Back testing is tough, back testing is boring. Back testing is #$%^&. No doubt. But while I was looking back charts, programming codes, ideas and thought flowing into my mind.
I was having some thoughts over the process of developing my trading system. At the earlier days when I first started Dummy trading, I simply jumped into a trade as long as a valid dummy spot appeared. I can’t say that was right or wrong. But by trading this way, it reduced the expectancy and accuracy of my system (well, I thought I did have a system! lol). When you have a system with positive expectancy, chances are you will be profitable in the long term. But when you have a system with low accuracy, you will have a tough time fighting with yourself to handle the strings after strings of losses (even though they were just small losses!). I had a discussion with Dave on this topic recently as well.
I decide to have a look at the core element of dummy trading. TREND. Yup, basically, I am looking to buy pull back or consolidation within an established trend. During the choppy day, I will be bleeding. I will be better off if I can analyze the current trend before I establish a position. Again, I can spend the whole day drawing lines, calculating moving averages, but day trading requires fast and accurate decisions, so , it will be helpful to have something built to reduce number of decisions I need to make during trading session.
I have drafted out a high-level chart to illustrated my work now.
My task now is developing a Trend Evaluations system. Of course, by no mean, this is a holy grail. It might be just a Moving Average crossover system, drawing some lines or just an ADX calculation system etc. No rocket science. It just to assist me to understand and see the TREND in a faster, and handy way. I am trying to find a method that will keep me at the right side of the market, (well, of course, it can’t be 100% accurate, at least, it put me at the high probability side of trade) regardless of what other system rules (such as Dummy) I might be following. So the end result will be, I have a “system” that takes a position as soon as I find a trend and then exits when the trend is over.
The reason is, if I start from a basis where I am already on the right side of the market, then whatever entry and exit rules I superimpose on my calculation of trend SHOULD only continue to improve my results.
Well, work is still in progress. Just to share something that I am currently doing, and I hope it helps me to grow as a better trader.
I’ve started watching ADXs lately for the same purpose. The jury’s still out on whether it’s helping, in my case.
I’ve started watching ADXs lately for the same purpose. The jury’s still out on whether it’s helping, in my case.
So far, in my paper trading and back testing, ADX does help. No conclusion yet. Just to share with you. to assist in viewing ADX, I modified the indicator by adding a trigger line (basically, it is just a slower ADX).
So far, in my paper trading and back testing, ADX does help. No conclusion yet. Just to share with you. to assist in viewing ADX, I modified the indicator by adding a trigger line (basically, it is just a slower ADX).
I’ve tried using ADX but I found it to be a lagging indicator. By the time the ADX was rising and over 30, I already missed most of the move.. I think the best filters you can put on your dummy trades are to take them in the direction of the stock AND overall market trend.. If you have both of those things going for you, that’s your edge. You also want to make sure the trend is young and not overextended.
Hey Kevin,
Thanks for the input. I agree with you ADX is lagging. In fact, almost all indicators are lagging. I am not using ADX alone for my system. It just one of the components.
I would prefer to look at the trend of ADX instead of some specific numbers.
Gav
Hey Kevin,
Thanks for the input. I agree with you ADX is lagging. In fact, almost all indicators are lagging. I am not using ADX alone for my system. It just one of the components.
I would prefer to look at the trend of ADX instead of some specific numbers.
Gav
Hey Gav, what’s your take on Thailand’s currency restrictions and its impact on Asian markets? What are people saying in Singapore? Thanks
Hey Gav, what’s your take on Thailand’s currency restrictions and its impact on Asian markets? What are people saying in Singapore? Thanks
Well, it begs the question.
What is the definition of a “Trend”? Is it…
1) The inclination for future prices of some time-frame to move in the same direction as a specific set of past prices in that time-frame? [General]
2) (Uptrend) When prices of successive rallies close at levels higher than those of previous rallies and lows occur at levels higher than at previous lows? [Dow Theory]
3) (Uptrend) When today’s close is above yesterday’s close? [Simplistic]
One really important question: How small can the smallest trend be? (ie. is the 12/19 5-min bar chart from 9:40 to 11:00 a trend?) It covers only 1 hr 20 mins, captures about 7 points. Is it long enough to be considered a trend on a 5-min chart?
Well, it begs the question.
What is the definition of a “Trend”? Is it…
1) The inclination for future prices of some time-frame to move in the same direction as a specific set of past prices in that time-frame? [General]
2) (Uptrend) When prices of successive rallies close at levels higher than those of previous rallies and lows occur at levels higher than at previous lows? [Dow Theory]
3) (Uptrend) When today’s close is above yesterday’s close? [Simplistic]
One really important question: How small can the smallest trend be? (ie. is the 12/19 5-min bar chart from 9:40 to 11:00 a trend?) It covers only 1 hr 20 mins, captures about 7 points. Is it long enough to be considered a trend on a 5-min chart?
Ok, so I’m the master of the rhetorical question, which can get pretty aggrevating!
There are no hard and fast answers that fit all. Lets take the Dummy system. What is the “Edge” you are trying to capitalize on? It might be defined as “My edge is that I have a set of indicators which gives me an above average chance of capturing a strong move within a trend with a small risk of loss.”
Great, so the nature of the edge is capturing “a strong move” with respect to a “small risk of loss” but you require “a trend”. So, you better danged well have a trend or else you’ve automatically given your edge away! I’d say in this case, you want to be late in concluding you have a trend, it has to be really obvious that a trend exists.
That means you want a very high value for the minimum trend length. Say you’re using 60-minute bars to measure trend. Looking at 10 bars and saying “I see a new trend forming” is going to kill you in such a system.
The downsides of having a high value for the minimum trend length are (1) you’re late to the party…the trend is already well established before you start hopping on board and (2) you’re still going to be trying to buy pullbacks well after the trend is over.
The upsides is that while it’s trending, it should be like printing money. You’re riding on the coattails of a trend, the wind is to your back and you should have a really impressive positive R-distribution.
Now, what if…there were a way to define trend without any lag? What if you could “predict” a change of trend? (Uh oh…sorry more rhetorical questions! Just can’t help myself).
Ok, so I’m the master of the rhetorical question, which can get pretty aggrevating!
There are no hard and fast answers that fit all. Lets take the Dummy system. What is the “Edge” you are trying to capitalize on? It might be defined as “My edge is that I have a set of indicators which gives me an above average chance of capturing a strong move within a trend with a small risk of loss.”
Great, so the nature of the edge is capturing “a strong move” with respect to a “small risk of loss” but you require “a trend”. So, you better danged well have a trend or else you’ve automatically given your edge away! I’d say in this case, you want to be late in concluding you have a trend, it has to be really obvious that a trend exists.
That means you want a very high value for the minimum trend length. Say you’re using 60-minute bars to measure trend. Looking at 10 bars and saying “I see a new trend forming” is going to kill you in such a system.
The downsides of having a high value for the minimum trend length are (1) you’re late to the party…the trend is already well established before you start hopping on board and (2) you’re still going to be trying to buy pullbacks well after the trend is over.
The upsides is that while it’s trending, it should be like printing money. You’re riding on the coattails of a trend, the wind is to your back and you should have a really impressive positive R-distribution.
Now, what if…there were a way to define trend without any lag? What if you could “predict” a change of trend? (Uh oh…sorry more rhetorical questions! Just can’t help myself).
wow..it’s Dave again. lol
I’ll shoot you an email and organize the reply as a new post later.
wow..it’s Dave again. lol
I’ll shoot you an email and organize the reply as a new post later.
OK, one last post here….
One of the things about tracking trends is that conventional methods tend to be slow to recognize that the trend has changed.
I’m not offering a holy grail of trend changes here, but another way of looking at changes in market direction is time cycles.
For instance, we have had had mostly bullish days on the ES for about a month now. But using cycles, I see us coming in for a good day for shorting tomorrow (12/20). This is based on a couple of cycles.
Does it have to happen? No, the market CAN AND WILL DO ANYTHING! But my edge tomorrow is that I feel I have a much better than 50% chance of making money by being short. It might be another up day and I’ll be wrong….so it’s not like I would put an extra large position or anything. It is just that I have to go where I think my edge is and I’m betting on the short side tomorrow.
This is an example of a change in price direction (maybe too much to call it trend here…I don’t know) without using typical trend-following methods like moving averages.
-MagicT
OK, one last post here….
One of the things about tracking trends is that conventional methods tend to be slow to recognize that the trend has changed.
I’m not offering a holy grail of trend changes here, but another way of looking at changes in market direction is time cycles.
For instance, we have had had mostly bullish days on the ES for about a month now. But using cycles, I see us coming in for a good day for shorting tomorrow (12/20). This is based on a couple of cycles.
Does it have to happen? No, the market CAN AND WILL DO ANYTHING! But my edge tomorrow is that I feel I have a much better than 50% chance of making money by being short. It might be another up day and I’ll be wrong….so it’s not like I would put an extra large position or anything. It is just that I have to go where I think my edge is and I’m betting on the short side tomorrow.
This is an example of a change in price direction (maybe too much to call it trend here…I don’t know) without using typical trend-following methods like moving averages.
-MagicT
I've tried using ADX but I found it to be a lagging indicator. By the time the ADX was rising and over 30, I already missed most of the move.. I think the best filters you can put on your dummy trades are to take them in the direction of the stock AND overall market trend.. If you have both of those things going for you, that's your edge. You also want to make sure the trend is young and not overextended.