Archive for the 'Trading' Category
New month and a new trading plan. So first the post mortem. What went wrong with the last plans? Three things. 1) Volatility, 2) Reliance on Technical Indicators, 3) Lack of understanding of the fundamentals.
1) So first volatility. FOREX is volatile. period. It turns out that currencies trade in a quarter cent range most of the time and then jump ranges by quarter cent multiples. Quite often a whole cent. Now if you are using the old “keep your stops tight” adage as an attempt at minimizing risk you are most likely trying a five or ten pip stop loss. Well forget it. You stop loss must be outside the current price range so that it will only be hit if the currency revalues – not on minute to minute volatility. So for me that means the stop loss is not 30 pips. Assuming I am careful with order placement and I am coming in at the lower end of the price range then the stop will only be hit on a move against me by more than a whole price band.
2) Technical Indicators. Well after years of using them I have concluded they don’t work. Why? because if you offset them by a random number they also appear to provide support and resistance. They also ignore fundamentals. If your technical indicator tells you that support is right in the middle of a trading range you are screwed. Even pivot points and bollinger bands are dodgy.
3) Fundamentals. By this you need to think about what is happening. Banks value currency based on interest rates, GDP, Risk etc and come up with a maximum price they will pay for a currency as well as a view of whether they think it will gain in value or whether they want to reduce their exposure to it. The desire to lessen their exposure to risk is as big or even bigger than their desire for profit. Even if they don’t make money they are under pressure to not lose money. So banks come up with a price and a direction. Above this they will sell and below this they will buy. Banks don’t use weird numbers like 1.2913 Its either 1.29 or 1.2925. When the price changes it will change by a quarter(0.0025) a half(0.005) or a whole unit (0.01). Not some piddly little offbeat number like 0.0017. So avoid piddly little numbers – you are just setting yourself up to have some institutional trader sell to you at too high a price or buy from you at too low a price.
In the following chart I have placed in quarter and half cent lines (darker). It is obvious that these provide support and resistance. By entering at the lower side of a range and ensuring the stop loss is more than a whole band away from the current range should ensure that the stop will only be hit if the price changes bands rather than on spurious wobbles.
In fact it is these levels that cause pivot points to appear to work. The highs and lows are defined by quarter cent ranges. Which means that if it turns out that the highs and lows are a half or whole unit apart then the central pivot which is half way between them will end up being on a quarter unit line. So Pivot points appear to work when they happen to correspond to a quarter unit line.
The Plan
Determine the current range and place an order at the lower (or upper) side of it with a stop loss at least a trading band away from volatility. Trade in the direction of the underlying trend.
September 04 2010 | Trading | No Comments »
In this day of computers traders rely heavily upon technical indicators and I have to admit I have been a firm believer in them. To see the problem open up a chart of some reasonably liquid currency or future and draw a random line across the price. I bet that you will be able to find points where that line provides support and resistance.
The problem started with Bollinger bands. I found that I was keeping charts on several different time scales in order to determine what traders using the different time scales would be seeing. Then I noticed they all seemed to work. I am a pure maths type of person and that is not possible.
Then I discovered that if you offset the Pivot points by a random amount they also appear to work – for any random offset.
Okay now try moving a horizontal line down over a chart. I bet you will be able to find a support and resistance level somewhere on the chart at every value you place the line at.
So we have a problem here. Its an optical illusion of sorts. Your brain sees the points a reversal occurs at and flags the points that it doesn’t occur at as a “price broke resistance” level.
You may as well choose a random price and say that it will be a support level.
In fact I have pretty well convinced myself that it is impossible to predict support and resistance levels to any higher accuracy than choosing a random number.
Now trend information that is different. At the moment I am of the opinion that with some level of accuracy the broad direction of the market can be predicted within a small window about any time. That argument is based around a loose intermediate value theorem argument. If the price is currently going up at 10 pips per hour on average for it to reverse and start moving at -10 pips per hour the average price will have to go through 0 first.
So how do people trade off indicators? Most likely good money management.
Pretty well screws up all my trading plans that were based on entering at particular support levels.
September 02 2010 | Trading | No Comments »
The tendency for traders to turn to scalping as a method is interesting in its own right. I am tending towards the opinion that traders converge towards a scalping method as the iteratively improve their trading plan in order to optimize their entries and reduce their risk. As a result there appears to be a collection of rules that through much pain you eventually end up with.
To date we have.
- Risk Reward needs to be greater than or equal to 1:3. Anything else makes it hard to pull ahead. Even at 1:2 you are only ahead by your at risk amount assuming a 50% fail rate.
- Keep your stop loss a bit larger than 1 standard deviation on the time scale you are scalping under.
- Don’t change your per trade risk. Scalping is a numbers game. It relies upon a sequence of trades to work. If for example you have a 5 pip stop on one trade and a 10 pip stop on the next then you have just screwed up your risk reward ratio. The loss on that single 10 pip trade is the same as two consecutive 5 pip losses. You can increase your take profit amount for the same reason.
- Use Stop Orders for trending markets. Limit Orders for range bound markets.
- Let the price come to you.
- Less is more. The more indicators you have the more conflicting signals you will see. If the information provided by an indicator is already visible on your chart don’t use it. If it conflicts with other indicators, don’t use it. The hard part of this game is reading entries. The whole point of using indicators is to filter out unwanted possibilities and to highlight good possibilities. The last thing you want to do is to increase the number of possibilities and hence your own internal confusion.
- Support becomes resistance and vice-versa in trending markets.
August 25 2010 | Trading | No Comments »
What order type to use. Limit Orders or Stop Orders. They are both different and have very different uses. Get it wrong for your trading time scale and market and you will be in a world of misery.
Limit Orders.
These are probably the most common used by newbies. The Limit order will enter if the price is at or better than the price you have selected. The big problem with a Limit order is that the price is by definition moving against the direction of the order when it is opened. Secondly we all use stop losses correct? Stand back and think about this for a minute. the limit order together with the stop loss order defines a narrow range within which you are making the assertion that the price will not only enter but will also turn around. Is your analysis that perfect? Have you allowed for volatility?
Keeping in mind the random forces acting on price this is almost impossible to do unless there is some form of brick wall the price will hit and not move past. It is effectively the same as throwing a wobbly dart at a small target. Good luck.
For Limit Orders to work they require a reasonably large stop loss together with a major support or resistance level. But even with something like an outer Bollinger band or a pivot point you will still need to allow for a standard deviation or more for your stop loss.
They require a high level of skill to use but if you get it right they are also the most profitable.
Use with range bound markets with clear support and resistance levels.
Stop Orders
Stop Orders open when the price crosses the line. They act as a trip wire. Unlike limit orders they assume the price is moving in the direction you have chosen.
The problems with Stop Orders are slippage, the amount the price moves between when the line is crossed and when the computers actually manage to get the order into the market. This can be huge on economic announcements or illiquid markets. They also don’t get the best possible price.
This has to be traded off against not getting any price at all. It is very easy to place a limit order and simply not be hit at all.
The caution with stop orders is when they are placed just outside a convergence zone. When breakouts occur the price has a habit of kicking back to the support level within a tick or so of the breakout – not talking 5 minute ticks here, am talking sub second tick by tick charts. So care has to be taken to ensure the stop loss is clear of the convergence area.
You will need to check with your provider to see how slippage affects your stop and take profit levels. Will they be placed at offsets from your original order or offset from the actual price obtained?
Use with trending markets. Care to ensure your stop loss is not likely to be hit on the inevitable micro retracements.
Market Orders
Don’t use them. Slippage is likely to be large from when you click your mouse and the order eventually makes it into the market. The big issue is your stop losses and take profits will be offset from where the price was when your order ht the market not where they were when you clicked your mouse.
August 25 2010 | Trading | No Comments »
Not even the end of the month and I am changing the plan again. OK what went wrong. The last plan aimed at capitalizing upon the most profitable moves of the day by looking for the set up on the EURUSD pair on London Open. It used a a couple of sets of Bollinger Bands to define the days trading range and relied upon trading the London session from the support/resistance that the market had settled at after the Asian session toward greatest probability.
This in itself was good. In fact still is. The problem is that we have had a skittish market of late. The broader trend of the last few months has ended and we have had a slew of economic announcements that pretty well sidelined me. When viewed in hindsight the plan works well. In practice it had been hard to read which level it is settling out at. It pretty well sets me up for one trade a day which if I miss for whatever reason leaves me with no income.
So what I need is a more flexible method that is less dependent upon economic context that can handle some volatility and that allows more trading opportunities. I also need to do something about dead hours during the day which just doesn’t really work.
So for that we turn to scalping. I have done some, in fact a lot of analysis about volatility and risk reward ratios and have pretty well come to the conclusion that risk/reward is independent of time scale. As mentioned in another post the whole concept of “buy low” “sell high” boils down to buying at the extremes of the Gaussian distribution below average price and selling at the upper end of the distribution above the price. This has to be combined with an allowance for “wiggle room”. Both of these things are defined in terms of a single quantity, the standard deviation. The minimum stop loss needs to be about 1 standard deviation and the maximum reward is a trip from one side of the bell shaped curve to the other, which is about 4 standard deviations giving you a maximum risk reward ratio independent of time of 1:4 .
Or more practically I found I needed a stop loss of about 25 pips to pick up the average London open move of 100 pips. But I could have obtained the same risk reward with a trade of 5 pips as a stop and 20 pips as a reward. The difference is that there are far more 20 pip moves per day than there are 100 pip moves.
You can work through the math, but it works out that if you divide your daily risk into many smaller trades the probability of a loss on the day assuming a risk reward ration of 1:3 is almost negligible. In fact if your per trade probability of a win loss is 50% then the probability of 10 successive losses is
Whereas if you put it all on one trade the probability is 50%.
The Plan
Well that’s enough for background. The new plan is in a sense the same as the old. It keeps the different behavior during different session and keeps the use of broad Bollinger bands. The difference is that it uses Bollinger bands that have been set to show what a 15 minute trader would see. In other words 20 x 15 minutes which is 300 ticks on a 1 minute chart. It also overlays the pivot points.
It uses normal bars rather than candles because I am looking for convergence rather than candle formations.
This chart is of the Australian SPI200 cash, but it also applies with the EURUSD etc.
The general idea is to scalp through the day on the SPI200 and then in the evening pick up the first couple of hours of London open trade on the EURUSD.
I am assuming say 10 trades per day, risk reward is small. The stop is about 5 pips including spread for the SPI200 (1 pip spread) and about 7 for the EURUSD (2 pip spread). Reward is correspondingly tight using a risk reward of 1:3.
The at risk amount for the day is the same as before – (not posted here) but is spread out over an assumed 10 trades per day. Risk for the day is a small percentage of total capital with returns compounded back in.
I have modeled it out using a spreadsheet assuming 10 trades a day over a couple of months using a coin toss technique for success or not per trade. It breaks even more or less at a 70% per trade fail rate. Provides a reasonable income at 50% fail rate and is massively profitable with a 30% failure rate.
The key driver is consistent income rather then the promise of get rich quick. Its a job not a casino.
August 21 2010 | Trading | No Comments »
Here is an interesting one to put into the debate about trading time scales. Does it matter what time scale people trade? The answer it turns out is strangely enough a no. The reason ends up being the old bell shaped curve. To trade effectively you will need to use a stop loss which has to be placed outside the normal volatility range of the price action in order to give the price some wiggle room. Technically this means the stop has to be outside one standard deviation from the price. A standard deviation is a measure of how much something deviates from average. The price action also moves according to a normal statistical distribution – the “bell shaped” or Gaussian curve. That means it moves around a mean by about two standard deviations each way.
So if we assume you buy low and sell high, then the best you can reasonably expect is for the price to move from one side of the bell shaped curve to the other. In other words to make four standard deviations. Your risk will be one standard deviation which gives a risk/reward ratio of 1:4 independent of time scale.
Now of course there are exceptions. You may be lucky enough to pick up the beginning of a longer trend. Remember though that this is luck. You cant know in advance that you have picked up that exact price and of course if it starts trending then it will affect all time frames.
The following pictures of the EURUSD demonstrate. The first is on a one minute time frame, the second a 15 minute time frame. Notice how if I didn’t tell you the time frame they pretty well all look the same. But more importantly notice the relationship between how much “wiggle room” you would need to allow compared to reasonable reward. The coloration of the background reflects the session dominating.
The same applies to daily and even weekly charts. Interesting.
1 Minute EURUSD
15 Minute EURUSD
August 18 2010 | Trading | No Comments »
3 September 2010: Having recovered from the discovery that pivot points and Bollingers despite their popularity don’t actually work have made a discovery that does. Well at least on forex. Quarter number intervals.
If you think about what is happening with Forex it is obvious. A bank will set a price for say the EURUSD of say 1.2900. They wont be saying 1.2913. They might say 1.2925 or 1.2950. Someone trading for the bank will have a maximum price set for them by management that they can pay for a currency unit. Under this level the bank considers the currency is cheap and above which the bank can sell the currency at a profit.
Popped a grid of quarter cent lines around the price up on the chart and guess what – it fitted. So where now. Well after last nights mistake of trying to fight the chop around an announcement will stay out. Its non farm payrolls again.
But longer term the trick seems to be to identify the half cent or so range that the currency is locked into – staying a long way away from the middle where all the volatility is. And guess what I have been doing – placing my orders right into the middle of the trading ranges thinking I have picked a top or bottom. Hence why I get stopped out so much.
Instead I should place orders at the extremes of the trading range. Pretty obvious huh – damn I hate hindsight. Why cant I have the hindsight before I try something.
continue reading »
August 09 2010 | Trading | No Comments »
Looking back over this blog it has been three months since I wrote a post bemoaning the massive Learning Curve associated with Forex trading.
Since then I have been through a couple of trading plans, have beaten the psychological hurdle of actually placing a live trade – imagine trying to reach into some complicated machinery with spinning gears and slamming doors to grab a coin, make even the slightest error and its going to hurt…
Have made numerous stupid errors: Traded against the trend. Canceled good orders minutes before the market puts in a solid 150 pip rise. Gone to the bathroom just when the price has decided to move. Have astutely watched the market go nowhere for hours on end to nod off just when it wakes up. You name it, I have done it.
Now that I have finally reached the point where I can with reasonable accuracy make some sense out of the EURUSD pair it is worth going back and having a look at where I started.
The problem of making money boils down to determining the time and direction of the larger moves in the following chart. Ensuring that the price does not move against you by the smallest possible amount after you enter and ensuring that you get out before it changes direction on you.
Easy huh….
The second chart is the same time period, but this time marked up.
Notice how the larger moves tend to occur on session boundaries. Normally Green to Blue (London Open) or Blue to Pink (New York) . This takes care of the “when” the move will occur.
See how the price moves tend to occur between the bands across the chart. That takes care of the minimum and maximum price as well as the direction and for that matter the stop loss distance.
I challenge you to go back to the original chart and find the same information.
As a rule set. The EURUSD Forex market is a random process about a daily average price. It trades between support and resistance levels situated 1, 2 and occasionally 3 standard deviations from the daily mean. It trades in the direction of greatest opportunity around this trend line. It always returns to this mean. It puts in its largest moves from session open (unless there is major economic news expected) .
August 08 2010 | Trading | No Comments »
All traders should use stop losses. Period. We all know that and we all probably know the feeling of seeing the market reach down, hit your stop level and then reverse up to make some massive new high.
In fact it seems that no matter what you do the market will find a way to stop you out and take your money from you. Sound familiar?
From a naive view you would think placing an order would have a 50% probability of being right or wrong. You place an order and the market will go either up or down. In fact based on this you would expect a coin toss strategy of tossing a coin and going long or short would be successful. If you are wrong you lose a fixed amount. If you are right you would win more than this. Over time you would pull ahead.
The reason this does not work is conditional probabilities. A successful trade is a trade that “Goes in The direction you choose” AND “Does NOT cross your stop loss on the way there”. To put this in the naive picture – and the math is wrong by the way, but it is only a picture. This is the
(Probability of NOT stopping )
(Probability of going where you want).
which if you think of your high school maths is
(1-Probability of stopping) (Probability of going where you want).
So naively if there is a 50% probability of it going one way or another from one moment to another you would have something like a
(1-0.5) (0.5)=0.25 =25%
probability of a successful trade. Far worse than a simple up or down question.
Drunkards Walk
Now in truth the probability depends upon the proximity to your price. In fact you could assume a Gaussian from moment to moment if you wanted to delve into the math. The math is non trivial by the way and depends on tick size and all sorts of other things. To do it properly you have to assume a Gaussian at every tick and look at the probabilities after a sequence of ticks which soon brings in the question of how many ticks, how much it will move from tick to tick etc etc and the problem gets bigger and bigger. It ends up being Einstein’s drunkards walk problem or in fact Einstein’s drunkards walk across a sloping hill side where you want the drunkard to get to his destination without bumping into some arbitrarily placed wall on the way. Needless to say if where you want the drunkard to get is against the slope of the hill you are screwed before you start.
Suffice to say the closer your stop is to your price the higher the probability that the price will hit it and the lower the probability of a successful trade.
In fact the way things work out is that when using stop losses there are far more ways to lose money than there are to make it. The closer your stop is to your price the higher the probability of it being hit. The benefit of course is that your loss is limited which is a lot better than the 50% probability of a HUGE loss if you didn’t have a stop at all. When trading Forex and futures you are using someone else’s money so the option of riding out a loss in the hope it will turn out in your favor simply does not exist.
The Solution
To beat this you need to change your thinking to make your stop placement the key thing that will make the difference between a winning trade and a losing trade. You like me probably look at the price action from a view of where the price currently is and where it could go. That is you are looking at how much you could make.
You are only going to make this amount of money IF the price does not hit your stop loss on the way there. So what you have to do is look at where the price is NOT going to go and put your stop there. In simple terms you have to optimise the relation
(Probability of NOT stopping ) (Probability of going where you want).
There are lots of ways of doing this. Placing your stop below the recent high or low opposite to where you are trading. Placing it outside the Bollinger bands against the direction you are trading. Using a fixed amount that is large with respect to the Average True Range over the time period you expect to be in the trade. Placing your stop behind a support or resistance level.
No matter what you do you need to have your stop where it will be hit ONLY if your are wrong about the trend of the market.
What you can’t use is the old adage of “Keep Your Stops Tight” because that equates to almost a 100% probability of being stopped out.
This is counter intuitive. Your natural desire is to minimize your loss which in fact will cause you to place your stop close to the price increasing your probability of a loss.
So next time you look at a chart look for two things. Where the price could go and where it couldn’t. Where it could go is your target and where it couldn’t is your stop.
August 07 2010 | Trading | No Comments »
Well the previous plan was a dud. Essentially the problem was too many false signals and a lack of a clear view of where prices where with respect to “Is Cheap” and “Is Expensive”. The stochastics provided timing but not enough information about where prices were relative to the days trade.
The good part of the old plan was the session boundaries.
I was also playing with longer term stochastics to give a broader trend direction which was sort of there, but not quite.
The new plan keeps with the concept of statistical moves but uses Bollinger bands rather than stochastics. It is far more visual as well.
In the chart below the first thing you notice are the coloured bands. What they are are the different sessions. Yellow is Sydney, Green is Hong Kong, Blue is London and Red is New York. The Grey after the Red is when no market is open. It can be seen that the biggest price moves occur in particular sessions. Normally London but this week New York has been figuring heavily.
On top of this are a big set of three Bollinger Bands. These are set to collect the last 24 hours trading data which is 288 ticks on a five minute chart at 1, 2 and 3 standard deviation intervals. You can see they provide good support and resistance lines for the intra day action. What I was missing was context of where prices were in terms of cheap or expensive. This fills that in.
Also on the chart is a five tick Heikin Ashi inside a normal set of 20 tick Bollinger bands for entry exit timing.
The idea is to trade the larger momentum moves that occur during the London Session (Blue). New York opens at about 11:30 PM my time and as much as I seem to like staring at a computer, following a trade through from midnight to NY close is out.
Also on the chart at the bottom is a 1.5 times the standard deviation with a 200 tick moving average through it to help with stop loss sizing.
In words it boils down to start an hour or so before London open. Wait for the market to pull back to a support level – it always seems to. Not sure why, could be Asia closing, could be people setting up a stop hunt, could be anything. Place the order and wait. Exit when it is on or close to the normal (2 Std Devs) Bollinger bands.
The third set in grey are a guide to an overcooked market – likely to reverse.
Pivot points are also there as a guide to other resistance levels.
August 06 2010 | Trading | No Comments »
Next »
