Market Inefficiencies Thread | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Lukas1's Market Thread 13 replies

Market Frustration Thread 8 replies

Arbitrage – Exploiting Inefficiencies in the Market 3 replies

Time in the Market, Beats Timing the Market? Does this work in FX?? 7 replies

Market Inefficiencies Thread

  • #1
  • First Post: Apr 12, 2008 5:53pm Apr 12, 2008 5:53pm
Hello.

I feel compelled to start this thread.

Basically, I started a thread earlier today about how all we do is come up with the same entry criterias.... Fibs, MA crosses, hedges, BB squeezes, etc. Should I continue?

So I want to step back and ask, what trades do you take rooted in INEFFICIENCIES.

This thread is for talk about that.
  • #2
  • Apr 12, 2008 5:55pm Apr 12, 2008 5:55pm
The first inefficiency (and only one that I am aware of) is trading nondealing desk broker spikes. It works like this. Take 10 brokers and compare their feeds. When one travels far away from all the others, take a trade.

It's likely that the price will return back to the others.
  • #3
  • Apr 12, 2008 5:57pm Apr 12, 2008 5:57pm
A second type of inefficiency is "stop hunting." Basically, you find out where the stops on currencies are based on a news service. When there is a lot of news about a particular stop level, you can watch for those stops to be broken by the banks as they search for liquidity. If stop levels are breached, often price will significantly rise or fall past those levels (hence, support and resistance as chartists say.)

These are the basics.... I don't have details, but rumor has it that there are some folks trading this way.
  • #4
  • Edited 6:17pm Apr 12, 2008 6:08pm | Edited 6:17pm
3rd inefficiency : information asymmetry

4th inefficiency : space-time analysis applied to the market.
  • #5
  • Apr 12, 2008 7:51pm Apr 12, 2008 7:51pm
Possible inefficiency: The rollover strategy Brijon talks about. Too early to tell, but there might be some meat on that bone.
  • #6
  • Apr 15, 2008 1:53pm Apr 15, 2008 1:53pm
Good one! Here is an inefficency that i am aware of. Trading the news spikes. There will most often be a retracement in the market after a news spike if the spiek was driven by traders\' sentiment. I gave a detailed post in another thread that I have attached here.
These are some of my thoughts on trading the news.

The movement in the currency market around news time is because of the way people perceive what the news is not what the news actually is. From studying many experienced traders and my own observation the fundamental movement from the news announcement is impactful but only temporary. The news will make a move based on a sentiment then return back to the over all trend of the market.

To see and example of this take a look at an area ion a chart where the market is trending in a 4 hour chart and mark the areas where a news announcement took place. Now look to see if it came into the overall trend of the market.

If the trend is up and the news drives the price down notice that it usually comes back up to the direction of the trend. And when the news is up in an up trending market the price will go up faster many times at these points.

So if you trade the news one idea is to wait until the market makes a move between 5 to 30 minutes and you get some good signals to enter in the direction of the overall trend. Take it for what it is worth as a common inefficiency we deal with.
  • #7
  • Apr 19, 2008 11:58am Apr 19, 2008 11:58am
There are a lot that I've mapped.

The biggest I've seen revolves around trader psychology: fear and greed. Resistance and support are good examples of this, at support people are too greedy and at resistance people are too fearful. Overpriced and underpriced numbers abound on the trading floor. When people get tired of that balance they start pushing it... the result is a huge emotional spike that drives price, then a sudden reversal of that (Oh no, did we overprice?!?) then causes the previous area to be tested.

This means that any manor of traded R/S, such as pivots and even some MAs, can provide a tradeworthy edge.

Yea, stop hunts are a fun edge. You can trade along if you want since most people place stops at round numbers. Round numbers are a fun inefficiency anyway.

Another is the way the dollar behaves in the bear market. For a while there I was just following the DJI.

Another is the way opening markets handle new pricing. News even comes out overnight and the markets respond as they open, but without all of the spread worries.

Another is the natural swings in volatility that occur during lunch and how a lot of traders prefer to go flat before they go when they're unsure of the trend. Not only is it good info, but it's an edge all by itself.

There are hundreds. Most are at the M15 level, they happen fast and then they're gone. There are a few that occur at higher intra-day levels, a tiny few that occur at the daily.

Oh, and some brokers will kill your account for feed hunting.
  • #8
  • Apr 19, 2008 12:17pm Apr 19, 2008 12:17pm
What you call an edge, I call gambling.
  • #9
  • Apr 19, 2008 12:30pm Apr 19, 2008 12:30pm
Quote
What you call an edge, I call gambling.
Fair enough, but anything over 50% is a statistical edge. Sit down at a black jack table and you're gambling, but start counting cards in a single deck game and you have an enormous advantage. IMO gambling is where you have less than 50% edge.

I know of no system, edge, or inefficiency that gives you a 100% certainty. Stop hunting can reverse, feed hunting can reverse (and get your account kicked), even long term cycle trading can be prone to misjudgment.

From our position as desk traders we don't have the luxury of small near-certain edges, if we were floor traders we could do market arbitrage, make spreads or profit from slippage, but we aren't... our edge has to come else where. Of course even those aren't 100%.

IMO periods of low liquidity provide inefficiencies, periods of irrational trading provide inefficiencies, periods of herd activity do as well. They discount one possibility over the other, effectively providing "support" to one outcome over another.
  • #10
  • Apr 20, 2008 9:05am Apr 20, 2008 9:05am
Quoting Rabid
Fair enough, but anything over 50% is a statistical edge. Sit down at a black jack table and you're gambling, but start counting cards in a single deck game and you have an enormous advantage. IMO gambling is where you have less than 50% edge.
.
In blackjack, card counting gives you about 2% edge over the house (max). So it's not a huge edge, but enough that it'll get you booted from a casino if they think you're doing it.

One project I have running in the background is measuring the statistics of how well indicators predict the future. Most indicators are a coin flip, but I have a few on the longer time frames (4 hour to day) that show a 5% to 8% advantage. Time frames below 4 hours are almost always a coin flip, at least with what I've tried so far.
  • #11
  • Apr 20, 2008 11:18am Apr 20, 2008 11:18am
I'd like to see you start a thread on this study you are conducting.

The problem with this approach is "curve-fitting."

You can have a completely random data set (the market) and find what appears to be an "edge" with an indicator that would have done very well. In fact, you can have 5 or 6 years of data, and still have nothing more than coincidence.
  • #12
  • Apr 20, 2008 1:07pm Apr 20, 2008 1:07pm
Quote
In blackjack, card counting gives you about 2% edge over the house (max). So it's not a huge edge, but enough that it'll get you booted from a casino if they think you're doing it.

I always heard it was closer to 1%, lol. Which is quite huge given large enough trials, hence the reason they'll boot you. Give me a guaranteed 1% advantage in the markets and I'll own the world. =)

Quote
One project I have running in the background is measuring the statistics of how well indicators predict the future. Most indicators are a coin flip, but I have a few on the longer time frames (4 hour to day) that show a 5% to 8% advantage. Time frames below 4 hours are almost always a coin flip, at least with what I've tried so far.

Very nice, when you're done do let us know. What indicators have you tested?

I suspect most indicators don't read anything you can't already see via price. Of course you don't need to always predict direction, you only need to predict one element of a trade (direction or momentum) to have an edge.

Quote
The problem with this approach is "curve-fitting."

But aren't all techniques curve-fit? I would surmise there exist no techniques today that are guaranteed to work in all markets for the remainder of time.

Quote
You can have a completely random data set (the market) and find what appears to be an "edge" with an indicator that would have done very well. In fact, you can have 5 or 6 years of data, and still have nothing more than coincidence.

The thing about random data is that there are patterns within it. Take for instance simple standard deviation from the mean.

Put your entry orders and SLs based around standard deviations, use good money management... sure some will stop you out, but most will reverse before the stop loss and go in your favor. Yes the data is "random" to a point, but the random data is generated by something non-random (with a set number of outcomes). The reverse approach could work too, or any number of variants.

That's the humor in the "random number" argument. Just because you cannot predict the exact next flip of a coin doesn't mean you can't predict with at least a 50.01% edge what the average of the next X coin flips will be given the right data. Over a large enough series of trials the odds always regress to the mean, then away from the mean at some point.

The irony here is that it would actually be easier to make money in a purely random data set since we know in advance the number of possible outcomes and it's mean. This is not something we can know in the markets. Ie: random data can be effectively curve-fit, the market will defy that.
  • #13
  • Apr 20, 2008 2:20pm Apr 20, 2008 2:20pm
I'll argue back and say that you never know when the regression to the mean will happen. A strong trend can keep the price off of a moving average for weeks at a time.

As for curve-fitting.... if indeed the market is truly random (which I don't believe there is any question that it is) then expecting a curve fit strategy to produce pips is a logical fallacy. Its an incorrect manmade assumption that history will repeat itself. Hell, just look how different your charts are from year to year on the same instrument!

Quoting Rabid
That's the humor in the "random number" argument. Just because you cannot predict the exact next flip of a coin doesn't mean you can't predict with at least a 50.01% edge what the average of the next X coin flips will be given the right data. Over a large enough series of trials the odds always regress to the mean, then away from the mean at some point.

The irony here is that it would actually be easier to make money in a purely random data set since we know in advance the number of possible outcomes and it's mean. This is not something we can know in the markets. Ie: random data can be effectively curve-fit, the market will defy that.
  • #14
  • Apr 20, 2008 3:26pm Apr 20, 2008 3:26pm
Quote
I'll argue back and say that you never know when the regression to the mean will happen. A strong trend can keep the price off of a moving average for weeks at a time.
Absolutely, but does it matter when it will happen? In a truely random set of data there is no strong trend, merely a trend that hasn't regressed yet.

Quote
As for curve-fitting.... if indeed the market is truly random (which I don't believe there is any question that it is) then expecting a curve fit strategy to produce pips is a logical fallacy. Its an incorrect manmade assumption that history will repeat itself. Hell, just look how different your charts are from year to year on the same instrument!
The random walk theory has been disproven for years, you need only look at human behavior to see the patterns. But that argument is tangental, in a completely random data set a curve-fit strategy is exactly the best way to produce results.

The decay of a single atom is random but the half-life of a material is still accurate enough to fuel a society.

I've gone back 100 years on some instruments (look at commodities, there's some good info there), and over 20 on any I actively trade (with the obvious exception of the euro). Yes they change, but they change in cycles. You may not be able to predict when the cycle will change, but you can recognize when it has changed or is in the process of changing.

History does repeat itself tho, even in a completely random environment. If you flip a fair coin (50/50) and tick up when it's heads and down when it's tails... you'd get a random data set, right? But there are still only 2 outcomes, heads and tails. You won't get a coin landing sideways, or spinning, or an uncertain result or an unpredictable result.

In that respect the market is most certainly not a coin flip, it's closer to a roll of the dice where occasionally people ignore certain rolls. My old "9 or greater" analogy works well here.

Imagine a game where you roll a die twice. If the dice roll is 9 or greater then you get whatever you've bet returned plus that amount (1:1 payout), if the roll is less than 9 you get nothing. You can bet before the first roll, or after the first roll, but obviously cannot bet after the 2nd roll.

You cannot predict which outcome will result from any given roll, but you can still create a profitable system by waiting for a 6 on the first roll before betting. Work out the odds, you'll see how it works. There's no guaranteed win on any given roll, and there's no guarantee that there won't be a streak long enough to wipe out your pot, but given thousands of 6 rolls the odds will eventually regress back and give you your edge. It's just math.

Even tho the individual rolls are completely random (assuming a fair dice and a fair roll), you still have an edge.
  • #15
  • May 3, 2008 7:38am May 3, 2008 7:38am
Quoting tdion
Possible inefficiency: The rollover strategy Brijon talks about. Too early to tell, but there might be some meat on that bone.
yes. maybe there is something. The problem is the way to find WHEN the run up will start. Price can fall 3000 pips 1 hour before and then the run up starts. Using indicator is not a way for sure.

ps: i sent a pm for you about another thing
  • #16
  • May 3, 2008 7:44am May 3, 2008 7:44am
Quoting tdion
The first inefficiency (and only one that I am aware of) is trading nondealing desk broker spikes. It works like this. Take 10 brokers and compare their feeds. When one travels far away from all the others, take a trade.

It's likely that the price will return back to the others.
could you explain your view? Because it differ from what i read before. The way i know should be: watch the prices in the "real" feed data like Reuters, then put the trades on the MMs as they are lag. So you will trade in the MMs accounts, not the contrary. But i read that it don't work anymore.
  • #17
  • May 3, 2008 7:51am May 3, 2008 7:51am
Quoting tdion
A second type of inefficiency is "stop hunting." Basically, you find out where the stops on currencies are based on a news service. When there is a lot of news about a particular stop level, you can watch for those stops to be broken by the banks as they search for liquidity. If stop levels are breached, often price will significantly rise or fall past those levels (hence, support and resistance as chartists say.)

These are the basics.... I don't have details, but rumor has it that there are some folks trading this way.
could you give a example? this is not the way i though it works. Maybe is my english and the meaning of "news service"

thanks in advance.
  • #18
  • May 3, 2008 7:53am May 3, 2008 7:53am
Quoting dito
4th inefficiency : space-time analysis applied to the market.
i suppos you not talking about Gann. Could you explain that in detail?
  • #19
  • May 5, 2008 11:30am May 5, 2008 11:30am
Very interesting thread.

I have experimented with sudden spikes but I found that the broker which makes these artificial spikes never filled my orders without slipping me in my disadvantage and making this strategy worthless.
  • #20
  • May 5, 2008 10:00pm May 5, 2008 10:00pm
Hello FF,

I have been a completly passiv member of ff, however I did and do enjoy the highest qualitly of this forum.
I have been active in the markets for a couple of years, and finally got the guts to speak up in this excellent forum.

Efficient Market Theory:
Is the market random? Well do traders trade randomly? My guess is no. Furthermore there is to my thoughts a connection between the market movement and the behavoir of traders.

Market Inefficients:
I believe I have certain knowledge of market inefficiencies, and have traded certain market inefficiencies. What are market inefficincies? Market inefficincies is a systematic behavoir which is finit.
A example could be Merger Arb., CEO are under pressure, thus they try to land a coupe by taking over a company, often paying a too high price. These flaws/irrational behavoir/inefficiencies get noticed by someone, and he/she makes a ton load of money. Is Merger Arb. coming to an end? Probably!
There are probably many inefficieny in the global markets, and it often boils down to human behavoir, at least my studies give me that gut feeling, however I have not yet found the proof!

I could go on and on forever, however I ll make my first stop here, feel free to challenge me in anyway you want.

Best regards
Oliver
0 traders viewing now