Well, you asked a question. It did not seem to me that you got your answer. And it also seemed to me that in dialogueing with you I might learn something myself. I have not asked you anything that I thought might be proprietary. But I was asking step-by-step questions to learn what your process is to get to your answer. Where I was going was the process by which you determine stop loss and profit targets to determine if there is some ambiguity there that might account for the fact that live trading is not reproducing testing. That is just one direction I would go. Another possibility is that you are suboptimized or over optimized. Yet another possible answer has been offered by another poster, that your backtesting was done over such a long time period that current volatilities do not match. But I cannot get there without knowing exactly how you test and trade. Which has nothing to do with the strategy itself. The short answer, however, is that I am quite stupid and cannot follow a complex discussion, so I have to do it in baby talk.
Any pure mechnical traders or system traders?
- Thread starter feng456
- Start date
Looking back on your responses ...
If each month results in approx 15 trades, generating a total of about 6 pts, then your average trade is approx 0.4 pts.
For the sake of argument assume that your target and stop offsets from entry are somewhere in the range of 0.5 pt to 2 pts.
Compare this with the ATR of a 30-min bar: I trade stocks and not futures; so others can confirm what I estimate (by analogy with SPY) that 30-min bars of ES probably have an ATR (averaged say over a week or a month) somewhere around 3 â 4 pts (and a 5-min bar has an ATR somewhere between 1 â 2 pts).
If the above is correct (and Iâm about to be badly flamed if itâs not
) then one problem may be that your 30-min and 5-min data are too coarse to accurately assess your strategyâs past performance.
Better ask Art D about his 1-second data ...
If each month results in approx 15 trades, generating a total of about 6 pts, then your average trade is approx 0.4 pts.
For the sake of argument assume that your target and stop offsets from entry are somewhere in the range of 0.5 pt to 2 pts.
Compare this with the ATR of a 30-min bar: I trade stocks and not futures; so others can confirm what I estimate (by analogy with SPY) that 30-min bars of ES probably have an ATR (averaged say over a week or a month) somewhere around 3 â 4 pts (and a 5-min bar has an ATR somewhere between 1 â 2 pts).
If the above is correct (and Iâm about to be badly flamed if itâs not
) then one problem may be that your 30-min and 5-min data are too coarse to accurately assess your strategyâs past performance.Better ask Art D about his 1-second data ...
I am 100% mechanical.
You have a nonstarter. What this means is that you cannot make any progress in making money. This is a negative comment.
your basic insurmountable difficulty is that your indicator and the market are incompatible.
What you use to get signals has no precise relationship to a 30 minute bar in specific terms of the periodicity of each. The ratio of signal device information to 30 minute bar information is a very high number. You must make this ratio about 1.0.
You test with one ratio and in live trading another ratio exists.
Some other major problems:
1. Most of your trades do not precipitate "exits". You simply get out at 4:00 pm because your margin forces you out. This means that your design is inter day and not intraday. this is big.
2. Your entry is "tied" improperly to the open for a major reason. The data shift from close of day to open of day.
3. You are carrying data over from one day to the next for a portion of each day. This fouls up your indicator relative to the market.
When I use the word indicator it is expressed with respect to how data is used to set actual values subsequently.
You are discussing, primarily, why something you came up with in non live trading is different when it is used in forward real time live with money actual trading.
4. There is a major disparity between entry and exit order type. As Deco pointed out to you, you are not doing entries and exits in time correspondence to bar data points. You did not test with a clock and you do not trade by a clock. Your excuse mentioning 5 minute bars is not acceptable in the sense of thinking logically.
5. At some point you have to look at the whole picture. the market offers and you do not do any kind of reasonable "taking" of the offer.
Murray is only able to speak from his experience and the experience of those he knows. the range is narrow and not effective nor efficient as you found out.
Deco is operating in the 700% a year (compounded, I would gess).
I read all the posts before posting. Reread hunnymunny about 25 times.
You have a nonstarter. What this means is that you cannot make any progress in making money. This is a negative comment.
your basic insurmountable difficulty is that your indicator and the market are incompatible.
What you use to get signals has no precise relationship to a 30 minute bar in specific terms of the periodicity of each. The ratio of signal device information to 30 minute bar information is a very high number. You must make this ratio about 1.0.
You test with one ratio and in live trading another ratio exists.
Some other major problems:
1. Most of your trades do not precipitate "exits". You simply get out at 4:00 pm because your margin forces you out. This means that your design is inter day and not intraday. this is big.
2. Your entry is "tied" improperly to the open for a major reason. The data shift from close of day to open of day.
3. You are carrying data over from one day to the next for a portion of each day. This fouls up your indicator relative to the market.
When I use the word indicator it is expressed with respect to how data is used to set actual values subsequently.
You are discussing, primarily, why something you came up with in non live trading is different when it is used in forward real time live with money actual trading.
4. There is a major disparity between entry and exit order type. As Deco pointed out to you, you are not doing entries and exits in time correspondence to bar data points. You did not test with a clock and you do not trade by a clock. Your excuse mentioning 5 minute bars is not acceptable in the sense of thinking logically.
5. At some point you have to look at the whole picture. the market offers and you do not do any kind of reasonable "taking" of the offer.
Murray is only able to speak from his experience and the experience of those he knows. the range is narrow and not effective nor efficient as you found out.
Deco is operating in the 700% a year (compounded, I would gess).
I read all the posts before posting. Reread hunnymunny about 25 times.
A backtest of 100 to 200 trades is not a very large sample. If you optimized to this sample then it is not surprising that out of sample results are lower. Trading on the ES is competing with the best. I trade stocks and ETFs but have the best edge in stocks that are too small for many pro's.
You received several pieces of excellent advice.
If what people are explaining doesn't make sense, that's another issue altogether.
To build a 'system', it is not enough to crunch numbers and look for a backtest with positive expectancy. You also must have at least a basic understanding of trading and markets, which it seems is missing. Spend some time on understanding markets and trading, and I assure you that all the advice you received will make a lot more sense then it does now.
If what people are explaining doesn't make sense, that's another issue altogether.
To build a 'system', it is not enough to crunch numbers and look for a backtest with positive expectancy. You also must have at least a basic understanding of trading and markets, which it seems is missing. Spend some time on understanding markets and trading, and I assure you that all the advice you received will make a lot more sense then it does now.
feng,
Here is a post that does several things. Think of it as an open kimono type post. It deals with one of your recent unanswered questions to Hey:
hey what is your average monthly profit? im trying to see if im aiming too low or not
The standard of performance is taking the market's offer.
The market's offer, as yet, is unseen by you. We can compare the offer to your performance and find out why your work is a non starter so far.
Consider the following.
To see and understand the market's operation you have to have criteria and measures. Here are six that are worth using.
1. Sentiment (The 10 cases of adjacent bars are used to measure sentiment) (Long is HH and HL; short is LLand LH) (the other 8 cases (all are valuable as they form , as well) are used for other purposes in a strategy.)
2. Market PACE (deciles of volume PACE are deployed) (This is corrolated with two price variables of adjacent bars: overlap and volatility. Normal distributions result) (Volume leads price.)
3. Dominance or Non dominance (trends are either with or against the observable market)
4. Nested fractals (Three sontiguous fractals are a minimum: the fastest observable fractal, the trading fractal and the boundary of the trading fractal)
5. Continuation or Change (the market is broken into profit segments) (continuation is when money is made.) (Change is when sentiment reverses; it is the timing aspect of making money)
6. Finite math (Only a go/no go approach is possible in markets because of market granularity.) (Binary vectors turn out to be the go/no go that incorporates size and direction.) (Binary vectors measure time rate of change (the first derivative in continuous functions))
To determine peformance you compare the market's offered profit segments to the "taking" you achieve taking those segments.
A simple measure is the vertical length of the segments compared to the ATR. Use a zig-zag chart and Excel sheet to make this calculation. Use a monthly average of the ATR.
Now, we can look at your performance in detail. Use your secret monthly results for a contract in ES. Divide it by 22. Then divide it by the ATR. Now you have something against which to compare to the market's offer stated as a multiple of the ATR. Divide the monthly multiple of the ATR into your daily fraction of the ATR. You will find that the first significant number is far removed from the decimal point in this decimal ratio.
Here is a third comparison. Trading the slowest of the three fractals above yields 3/4 to 2 times the ATR. Trading the "trading fractal" yields 2 to 4 times the the ATR. Trading the fastest observable fractal yields up to 6 times the ATR.
Here are some other comparisons. You trade using 13 bars a day. The "trading fractal" has 81 bars a day.
On the fastest fractal you can see (observe) 25 to 40 trends a day.(See zig zag observations) These are simply price moves that last 2 or 3 bars (5 min bars) whereby the extremes of the trends occur at the extreme ends of the bars and never on the open or close of a bar (with minor exceptions).
It appears that you have used 30 min bar values. You have but 13 a day to work with. They are restricted to up to four values and some of these values are machinated to get specific values of price only. You use these for three purposes: enter, protection and targets. All entries are used; protection gives you losses always; targets are not hit too often and the 4:00 something value is used to take you out of the market. Mostly all CW oriented traders get to the point you are in this paragraph.
So you need to ask questions of like minded people. I'm not one of those, obviously; I trade to take all of the market's offer on a continuing basis. This is a HOLD/REVERSAL strategy that uses six of the market's most prominant raw data measures.
Your total sample represents one week of price moves for most mechanical traders and traders who are automated.
I feel that the progression of a learning trader (who can keep adding to his ATS) is to go through 10 levels of increased sophistication. About 12 days is a good enough time to digest the opportunity each specific level offers.
You are at the level of observer which is before the first beginning level. Read the posts from like minded people; they are contributing to your future welbeing and you need to get their input.
Here is a post that does several things. Think of it as an open kimono type post. It deals with one of your recent unanswered questions to Hey:
hey what is your average monthly profit? im trying to see if im aiming too low or not
The standard of performance is taking the market's offer.
The market's offer, as yet, is unseen by you. We can compare the offer to your performance and find out why your work is a non starter so far.
Consider the following.
To see and understand the market's operation you have to have criteria and measures. Here are six that are worth using.
1. Sentiment (The 10 cases of adjacent bars are used to measure sentiment) (Long is HH and HL; short is LLand LH) (the other 8 cases (all are valuable as they form , as well) are used for other purposes in a strategy.)
2. Market PACE (deciles of volume PACE are deployed) (This is corrolated with two price variables of adjacent bars: overlap and volatility. Normal distributions result) (Volume leads price.)
3. Dominance or Non dominance (trends are either with or against the observable market)
4. Nested fractals (Three sontiguous fractals are a minimum: the fastest observable fractal, the trading fractal and the boundary of the trading fractal)
5. Continuation or Change (the market is broken into profit segments) (continuation is when money is made.) (Change is when sentiment reverses; it is the timing aspect of making money)
6. Finite math (Only a go/no go approach is possible in markets because of market granularity.) (Binary vectors turn out to be the go/no go that incorporates size and direction.) (Binary vectors measure time rate of change (the first derivative in continuous functions))
To determine peformance you compare the market's offered profit segments to the "taking" you achieve taking those segments.
A simple measure is the vertical length of the segments compared to the ATR. Use a zig-zag chart and Excel sheet to make this calculation. Use a monthly average of the ATR.
Now, we can look at your performance in detail. Use your secret monthly results for a contract in ES. Divide it by 22. Then divide it by the ATR. Now you have something against which to compare to the market's offer stated as a multiple of the ATR. Divide the monthly multiple of the ATR into your daily fraction of the ATR. You will find that the first significant number is far removed from the decimal point in this decimal ratio.
Here is a third comparison. Trading the slowest of the three fractals above yields 3/4 to 2 times the ATR. Trading the "trading fractal" yields 2 to 4 times the the ATR. Trading the fastest observable fractal yields up to 6 times the ATR.
Here are some other comparisons. You trade using 13 bars a day. The "trading fractal" has 81 bars a day.
On the fastest fractal you can see (observe) 25 to 40 trends a day.(See zig zag observations) These are simply price moves that last 2 or 3 bars (5 min bars) whereby the extremes of the trends occur at the extreme ends of the bars and never on the open or close of a bar (with minor exceptions).
It appears that you have used 30 min bar values. You have but 13 a day to work with. They are restricted to up to four values and some of these values are machinated to get specific values of price only. You use these for three purposes: enter, protection and targets. All entries are used; protection gives you losses always; targets are not hit too often and the 4:00 something value is used to take you out of the market. Mostly all CW oriented traders get to the point you are in this paragraph.
So you need to ask questions of like minded people. I'm not one of those, obviously; I trade to take all of the market's offer on a continuing basis. This is a HOLD/REVERSAL strategy that uses six of the market's most prominant raw data measures.
Your total sample represents one week of price moves for most mechanical traders and traders who are automated.
I feel that the progression of a learning trader (who can keep adding to his ATS) is to go through 10 levels of increased sophistication. About 12 days is a good enough time to digest the opportunity each specific level offers.
You are at the level of observer which is before the first beginning level. Read the posts from like minded people; they are contributing to your future welbeing and you need to get their input.