with some work you could reasonably understand that the company would again be profitable in the same range if they had a clean balance sheet and insurance for the facility. If / when they return to profitability, the stock will return to its normal trading range.

That’s kind of a boring example of a pattern, so you could say the pattern is that panic selling that does not lead to permanent solvency issues or impairment of earnings power is a good opportunity to get long. There are hedge funds that do nothing but chase fires like this and some of them are able to beat the supposedly unbeatable market by doing just that.

The behavior is “they’re not going to earn anything right now and I can’t guarantee it will get back on track and/or there is uncertainty, so I have to panic sell.” This seems pretty stupid on the surface, but the reason it’s a pattern (and not just recurring stupidity) is that a lot of funds live and die by their short-term (probably quarterly but definitely annual) numbers which means they might not have time wait, especially if they are concentrated. So merely by understanding the mandate of most of the other people in the stock, you can begin to understand what would cause them to act in a certain way – and these ways, or patterns, recur. Fidelity, for example, has a much different mandate than SAC Capital because they have totally different investment strategies and investors even though they are both equity funds.

So to summarize, the pattern is long-term mean reversion because of short-term mandates and company-specific issues. You could even talk to the guys who panic sold and they would say, “Yes, I agree with you that the earnings will probably come back but I can’t sit around and wait for that because my clients would kill me.”

Bro, this is way too coincidental. I actually came across my TA workbook the other day from a seminar I attended. I plan on taking a look at it over the weekend to brush up on some facts.

I like the moving averages and RSI. To me, this is TA 101 and simply looks at intraday momentum pricing. It’s been some time since I’ve really cared to use TA, but I’ve found myself picking things up at cheaper prices when I’ve traded in my PA.

To the L1 Candidates reading, moving averages refer to how the stock is trading in the last 50 days compared to the last 200 days. You could think of this as looking at a tugboat around a ship. The last 50 days will often be a leading indicator of how the entire 200 days will turn.

RSI is Relative Strength Index which marks the closing price trading pattern in the past 14 days. Lots of selling pressure will bring this index to ~20. This is a buy low signal. Lots of buying pressure will bring the index ~80. This is a sell high signal. This is true if you’re contrarian though, if you’re a believer in the smart money, you’d do the opposit ;-).

I am not an active trader, but these simple items are not a bad way to look at WHEN to Trade combined with the CFAesque analysis of WHAT to trade.

For those of you who have a MTA/CMT chapter in your town, I’d encourage you to check out a meeting. Most of this stuff is very fascinating. Earnings can be restated rendering multiples and DCF models useless, trailing stock prices are always correct and historical. The tape is seldem restated.

I found myself surprised to be a fan of technical analysis. I initially figured that the market would at least be weak-form efficient, even if the other forms were hard for me to swallow (just because I’ve taught this stuff doesn’t necessarily mean I *believe* it - the trick is knowing when the ideas are useful and when they aren’t). But over time, I’ve come to believe that TA actually has some value in short and medium term time-frames. I think technicals do capture short and medium-term changes in sentiment, and I think (though don’t have proof) that these ultimately feed back into risk premia. So the idea of “buy when others are fearful; sell when they are greedy” can translate to a TA mindset.

I want to point out that looking for “the best technical indicator” is a lot like searching for “the best valuation metric.” Like valuation, it is very context dependent. Every time you come down to one indicator or valuation metric, you start to realize “it doesn’t really work well in this other context, so you need to use a different one here.”

In the case of TA, the context depends a lot on things like what your transaction costs are, how long you intend to hold positions, can you sit on cash while you wait for an entry, and - above all - your personality. Traders are pretty unanimous in mentioning the psychology part of it - you need to choose an indicator that resonates with your trading/investing personality, because if you don’t, you will second guess the signals it gives you and not have the discipline to follow the signals, cut your losses when they don’t work, take your profits at the appropriate time, and manage your position sizes appropriately. I suspect that TA improves people’s trading not so much because the signals are magic (some are better than others, for sure), but because using them enforces discipline and serves as a crutch against one’s emotions about making and losing money. It sounds really crazy, I know, but even the guys who use astrology may actually improve their trading (vs running with gut feelings) if their astrological analysis comes along with sensible stop-loss and position sizing disciplines.

As for personalities, LPoulin pointed out that most people break down into trend followers (momentum people) or countertrend traders (mean-reversion people); there are also some people who are carry types, but I haven’t fully thought about how well TA mixes with carry approaches. Momentum people tend to go with moving averages, MACD, RSI. Countertrend people tend to go with Stochastics, other similar “oscillators,” or Bollinger Bands for their entry signals.

Things like ATR (average true range) are useful for determining stop-loss thresholds and (depending on your system rules) profit-taking thresholds. You can also use them for determining position sizes, the basic idea being that no position should cost you more than X% of your capital if it goes against you, and you use ATR or other volatility indicators to figure out what size position will adhere to that risk budget. Of course, things can gap against you sometimes, but that’s one reason that X% needs to be pretty conservative, and you try to make up for it by trading uncorrelated assets so they won’t all gap at the same time (as much as you can avoid it).

Elder’s force index always appealed to the physicist in me, but in practice I find that it is a coincident or even lagging indicator - by the time I get a signal, I feel like most of the move is over. It’s useful as a diagnostic tool, but I don’t find a good trading tool.

Most technical analysis is basically three-stage: you have 1) a trend (or other chart pattern), then 2) you have something that signals a possible change in that pattern, and 3) you have second indicator or event that confirms the change in that pattern. Generally, you don’t actually act on a signal until you get to stage 3. If the signal turns out to be a false signal (and you’re back to the trend at stage 1), you cut your loss quickly and wait for another signal.

Ultimately, people tend to choose one or two indicators and run with it: many of these indicators catch more or less the same stuff, and the trade-offs tend to be that some catch moves faster, but with more false positives, while others are more reliable, but give signals somewhat late. If you need more reliability because your personality is that way, then choosing a faster one can work, but you’ll be lax to act on the signals. If you are happy to take a bunch of small losses for an occasional large gain, then you probably want a signal that is less reliable but signals earlier.

It is important not to use too many indicators simultaneoulsy, because you can easily over-optimize things and find a backtest that works beautifully but then performs horribly out of sample (i.e. going forward). This is especially true because some indicators condense many different measurements simultaneously and therefore consume more degrees of freedom than you’d expect with just four or five numbers.

A lot of people think that technical analysis is about seeing a pattern and then trading on the idea that it will complete, but actually it’s the change in patterns that is more important. If a pattern looks like it is going to be a cup or a head and shoulders or something like that, but then stops and turns in to something else, *that* is the signal to go long or short. What the patterns do is warn you to get ready to make a change.

For a valuation guy like Bromion, my guess is you are holding your positions for a long time, and so the real use of TA is to avoid entering long positions while they are still falling and to avoid shorting things when they are still rising. So obviously you may want to have a trend indicator like a moving average to spot the direction, and then mix that with an oscillator to try to detect when something has come off of a bottom. Bottoms can be jumpy though, so it’s important not to obsess about catching the very bottom, which is the hardest point to get, and if you are doing a value strategy with a long holding period, it is nice, but not truly necessary to get the exact bottom.

Don’t forget that TA can also be useful for exit signals once you’ve made an investment. If something that was low valuation has performed well due to some catalyst and is now at high valuation, it can be time to take a profit, and a technical indicator may help with pulling the plug. I think that this is a place that I haven’t heard a lot of value people consider TA… I hear about using TA for entries, but the discipline of when to exit isn’t always so clear, but is ultimately just as important.

It’s taken me a long time to get familiar with TA, so I was hesitant just to dump a bunch of knowledge on someone who comes here saying “hey, just teach me everything you know about something,” but bromion has made some great posts on valuation, value investing, and especially on identifying companies likely to blow up, so one good turn deserves another.

Thanks bchad.

I use moving averages for this reason and have started to experiment with RSIs.

One thing I find really interesting is that a lot of the best ideas in the stock market are invisible to the majority of investors until they start to work. For example, a company that has been dormant for a long time but is about to have explosive earnings growth will not necessarily screen amazingly well on P/E or other value factors. Likewise, it probably won’t have an exciting trending chart because the earnings haven’t exploded yet. Once the earnings do suddenly increase, then it begins to screen better, which in turn causes the trending chart to occur (or usually does at least).

Presumably someone is finding these before the break out, so I was wondering if there is a way to identify those on the technicals first. I think probably not because anyone who found them would of course try to buy them quietly. It also seems that by definition, the technicals are the lagging indicator since they are always backward looking based on actual price data.

The other area that interests me is why certain charts have successive gap ups instead of trending. You can look at CCIX for example which had 3 gap up patterns each time they reported earnings in the last 3 quarters, vs. a company like CREE that had a consistent up trend between October '12 and August '13 before they missed earnings. Both companies had surprisingly strong earnings growth with a positive outlook, so it’s interesting that the charts look so different. I suspect this is a function of both liquidity and the “story” around the stock (CCIX is a boring commodity industry but some people think CREE’s new LED lighting product is going to grow for a long time).

Most reliable (relatively speaking) technical indicators in my experience (this isn’t an exhaustive list… I’ve never investigated all of them in detail):

  • Exponential moving averages.

  • Fibonacci Retracements

  • Pivot Points (for intra-day trading)

Indicators which I no longer look at:

  • RSI and other overbought/oversold indicators (stuff can remain overbought/oversold longer than you can imagine)

  • Chart patterns like head/shoulders, double top/bottom, etc. (These are too subjective to be of any use and everyone seems to have different precise definitions.)

No question about that. Sometimes it’s amazing to see a stock that is statistically cheap with 8-10 quarters of pretty good sales growth behind it where the price hasn’t moved. I owned a stock like that where it got written up by someone on SeekingAlpha and moved 50% in two days. For some reason the market didn’t love it until someone wrote it down even though nothing changed in the numbers.

Likewise, you can see companies that are “obviously” committing fraud that are way up and have market capitalizations for hundreds of millions of dollars but which are actually worth zero or close to zero and it often takes the market a long time to figure it out.

Indicators are pretty much always going to give you late signals if you are using “the bottom” as your standard. It just comes with the territory. As far as crap companies that get discovered and earnings explosions that cause gaps, there’s not a lot that is going to trigger a signal in time to catch those big moves up that you want (and that’s probably why the moves are so big - there are only a few people on that side of market at that point, so a tiny change in demand can create a big change in price)

The traditional value technique to look for catalysts may be more useful here, perhaps combining with options plays to establish your position.

If there is any indicator that will be leading, it may be that some kind of volatility indicator ticks up just before a jump, as some people catch on to the catalyst while others are asleep. Probably selling out of the money puts and buying calls to offset them at moments like this is a sensible strategy, provided you are sure that the company isn’t going to zero.

(Edit: Declining volume is also a potential signal that whatever last piece of relevant news has been digested by the market and a jump requires a new piece of news, but like price, it can keep on declining until it suddenly doesn’t anymore. I do think that volume contains useful information that most non-technical approaches miss out on, though)

The technical strategy if you are using an uptick in volatility as your signal would typically be to buy/sell on a breakout up/down from the current range of volatility or ATR. Unfortunately, that will mean that you enter at the top of the current range (for a buy) and not at a true bottom on the chart, but nothing other than luck is going to give you perfection. Again, it will look “late” if you are taking the very bottom as your reference point, but it will probably give you more confidence to put on a position in size.

If the price level turns down again, you cut your position (or manage with futures or options) to keep losses small. If the the indicator worked, then you didn’t get the exact bottom, but you are still likely to have some momentum as people wake up and notice. And since you already have the company on your radar based on value criteria, then it should be “good enough” for you to hold in your portfolio for the long term.

In theory this is how you avoid losing money in any investment. Of course that requires that you be right and actually capable of holding on to a stock for a long time which most people are not (or would not).

Timing the buy is probably the hardest thing in all of investing (some would argue timing the sale is harder but I don’t agree with that). It’s very easy to buy things that “seem so cheap” and then sit there for a long time until (“if”) the market comes around to your way of thinking. If you own a good, growing company at a decent or good price though, at least it’s unlikely that your position will be impaired.

One key aspect of my business model is trying to find invisible stocks that are going to have a breakout. Sometimes I get it right for a quick kill (CCIX which I bought on March 5th) but other times I am too early and have to wait.

Are you in NY, Bromion? Perhaps we should chat about this stuff over beers sometime.

Unfortunately no, I live in the Bay Area. I will probably be in NYC early next year though on a fund raising trip. If that happens I will drop you a line first and we can grab beers, that would be fun.

I grew up out there. I’ll be sure to let you know next time I go back.

elliot wave theory is very interesting

I don’t have the knowledge on life and everything like Bchad, but I know some random stuff. Just be advised that this is mostly stuff that I’ve read. Asides from quick looks on RS, length of momentum and some contrarian thoughts, I have barely used this stuff, if at all:

Support and resistance: You probably all know what this is, but one thing many people ignore is that support and resistance are behavior related. Support is roughly a consensus target price to sell, and resistance is roughly a consensus target to buy. It doesn’t work a lot of the time, but it’s something interesting to have in mind. If a breakout happens accompanied by a paradigm shift (let’s say the company releases a stupid or brilliant strategy) there may be some extra room for change.

RSI: does not test well, even though it’s very popular. A lot of people confuse RSI with RS. RSI shows what is overbought or oversold, and apparently sucks. RS measures strength of a security against other security or the index, which works pretty well.

RS: can be taken as the most recent six month price change. The top decile in this criteria continued at above average levels for the next 3 to 10 months in studies. This seems reasonable - part of the market likes to buy things that are going up (either to enjoy momentum or just for being naive and dreaming of big riches).

What sucks about momentum is that it seems to be very strong for some time, but over a longer period the stocks tend to underperform (so long term investors can only bet on momentum once they’re convinced that catalysts will make the fundamental part catch up with the stock once momentum subsides).

For momentum, Jegadeesh and Titman (I think it’s Titman) have some pretty interesting papers. If I recall correctly, they proved that buying winners and selling losers got alpha for some time (up to 15 months I think) and they redid the study a few years later to find out that momentum was still beating the market. There’s also more research from a bunch of other guys such as Carhart (he improved the Fama-French model by adding momentum to it), and they all give momentum a thumbs up.

Haurlan Index: This makes sense. It is the number of stocks that advanced minus the number of stocks that declined, exponentially smoothed for a 3-day EMA. Then you see if the 20 day EMA of this number is crossing the 200 day EMA to get trading signals. The measurement seems a little weird, but it makes sense in my opinion because it gives us a different view of looking at the market. Maybe the S&P is going up because Apple, Google and a few others are hot but, if more stocks are falling then going up, maybe the waters are more dangerous. This tested without favorable results while a slightly modified version tested extremely well. Maybe that’s just noise, but looking at advances minus declines in any way seems fast and gives a different view of overall market trends. There are about 12 million variations of advance-decline line tecniques anyway…

News: I have no data, but this seems to work well as a contrarian signal. When news are really really bad, there may be a bottom - when they’re really really good, the stock is screwed. It seems sensible since if everybody is optimistic, everybody already bought the market/the stock. As anecdotes go, recently Shinzo Abe made the cover of The Economist as Superman, and a week later their stock market started a 15% plus fall . They also made an awful report o Gazprom and later the company went up about 50%. Of course nobody should use it as a definite signal. But, since everybody loves Tesla, I’d think twice about buying it (momentum may run out). I’d also not short it though (momentum may run over me). A local example would be Eike Batista companies - this guy had 30 billion dollares and was a brazilian superstar - now he risks being broke. He basically got rich from companies that were yet to produce anything.

Of course best sellers and politicians can go way dumber than The Economist, so they may be even better contrarian indicators. We’ve seen many end of the world books while stocks more than doubled from the bottom of the crysis - once we see a new Dow 50,000 or whatever book selling like hot cakes we’re probably screwed.

Buy on Monday and sell on Friday: The reasoning is that much more bad news are given on friday after the close than at any other time (companies want to avoid panic selling and they probably figure less people will be paying attention to the weekend news anyway). Mondays as a whole don’t do well historically, so maybe waiting until tuesdays for long term buys enhances the chances of grabbing a few bps.

January effect: This does not seem to work anymore (arbitraged?) but would come from window dressing and tax losses realization in December

Thing that I think are less reasonable/well explained:

Bulkowski: This guy (http://www.amazon.com/Thomas-N.-Bulkowski/e/B001IGOYSE) studied a bunch of patterns. Some had amazing historical results, but I can’t really tell what is data mining or not. If somebody want to try stuff, look for descending triangles - Bulkowski says their median was 34% vs 5% from S&P on the researched period.

Patterns: Data mining or no, Flags seem to have beaten almost anything.

Volume: Just my opinion, but I don’t like volume all that much because so many people trade in derivatives now, so whatever worked in the past is probably getting very different very fast (how will I know if vol is really down? - maybe people are trading structured derivatives like crazy on my stock). There are many volume indicators and I don’t trust any of them.

Stuff to read:

This paper http://papers.ssrn.com/sol3/papers.cfm?abstract_id=603481 basically reviewed most TA studies. It seems TA worked in the past, but not so much anymore. It lost strength in the US on the 80s and 90s and is losing strength on other markets within time. Of course it is one thing to look fo Dead Cat Bounces and another to do whatever the heck RenTech is doing to make all that money. This book http://www.amazon.com/Technical-Analysis-Complete-Financial-Technicians/dp/0137059442 feels like the core of CMT Level 1 readings. It has a pretty wide coverage and it is as balanced as a TA book can be, I guess. The others seem to be mostly repeating stuff from this or just diving deeper into small specifics (like the Point and Figure book).

The main thing I don’t like about technical analysis is that there is no unified framework between the different indicators. Ten indicators might say to buy, but then another might say to sell, but maybe those ten are worthless when the sell indicator is flashing.

At this point, the extent of the technical analysis I’m willing to do is to look at a chart and see whether something is trending up/down in the medium-term (couple of months) versus a shorter horizon (days to couple of weeks). And that’s really just used to get a sense of whether there are idiosyncratic risks that might not be captured by whatever models I’m using.

This happens with fundamental analysis too.

What do you do if the P/E is negative because the company is losing money, but it’s a good business trading at a low multiple of sales and below tangible book value where the company has a pretty good balance sheet and is not burning cash? Do you pass because the earnings are negative while the rest of the scenario looks good?

What do you do if the company has a long-term positive secular backdrop (like favorable demographic trends, for example) but sales are currently declining for some temporary reason?

What do you do if you conduct some convincing research on a company that tells you to buy the stock but the sell side recently downgraded the stock? Stocks statistically underperform after the sell side downgrades but your research indicates this stock is a buy right now.

With almost any stock, you can find a reason to like it and a reason to dislike if you look deeply enough. I have a feel for the fundamental stuff because I’ve been doing it for 8 years but I have no feel for the technical side when it comes to conflicting indicators and what I should pay attention to.

I think that’s a good point. I actually had considered writing that it was an issue for fundamental analysis as well, but I didn’t want to hijack your thread. I don’t do much fundamental analysis on individual stocks. I prefer to take that fundamental (or technical) information, build a model that can consider all that information jointly and then invest on the model’s predictions. If downgrades have significant information, the P/E is negative, but other ratios are favorable, let the model decide how important everything is (cross-sectionally across all stocks in the universe and over time). I would then look at the predictions and think about why is the model wrong or what isn’t the model catching and try to figure out how to make adjustments to my allocations based on my confidence in the model.

So what I meant by checking the medium-term and short-term trends was mainly to compare it to what the model is predicting. If the stock is ranked highly in the model, but has been falling significantly over the past few months, I might not take as much of an active bet than if the stock has been steadily rising (my models all incorporate a momentum variable, but momentum is different from a visual inspection of a stock versus a trend line). So the technical indicators I look at are mainly as a validation of the model’s recommendations.

I also don’t handle any execution, so perhaps if I were trading and looking at things on a very short-term basis, I might pay more attention to TA.

I appreciate your insights jmh and there is no real objective in this thread other than interesting discussion so I don’t mind at all if it goes in another direction.

One question I have for you is how do you know that the contents of your model are not already embedded in security prices? I think that’s unlikely for a lot of reasons, particularly in a multi-factor model, but it’s a key question.

It seems to me that the market (in general) is very good at discounting what is already known, meaning what is in the numbers already. If the earnings are $1.00, then unless there is some kind of accounting fraud or other outlier situation going on, the stock is not going to sell for say, $3.00, because the market isn’t that dumb. Everyone running low P/E screens will see that the stock is trading for 3x trailing P/E and so people will buy the stock. Probably some pretty good number of people would buy that stock without even doing any real analysis at all.

This is not a criticism, but I’m just curious how you have gotten comfortable with your screening approach if you’re able to discuss and share your thoughts.