I have been battling with stops every since I started trading. Stops can be so arbitrary. If you look at any chart, most likely you will find that there are many reasons for choosing or not choosing a particular stop level.
Then there is the eternal question to be answered: Mechanical vs mental stops. It seems like both work against you: Place a mechanical stop too tight, and you get stopped out. Place it too loose, and the more chance you have your stop will fall victim of mean reversion and you get stopped out at the worst possible price. Place a mental stop, and you watch the price blow through your stop before you have the time to act accordingly.
And finally, put stops at text book areas on the chart, and you will most certainly get stopped out, and after a few days you realize you were stopped out at the extreme.
Do other traders get these feelings or is it only me?
Statistically the effects of stop losses can be mitigated by also placing profit targets with the same technique, however, the larger the reward to risk, the less effective this technique becomes. Leaning towards Position and Swing Trading, this technique definitely has a big downside that it often results in premature exits on profitable trades.
Long story short, I have reconciled my stop technique to contain a mix of several techniques, and this seems to be working quite well for me now.
First of all, I use my stop for position sizing. This is basically the fixed risk method where the difference between your trade price and your stop determines the size of your position. This is fine as far as confidence is concerned, but as shown in a previous post, just this on its own may lead to wild fluctuations in position sizes which can affect the consistency of your equity curve.
Now, if that stop is calculated solely on the basis of volatility, say a multiple of the ATR (Average True Range), then the position sizing model becomes a volatility sizing model, which seems to perform well when compared to other models.
However, in most cases this volatility model puts a stop in nonsensical places on the chart, so a discretionary element is required. It won’t make sense to use a volatility stop if that stop is a few cents above a long term support line, for example.
My simple solution is to address as many of these issues as possible into one, compromising, stop technique. I place a stop on a place in the chart where it would make sense and consider that to be my mental stop. If that stop is breached, I will be following the price intently to see whether it was a fake out or whether there will be a continuation. I also place a hard stop 1 ATR below this level. The position size is determined on the distance between this hard stop and the entry price, hence it has both a volatility element as well as a proper risk element which is determined according to your stop. This method allows larger position sizing when closer to a turning point, which is consistent with the theory that you should place your biggest bets where you have the best risk/reward scenarios. Conversely it will raise a warning sign that you may be too late for a sensible entry if the position size results in too small a figure.
I find that I usually rarely hit the hard stop, so that becomes a true worst case scenario, which I am happy with as I consider hard stops as a necessary evil. However this level has a great impact on the position size which in risk terms is a crucial part of the equation.
Filed under: Trading Methodology | Leave a Comment
Under the DTN (IQ Feed) / Amibroker Platform I currently use, I am limited to 500 streaming symbols. I have chosen to use this limit as my “Market Universe Basket”, with several categorizations and watchlists defining various subsections.
From these 500 instruments, approx 100 are various indices and indicators, which leave me with around 400 stocks which are categorized by Sector and Industry. The requirement for a stock to enter my list is that it
Either (1) it is a good representative of the industry it is in or (2) is of special interest.
I do not try to include all industries, but do try as much as possible to keep this list balanced, with roughly the same number of stocks in each industry being followed.
As this list is usually full, a new entry would usually need to replace an older one.
In Amiboker, it is possible to navigate through this list by sector, industry. At this level, I use the built in classification of Amibroker to do this. I also have a set of explorations I use which give me an overall picture of what’s generally going on. These are much like Finviz’s maps as well as filtered lists such as:
How many instruments are trading above/below VWAP
How many instruments are trading above their most recent swing high.
.. and others.
This Universe basket will also supply my watchlists. As I mentioned in previous blogs, I do look, and give lots of importance to fundamentals (not limited to earnings reports but can pertain to any information about the stock which is not purely price/volume based). My watchlists are also a basis for what I call “Trading Lines” in my fund: Basically this are portfolios within the main fund for which I can track and report performance. First an instrument enters my watchlist, then it (maybe) is eventually traded, and the performance from that trade is tracked by a trading line of the same name as the watchlist.
These watchlists/trading lines are:
Growth: (Long Only) Companies which have performed well in the past and are expected to continue performing well.
Value: (Long Only) Companies which have either not performed well in the recent past, or have performed well but are out of favor, but are expected to turn around a corner in the near future. In order for a company to enter this watchlist, a valuation study must be performed which shows that the stock is cheap when compared to the general market.
Dividend: (Long Only) Companies with good fundamentals which offer a high dividend return. Typically stocks of such companies act differently to the overall market, and are also valued differently, offering an interesting diversification point from the Growth and Value watchlists.
Special Situations: (Long or Short) Companies, Commodities or Currencies which are experiencing a temporary shift in perception (can be both a positive or negative perception) which may offer a temporary opportunity in the market. Because commodity and currency markets are a zero sum game I consider all trades in these instruments as special situations.
Technical: (Long or Short) Where there is no other compelling reason to enter a trade other than from a pattern recognition perspective. Trades in this line are also executed in order to balance the overall portfolio (e.g. Shorting the SPY at technically sound levels when heavily long on the other trading lines) There are also some subdivisions of this line which include some automated programs I sometimes play with, but which I have not recently worked on due to time constraints.
Alternative: (Any direction) If I have a new idea which doesn’t fit into anything above, it starts off as a sub section of this line. This is where I dabble with more complex trades such as options calendar spreads and plays on volatility (which i am relatively new to). If a new methodology which is tangible and consistently profitable emerges from here, it will eventually take life as a trading line of it’s own.
I do not limit these lines by a fixed capital amount, but by assigning a certain amount of risk to each trading line. I work with a total “Risk Budget” which comes in the form of Fixed Income from my Bond Portfolio and then split into these trading lines as appropriate. I will explain this methodology in more detail in a future post.
Filed under: Amibroker, Trading Methodology | Leave a Comment
Starting Again
I started this blog with the intention of having an online trading diary. I have not kept this up. My bad. I guess I tried to write for others where I should have actually written for myself. I’m going to give this another shot. If it works out, I’ll carry on. If not, I’ll shut down this blog.
During this past year I have gone a transformation as a trader. For several years I had dipped my toe into various styles, ranging from buy and hold to intently following level II. I tried my hand at fully automated trading and fully discretionary trading and explored the various combinations in between. I’ve had some moderate successes and some failures, I followed some excellent traders and some not-so-good ones. Somehow in all the styles I have tried, there was always something which didn’t quite fit the bill for me.
Now, I have slowed down my search and somewhat feel at peace with what I am presently doing. I have found a trading style which jives with me. It’s a rather strange feeling 
Not unlike finding the right partner to share your life with. You don’t really know why feel like you can live your life with someone, but you just know you do.
What I would like to do in the next few posts is to try to journal what I am doing. It’s really for my sake more than anything else. The reason I want to post it is for reasons I think I mentioned before: writing publicly forces you to get your thoughts in order and write properly, opens your thoughts to outside critique, and somewhat strengthens commitment.
I cannot imagine I would be posting very frequently to this blog. Writing is not my strength. I will also try to use twitter as much as possible as an on-the-fly journal.
Just as an brief intro to the way I am now trading, here are the main points:
1. The majority of my capital is in corporate bonds and other fixed income instruments. I do not trade these and anything I enter is intended to be held to maturity. This fixed income portion provides a considerably constant and foreseeable income and supplies the risk money for my trading.
2. The risk money earned by the fixed income portion is used to build a “risk budget” for my actual trading. For this portion, I trade mostly equities. My philosophy is to aim to leverage every little bit of information from the market as possible to gain an edge. While I have found some success with pure technical trading, I have never been able to achieve the performance I hoped for. Possibly one of the main reasons for my lack of performance in this regard could be purely psychological in that I was never able to trade with enough conviction if I did not have a fundamental reason to do so. I am an investor, not a trader, at heart, and the reason I trade is to be able to model my equity curve to my risk requirements, and not to the fluctuations of the market. If I could invest in a position and hold it forever, I would, but the market seldom gives us the opportunity to do so without substantial drawdown in the process. To summarize, I enter a position for fundamental reasons, but time the entry according to my evaluation of supply and demand (hence technical). Exits and position sizing are very much technical based and are really a function of my risk control structure.
3. While I do not run a legal fund entity, I account for my trading operation as a regular fund. I have accepted some funds from some close relatives and friends and I provide them with a monthly NAV report and short commentary. I have done this in order to increase my discipline and accountability.
4. I have come to terms with the fact that I cannot look at the market on a full time basis. The most damaging thing I have found during the years I was exploring various styles, was looking at the market at irregular intervals (such as watching the market on a minute – by – minute basis when I had the time, but then not look at it at all for several days). I have found one of the most important thing is to stick to a ‘pace’. A chosen rhythm if you will. Now i try to view the markets on a daily basis with most of my research being done on weekends. Four time frames matter the most to me – the weekly timeframe gives me a clear indication of the overall price history, a daily timeframe for pattern recognition, a 30 minute timeframe allows me to define precise entry points, stops and allows me to calculate risk, and a 5 minute time frame is just to get a clear reference of the intraday structure of the past few days. I will trade out of the 30 minute chart, with reference to the longer timeframes, rather than to the 5 minute. The greatest challenge using this technique is sizing positions adequately in order to filter out what needs to be considered noise in this style of trading.
5. I will try to hold a position as long as I possibly can without compromising my risk profile. My risk is evaluated from day to day, and is not related to the current P&L provided by that position. Basically at any point in time a position has a specific perceived upside and downside and a position is exited if the probability to the downside exceeds that of the upside, regardless of how it performed in the past.
This is more or less it. I will try to expand as much as possible on these topics in my next posts, and will try to keep a day-to-day journal of my thoughts and trades on twitter.
Filed under: Trading Business | Leave a Comment
It’s been a long absence since my last post. I cannot say I’m a proficient blogger. It’s also been an extremely busy time for me during the past few months.
Anyway, i’m here now, and I’ve got a follow up to my previous study: A Peek into a Professional Trader’s equity curve. Originally, I set out to see what a professional trader’s equity would look like. What was the performance? what are the drawdowns? As a byproduct however, I also included 2 position sizing models to investigate the impact position sizing would have on the curve. The methods were: Fixed Size and Fixed Risk. All in all I felt that the results were not very far apart and both methods had their pros and cons. In some of the comments I received, I was asked to try out a percent volatility model.
I have done so, and the result is rather good. The study is based on the first theoretical one (No slippage assumed) so it must be compared to this. Just for clarity, the original Fixed Position study posted a 29.91% profit with a maximum drawdown of 5.90%, and the Fixed Risk study posted a 24% profit with a maximum drawdown of 6.15%.
The following percent volatility model was calculated by using an ATR over the past 300 minutes. A constant was used in order to bring the values close to the original exposure, as was done with the fixed risk model. This study posted a 32.63% profit with a maximum drawdown of 5.72% That’s definitely the best of the lot. Here are the detailed results (Click for a larger pix):
Filed under: Amibroker, Trading Methodology, Uncategorized | 1 Comment
Part II: Profit Distributions
While I am working on the Percent Volatility model which GHG suggested I work on, in a comment on my previous post, I am extending yesterday’s study with an examination of the profit distributions for each position sizing model. Why? Because I believe that the nature of profits/losses that come with every trade have a big impact on the emotional swings of a trader. Ultimately that’s what position sizing is all about. No position sizing technique will translate a losing edge into a winning one or vice versa. It simply molds the results into something more or less acceptable to the individual trader.
(The following charts are based on Study 1)
First of all, let us examine the Profit Distribution for the Fixed Position sizing model:
Distribution of Trade Results, using the Fixed Position Size Model:
Here we can see that the results follow very much a standard bell curve, with most of the trades resulting in a profit between –200 and + 450 (With the most common result being around the +100$ mark) There are a few outliers on the negative side, and slightly more outliers on the positive side than the negative side.
For the fixed risk model, we will examine the position size distribution as well as the profit distribution.
Distribution of Position Size, using Fixed Risk Model:
What I thought was interesting in the above chart is that the the position size varies wildly – from minuscule positions of under 400$ to a very large position of over 40,000$. It leads me to question, is fixed risk really fixed? Is it wise to put in a 80% of your capital into a single position into one instrument, just to follow the model? Are you exposing yourself to unknown risks by trying to contain known risk? Would it be a good idea to cap the sizing and introduce an upper and/or lower limit?
Distribution of Trade Results, using Fixed Risk Model:
As expected the results are more contained and there are considerably less outliers on both sides. Interestingly enough however, the fact that there are less swings does not lead to a smoother equity curve: Compare the equity curves in the previous post – they are pretty similar.
A series of 5 losses of 100$ still equate to the same 500$ drawdown as a string of losses of 10$, 300$, 90$, 30$, 70$. This may really sound obvious, but there is so much ‘conventional wisdom’ out there that different sizing models make a huge impact overall. At least this exercise is showing otherwise.
I am curious to see the results of the Percent Volatility Model. Maybe it will be a good compromise between the two models we have already seen? I hope I’ll be able to get around the technical hurdles and get it done…
Filed under: Amibroker, Trading Methodology, Uncategorized | 2 Comments
Being a risk averse person, I could not get a feeling of what sort of draw downs are acceptable v.s. what isn’t. Every time I hit a losing streak, I had nobody out there to tell me “hey this is normal” or “this is way out and you should stop”. Invariable I would panic and bail out, pausing to rethink my strategies, only to get back into more or less the same game as ultimately I found nothing wrong in what I was doing.
This was annoying me. It was like driving a winding road, not knowing at each turn whether you are going to drop off a cliff or whether it was simply a difficult road to negotiate. Hence, I set out to attempt to shadow a professional trader in order to get an idea of what is acceptable, what isn’t, and most of all, what does an “equity curve with an edge” look like.
So in July ‘09 I subscribed to Trader Stewie ’s alerts. Stewie is a professional trader who trades for a living. This year he launched an email service where he would call out his trades by email.
The following post is a detailed study of all his trades called out from 1st July up until 16th October 09. I compiled it for two purposes.
1. To study how a professional trader makes his trades. How many trades per day? What was the duration? (Now most of you here will say that this is very specific, and all traders are different and have different methods, but that’s no excuse not to pick up at least one successful method and see at least one example of what really works.
2. To find out what is the most efficient way of following a trader’s calls. Evidently it is practically impossible to match a trader’s performance by shadowing him. There will be delays, slippage, worse fills, and this gets worse the shorter term the trader is. In fact Stewie is a short term trader (at least to my standards). His trades average a holding time of 100 minutes (as you will see in this study), so timely entries following his trades are a challenge. So what is the best way? Limit orders and risking not taking the trade due to being late? Market orders and getting worse fills? Fixed Dollar Position sizing, Fixed Risk? other methods?
Ok so let’s start. All trades have been entered into Amibroker using a modified version of my simulator. The studies are based on 1 minute bar data. The studies will be as follows:
Study 1: All trades as published. Entries at published prices. The idea behind this study is to have a benchmark, and see what sort of performance figures are required for trading successfully full time. I have taken a portfolio of 50,000 USD as my benchmarks. Are the returns enough to live on? Can you do with less or do you need more?
Study 2: All trades as published. Entries at published prices. However, since the stops are executed at market prices, I will assume an 0.02c slippage on exit. We will see how this difference will impact the results in Study 1.
Study 3: Entries using the opening bar price of the following bar of the time the email is received. This is an attempt to simulate a real life scenario of what a market order would look like if the orders are taken within 1 minute of receiving the email. (Note: This does not mean that all trades are 1 minute after the email is received, as the email can be received any time during the previous bar, including right at the end of that minute bar, but it is the best I can do.). Stops are at published prices less 0.02c slippage.
Study 4: Entries using the published price, but checked for validity the following bar after the time the email is received. This is an attempt to simulate a limit order rather than a market order. The expectation is that the entry prices are better, but some trades would be missed as the entry will not get hit.
For each study we will attempt 2 variants: 1a) Fixed Risk Position Sizing and 1b) Fixed size position sizing, in order to see what impact these two techniques have overall. For each study I will publish the standard Amibroker performance report, as well as the equity curve. I will comment on the figures I consider of interest.
To the races! Here we go:
Study 1a: All trades as advertised. Entries at published prices. Aka The Benchmark. The study has been compiled using a 50K Portfolio and taking a fixed position of 12.5K (25%) on each position.
Summary:
The results are impressive. At first glance, the equity curve is the sort of equity curve I would be comfortable trading. Notice the relatively shallow draw downs compared to the gains. I also see from the Capital at Work chart, that we were keeping an very large amount of cash, and a 10K position could have easily been traded on a 25K portfolio, which would have given us twice the performance figures.
The total profit was a respectable 30%, which projects to an annual return of 144%. During that period the 50K portfolio made almost 15K, so around 4.2K monthly. Must be noted that the position size or risk remained fixed throughout the test, and not adjusted according to current equity, which would have made for better performance but also for more risk.
192 trades were taken over an 80 workday period. That would be an average of 2-3 trades (round trips) a day.
The average holding period was 104 minutes.
The Win to Lose ratio is 60/40
The average loss is 265 while the average win is 312. The largest loss was 1074 and the largest win was 1270.
The Maximum drawdown was 3555, less than 6% of the capital. This is also only half of the average monthly profit of 12%, so all things equal this trader can be profitable month in-month out with ease.
Note on Long/Short Positions: admittedly there have been more longs than shorts, however the ratio shown in this test can be misleading as quite a number of the longs are actually long in inverse funds. Hence this metric does not represent a real picture of shorts vs. longs and cannot be taken into consideration.
Study 1b: Same as 1a, but using Fixed Risk Position Sizing. In order to equate to the same overall risk, I settled on a fixed risk of 330.
Summary:
The results are surprising to me, and somewhat contradictory. Firstly I must admit I thought that fixed Risk would have yielded superior results but in fact fact this methodology seems to underperform in certain areas.
The total profit was 25%, in comparison to 30% in the previous study. The maximum drawdown is slightly more, which implies that the system was not as efficient.
The average loss and profit are less than in the previous study, which would imply that the results are a little less volatile, but honestly when looking at both charts, I wonder whether the differences are purely by chance, and that this system would sometimes underperform and sometimes over perform the fixed position sizing method. The one good point about this is that the largest trade loss was 584, almost half of the previous test, but one must wonder whether protecting for these few outliers is worth the loss of performance when using this method.
I am not yet convinced that I should be throwing out the Fixed Risk model just yet, so I’ll retain this comparison for the next tests. I will publish side by side from this point onwards.
—————————————————————————————————————————
Study 2: Same as above, but with an 0.02c / share slippage on exit. This simulates a stop at market. 0.02 c is my best guess average. I think some stocks traded will do ok with 0.01c but several stocks are rather illiquid so I raised this to 0.02c to make the test more realistic. Intuitively, I would expect the fixed risk position size to do better this time, as less liquid, more volatile stops are assigned a smaller position, which also helps reduce slippage issues. Let’s see….
Fixed Position
Fixed Risk
Well, surely the system did not lose all it’s edge and is still tradable, however we can see that slippage of 0.02c shaved off a whopping 7% on the Fixed Position Model and 5% on the Fixed Risk model, out of our performance. All variables have headed down, with the maximum drawdown now at around 7%, but still ok, I think.
This to me brings in the harsh reality of the damage done when overtrading. The loss of performance here is 5-8% on 2-3 trades a day. If we traded 4-6 trades per day, this would have doubled, and already we would have turned one month into a losing month. It may not seem like much when trading one trade at a time, but when shown like this, the reality really sets in.
The Fixed Risk position has also gained some ground here. This model is still underperforming, but it lost less. The overall maximum drawdown is now less in the fixed risk model than the fixed position model.
In conclusion, I think this study represents a good picture of the sort of performance you can achieve as a trader.
—————————————————————————————————————————
Study 3: Entries using the opening bar price of the following bar of the time the email is received. (Simulating a market order entry with an arbitrary delay of up to 1 minute after the email is received)
Fixed Position
Fixed Risk
Further performance creep here, but that was to be expected. In fact, it demonstrates that indeed there is a short term edge on the original entries. If they had no edge, then entering a bar after would have not shown a significant impact as some would have better prices and some worse.
On the fixed position model, the maximum drawdown now is over 8.5%, so entering in this way would have shown a loss in one month. On the fixed risk model, the drawdown is 7.5%. Fixed risk is holding it’s own on this study.
—————————————————————————————————————————
Study 4: Entries using the published price, but checked for validity the following bar after the time the email is received. (Simulating a limit order entry)
Fixed Position
Fixed Risk
Well what do you know? Of course this study results in less trades as some trades are missed due to them not hitting the limit order. However, we can see that the result is better than the market order study. Honestly I was not expecting such an improvement though. I think what can be taken out of this study is that it is better to stick to original order entries than to chase prices upwards. Missing trades also results in less slippage losses because less trades are taken.
All in all, the Fixed Risk model produces very close results to the Position Model. Had I not performed these tests I would have thought differently.
—————————————————————————————————————————
Overall conclusion: My opinion is that following a (good) trader can be a worthwhile exercise, however it is very, very easy to lose tons of performance if entries are carelessly shadowed and one allows too much slippage to creep in. The hope is that, after some time shadowing a trader like Stewie one can start slowly putting in one’s own ideas and compare performances until one can do it on their own.
Many thanks to Trader Stewie for allowing me to put his work out in the open for an honest evaluation of his trades. Honestly, it is not easy to find somebody as open and as willing to share their full track record the way Stewie does.
Filed under: Amibroker, Trading Methodology, Uncategorized | 14 Comments
Note: This is not related to my typical style of trading. I always try to supplement my trading with longer term investing, and this post relates to this.
As I do not wish to do any disservice to anybody, (and also as I do not wish to get into trouble) I will not divulge the details of the stock advisory service I mention in this post. In fact the purpose of this post is really to highlight the general scene, and indeed it is not specific to the particular one featured in this post. The moral of my story is to ensure that any services you subscribe to come with a performance log and even if they do come, it is probably worth the effort of scrutinizing that log to see whether it is realistic to replicate the performance when following it.
About 2 years ago I subscribed to a certain stock advisory service. It was not cheap, had a good pedigree (this was not some penny stock touting letter) and had some excellent recommendations from several high profile writers. I prepaid a subscription for 2 years but I was never really able to follow it.. I don’t know why.. my nature (and the reason i trade for myself, and that I rely heavily on back testing any ideas I come across) is that I cannot bring myself to put money into something I am not totally comfortable with. Lately I have received a notice that my subscription is running out and I decided to go through all the data I received and see how I would have done if I followed it. Based on this I would decide whether to renew my subscription or abandon it.
While, during these 2 years, I did not follow the instructions the service gave out, I did read it in order to get some ideas on what is hot/what’s not in the current investing world. I was under the impression the service was doing great. Every email I received showed some recent wins such as
Stock X – Up 80% since recommendation
Stock Y – Up 73% since recommendation
and so on…
Sometimes, I’d also read through the email and come across phrases such as “we are up over 80% while the general market is up a measly 4.3 % to date”. However, the service did not provide a concrete performance evaluation.
This week I have been away from my trading desk and the markets, so I decided to evaluate the service properly. I picked up all the emails I had received to date, and picked up all the buys and sells. The emails are sent out approximately on a weekly basis after the close of a particular weekday. The service considers itself fully invested when it has 10 positions, and does not give out specific position sizes so I performed my test assuming a 10% position size for each trade. I used a modified version of my Amibroker Simulator to log a trade at the next day’s open, at the open price. I did not account for commissions or slippage. Here is the result of investing 10,000 dollars at the start of 2008, to date:
Some additional data:
The average holding period for the trades is 63 days, but this was anything from 3-4 days (the service advocates cutting losses short) up to over 200 days for winning trades (conventional wisdom is certainly present).
The total number of trades was between 50 and 60 during the 2 year period.
The service lost 35% in 2008 and gained 39% to date, in 2009.
Also worth noting is that this study takes this service through the testing times of the 2008 environment.
My Conclusion:
The service might not have been a disaster, but my question is, is it really worth the high price, and effort it takes to read through the letters and take the trades? Does it really have any performance characteristics which outperform simply holding on to an index ETF?
I am however most annoyed with the way the service advertises itself to it’s own subscribers in it’s own letters. Why tout all the winning trades and not mention the offset of the losing ones? Where on earth did the “we are up over 80%” come from? I really had to go over the data again and again as initially I was sure I made a mistake somewhere, but indeed I did take all the trades and to my knowledge I did not miss anything. I also feel for the subscribers who I am sure are left wondering why they are not able to replicate that performance themselves.
The bottom line is that we must really be careful about what we choose to follow. I was shocked to see a drawdown of almost 50%. I am sure that if I did follow the service I would have abandoned it at some point throughout that drawdown, never to recover. If a service is not able to give you a good indication of what sort of drawdown it would tolerate, and giving you the opportunity to scale your exposure accordingly according to that drawdown, then it’s probably a better idea to allocate some money into some ETF’s and manage it yourself – reducing and increasing exposure to the ETF’s when you deem appropriate.
…and most of all, if the service does not give an accurate performance history of what it does, it’s probably for a reason.
I would welcome any feedback, as always.
Filed under: Amibroker, Trading Methodology, Uncategorized | 3 Comments
Notes on Tape Reading
A couple of months ago I started a project to learn how to read the tape. I wanted to add another dimension to my trading by adding Level II and Time & Sales windows and trying to make sense of these. It took a while to get a feel, but as I move along this journey, things are starting to become a little more clear in this area. Here are some notes on my observations to date:
- I never fully understood why we should be trading “in play” stocks. Especially because my position sizing methodology will reduce the size of anything volatile, resulting in very much the same percentage performance of a stock which hardly moves. After (many) sessions watching the LII screen it suddenly dawned on me. The tape was much more clear. You could see the big players coming in. You could see when there was an order out there which upset the normal equilibrium present 99% of the time.
- The LII screen is “noise” 99% of the time. I cannot really figure out anything useful by watching the LII when there were roughly equal sizes on both sides. Equilibrium on your Level II display does not mean the price will not move, even at extreme levels.
- Support or Resistance can be identified when there is a huge order approaching Level I. Most of the times these orders appear only when that level is close. For example if there is a huge buy order at 34, then this order may only appear when the price approaches 34.03 and below, and disappears when it goes above that value. (Hence it needs to be watched closely and consistently)
- Very important is the way that order is executed when it gets to the battlefield. This is where there will be a true supply/demand imbalance. Say for example there is buy order for 100,000 shares about to hit level 1. On the sell side there are lots of 100, 200,500 share lots, but none the size of this one. What happens once it starts executing. If the 100,000 gets bitten into but the sellers are quickly exhausted, the price will bounce exactly off that level. If the sellers persist and eat out that entire order, or eat enough of it, fast enough, to change the buyers mind who promptly pulls the remaining order off the market then most likely support will break.
- I have read lots of material which states that Level II is not reliable because there is a lot of information which is not displayed (dark pools, fakes etc…), but I think the key to this is that you need to react on the tape you are seeing, knowing fully well that there are orders out there which you cannot see. Finally, Time and Sales don’t lie, so watching how LII is flowing into the Time and Sales window is also very helpful. Take the example above: if there was an equal, invisible, large order on the sell side, then you would see the buy order getting executed instantly and effortlessly, still indicating that support is about to break.
- Watching the Tape is more a risk minimization technique as opposed to a strategy. It will allow you to design entries with very tight stops.
- Because the tape is 99% noise, it is much riskier to use to exit a trade than it is to enter. I have had various instances when I nailed the entry to the cent, but shortly after “the tape stopped talking to me” and I could not determine an exit point in the same way. As mentioned before, when the tape is not showing much, it does not mean big price moves don’t occur.
- It is very helpful to watch a classic 2 column Level II window in conjunction with a DOM style price ladder which shows the sell orders above the current price and the buy orders below the current price. This gives a much clearer view of the spread of the orders. LII does not visually show a difference (apart from the numbers) if the next order is 2c or 20c away.
Here is an example of Interactive Brokers’ DOM and Integrated Stock Window showing Level II and Chart. Your own limit orders can be placed by point-and-click on either window. Your own orders appear highlighted in relation to other orders.
Comments welcome! I am sure I have a long way to go before becoming a proficient tape reader.
Filed under: Trading Methodology | 1 Comment
A resurrection
I have not been posting to this blog lately because I have lost sight of it’s original purpose, which was to open myself to criticism and ridicule(thank you for putting me back on track Anne Marie), and instead I turned it into carefully thought articles on technical stuff. Well I need to change this, so from this point forward, hopefully this blog will be more frequent, more spontaneous, will have more bad grammer, and most of all I hope it will inspire some feedback and discussion. So I will start with a note to self which resulted from some brooding this morning. I want to look back at this blog in a year’s time and see how I evolved as a trader.
Filed under: Uncategorized | Leave a Comment
Making Sense of Volume Part II
First of all, thank you Trader Stewie who’s subscription service is proving to be very valuable, both for an incredible number of successful trades he keeps churning out, and also for being a downright honest educator. Both instruments displayed below have been a result of his alerts. Following what he does is proving to be most enlightening.
In the original post Making Sense of Volume I attempted to create an indicator which speaks to me a little more than the standard Volume histogram as found on most charts.
When I create indicators I find that I tend to ditch most of them after a while. Not many stick, but this one really seems to work nicely. It really has one purpose: to efficiently identify if a stock is making move with heavy volume relative to it’s norm.
I find there are two main issues when looking at a standard volume chart: it does not clearly show whether the volume is heavy relative to the time of day (intraday volume has a nature “smile” – heavy volume at the start and end of the day with the lowest volume being roughly around the middle), and it does not clearly show whether the high / low volume is a result of the normal volume volatility of the instrument, or whether it is really a sign of heavy volume.
Both are of course possible to detect by eyeballing a normal chart, but volume studies are all about speed of detection. You want to detect signs of strong volume as early as possible into a move.
In addition, having a clear display also means (at least in my implementation of this indicator in Amibroker) that these can be programmatically detected, backtested, realtime alerts created out of them etc… In this case it becomes easier to, for example, send out an alert if a particular price level is reached with heavy or light volume.
I have also done several improvements on the original version I posted back in my original post. This is what it now looks like.
Here is a chart of the last two days action of CTRP. This is not necessarily the best example, but is one of those which I took action on and traded successfully.
First of all here is a list of the components of this indicator, since it actually changed a little since the previous version:
1. The red histogram indicates the average volume per 5 minute segment across the past 20 trading days. The red histogram is flanked by a dark red and pink histogram (dark red is not very apparent in this chart: it is the lowest one), which indicate the average volume plus/minus a standard deviation. The wider these areas, the more volume volatility there is.
2. This average volume +/- one standard deviation is also displayed as three, straight, horizontal parallel lines in light blue in the middle of the chart.
3. A cumulative relative volume is the light blue squiggly line (in this case rising). A steadily rising line shows that heavier volume was present throughout the day while a declining line (in which case it turns brown) shows a constant lack of volume throughout the day.
4. The actual volume is displayed in two places: First, superimposed over the average volume. This always shows green bars. Bright green bars show a higher than average volume (exceeding average + 1 standard deviation) and dark green shows lower volume (less than the average – 1 standard deviation). Second, as a relative difference to the average volume. If the volume occurred on a down bar (price wise) then the bar is displayed in red and if it occurs on an up bar, green.
It is this last bit which I find most useful as it really displays the interest (or lack of it) during that particular move.
Note how clearly the three high volume areas between 10:30 and 11:30 AM on the 20th July jump out.
Taking a look at yesterday’s action in GMCR, this time with a standard volume chart. Just like this, it is not quite clear whether the high volume at the beginning and end of the days are a result of the normal volume ‘smile’ or whether it is the result of unusually high volume. At this point you may say “it would be evident if you scope out a few days”, which is 100% true, however the point is to increase efficiency when reading charts. Something to grab your attention, especially when scrolling through several 100 charts.
Here is the same chart but this time with the relative volume indicator.
For anybody out there who is using Amibroker, here’s the code:
_SECTION_BEGIN(“Intraday Average Volume”);
if (Interval() >= 300) {
TimeFrame = Interval();
n = (23400)/timeframe;// number of bars in day
//TimeFrameSet(timeframe);
Avgvol=0;
StdDev=0;
d = 20;
barnum = BarsSince(Day()!=Ref(Day(),-1)); //bar number from start of the day
for (i=0; i<n; i++) {
VarSet(“vol”+i,IIf(i==barnum,V,0));
}
for (i=0; i<n; i++) {
Avgvol=IIf(barnum==i,Sum(VarGet(“vol”+i),n*d)/d,Avgvol);
VarSet(“AvgVol”+i,IIf(i==barnum,Avgvol,0));
}
for (i=0; i<n; i++) {
StdDev=IIf(barnum==i,sqrt(Sum((VarGet(“vol”+i)-VarGet(“AvgVol”+i))*(VarGet(“vol”+i)-VarGet(“AvgVol”+i)),n*d)/d),StdDev);
}
CurrentVol = V;
BarsSoFarToday = 1 + BarsSince( Day() != Ref(Day(), -1));
CumVolDiff = Sum(Currentvol,BarsSoFarToday)-Sum(AvgVol,barssofartoday);
Currentvol = CurrentVol;
//AvgVol = Ref(AvgVol,1);
//StdDev = Ref(StdDev,1);
//CumVolDiff = Ref(CumVolDiff,1);
//TimeFrameRestore();
//AvgVol = TimeFrameExpand(AvgVol,timeframe);
//VolStdDev = TimeFrameExpand(StdDev,timeframe);
//Currentvol = TimeFrameExpand(Currentvol,timeframe);
//CumVolDiff = TimeFrameExpand(CumVolDiff,timeframe);
Plot(CumVolDiff,”Cum. Diff”,IIf(CumVolDiff>0,colorLightBlue,colorBrown),styleThick|styleThick|styleOwnScale|styleNoTitle);
//Plot Volume Differences as a % of the current std deviation
Plot(0,”ZeroLine”,IIf(CumVolDiff>0,colorLightBlue,colorBrown),styleLeftAxisScale|styleNoTitle|styleNoLabel);
Plot((((StdDev+(V-AvgVol))/StdDev)*100)-100,”Diff”,IIf(C>O,colorDarkGreen,colorDarkRed),styleLeftAxisScale|styleHistogram|styleThick|styleNoTitle);
Plot(100,”Diff”,colorLightBlue,styleLeftAxisScale|styleNoTitle|styleNoLabel);
Plot(-100,”Diff”,colorLightBlue,styleLeftAxisScale|styleNoTitle|styleNoLabel);
Plot(400,”Diff”,colorLightBlue,styleLeftAxisScale|styleNoTitle|styleNoLabel);
Plot(-400,”Diff”,colorLightBlue,styleLeftAxisScale|styleNoTitle|styleNoLabel);
Plot(Currentvol,”Volume”,IIf(Currentvol > AvgVol+StdDev,colorBrightGreen,IIf(Currentvol < AvgVol-StdDev,colorDarkGreen,colorGreen)),styleHistogram|styleThick);
Plot(Avgvol-StdDev,”Lower”,colorDarkRed,styleArea|styleThick|styleNoTitle);
Plot(Avgvol,”Avg Vol”,colorRed,styleArea|styleThick);
Plot(Avgvol+StdDev,”Upper”,colorPink,styleArea|styleThick|styleNoTitle);
Plot(0,”ZeroLine”,colorBlack,styleNoTitle|styleNoLabel);
}
_SECTION_END();
Filed under: Amibroker, Trading Methodology, Uncategorized | 5 Comments
Recent Entries
- A stop Technique which i have recently implemented
- The Market universe and stock selection
- Starting Again
- Percent volatility model: A little better
- Part II: Profit Distributions
- A peek into a professional trader’s Equity Curve
- A review of a Stock Advisory Service
- Notes on Tape Reading
- A resurrection
- Making Sense of Volume Part II
- Managing Trading Time
Categories
- Amibroker (11)
- Indicators (2)
- Trading Business (5)
- Trading Methodology (8)
- Uncategorized (6)
