consitancy

[renkoben]

I'm developing a strategy with about 7 entries.

I did a backtest on Only one entry

Briefly speaking ES 5min chart entry 7 only take profit at 12 ticks

I did a back test after an optimization for the 12 months of 2012. the BT got the same results as the optimiz. (that's good, it's supposed to be the same)

I got 195 trades for the 12 months 70% profitable sharp ratio 1.14 profit factor of 1.74 Total net profit of $ 8,350

THEN I took this optimization settings and placed them in for the 12 months of 2011 the previous year. I was hoping to get similar results. I was hoping for consistency.

NO GOOD

I got 363 trades for 2011 using the same settings as 2012.

57% profitable sharp ratio -.25 profit factor of .93 Cum profit of -3.45
and total net profit of $-2,025 I LOST MONEY

If there is no consistancy from one time period to another than this is a waste of time

what do you advise?

What do you think?

Ben

[sledge]

If there is no consistancy from one time period to another than this is a waste of time

what do you advise?

What do you think?

Ben

I think you are right that there is no consistency from one time period to another.

The Markets change and are generally never the "same". The VIX was "wild" in 2011 (compared to now), over 20-40 at times.

Now the vix has been 14-18 for sometime, smaller daily ranges, only a couple intraday reversals if that.

Back when VIX was pushing 80 (2008/) there were many intraday crazy reversals. Your strategy would probably have lost bigger then. Especially taking only 3 points profit. (Not sure what your stop loss is).

[renkoben]

If there is no consistancy from one year to the next (12 months of 5 min data) and that's a lot of data to even (try to) out the extremes, then this says that technical analysis is bunk.

After I get all 7 entries working I'll probably have 6 trades per day and 6% profit per month (with no leverage) and about 64% profitable trades but that's the optimization. Reality is another thing

[renkoben]

with the above results, should I be making money with such an optimization? What do I have to do? have 5 entries with 85% profitable trades and 6 trades per day. I can't do it.

[sledge]

If there is no consistancy from one year to the next (12 months of 5 min data) and that's a lot of data to even (try to) out the extremes, then this says that technical analysis is bunk.

After I get all 7 entries working I'll probably have 6 trades per day and 6% profit per month (with no leverage) and about 64% profitable trades but that's the optimization. Reality is another thing

You might want to check out this thread for some comparison...

http://www.ninjatrader.com/support/forum/showthread.php?t=54193&highlight=let%27s+compare+strategies

[koganam]

If there is no consistancy from one year to the next (12 months of 5 min data) and that's a lot of data to even (try to) out the extremes, then this says that technical analysis is bunk.

After I get all 7 entries working I'll probably have 6 trades per day and 6% profit per month (with no leverage) and about 64% profitable trades but that's the optimization. Reality is another thing

Unfortunately, that is the nature of markets. We cannot impose our ideas of consistency on the market. Attempting to do that is just another way to blow up an account or two.

It is we who have to adapt to the changing market, not the market to us. That means that you have to select your optimization period to be such that it identifies that nature of the market, and you then have to realize when the market's nature changes. If that means having a set of optimization parameters for different market types, then that is what you will have to do. Something of a bummer, right?

As an example of what I mean, look at my TimeSliceRange indicator, which looks at the average (in this case, median) range of each 30 minute time slice over a statistically significant lookback number of days.

Given this value, we can test it against today's values. If you look, you will see that it allows us to calculate the likely limits for any 30 minute period, and by looking at the cumulative value for the day, we can see what the day is looking like.

Today for example, had a normal kind of range. I have seen days when the Cumulative figure was 65%, meaning that I would then be expecting median ranges for each next period to be only 65% of normal. I may choose to have those smaller targets, or they may be small enough that no trades meet my risk parameters, and I just SOH. Typically, if the Cumulative is <75%, I know I am in a scalping day, and trade accordingly, if I choose not to SOH.

For all practical purposes, this is a quasi-continuous optimization based on the current market ranges, with a continuous fixed lookback period, and a lookforward period of one.

The jagged orange line is the 100% mark: the bars are the actual values for the day, so I can even see if what started as a narrow-range day has suddenly expanded, as for example happened at around midday on 12/12/2012.

[sledge]

with the above results, should I be making money with such an optimization? What do I have to do? have 5 entries with 85% profitable trades and 6 trades per day. I can't do it.

If this was the case, then everyone who signs up, runs the optimizer, will be rich and find the magical MakeMegaBucks with ease.

[renkoben]

what is SOH? Maybe I could use ATR average true range logically or Boll Band width using a 30 min time frame or something. Also DM with the red and blue lines with ADX, this could identify bull/bear trends or no trends.

when you say you use 30 min to measure the character of the market what is the corrisponding time frame that you actually trade? It seems that you are pointing to multi time frame analysis. Using the longer time frame to identify the lower (trading) time frame characteristics.

[koganam]

what is SOH? Maybe I could use ATR average true range logically or Boll Band width using a 30 min time frame or something. Also DM with the red and blue lines with ADX, this could identify bull/bear trends or no trends.

SOH = SitOnHands.

I cannot comment on your use of indicators, as I do not use those indicators.

when you say you use 30 min to measure the character of the market what is the corrisponding time frame that you actually trade? It seems that you are pointing to multi time frame analysis. Using the longer time frame to identify the lower (trading) time frame characteristics.

Not quite. The 30 min TimeSliceRange is to tell me where price is likely to reach in that 30 minute slot. That will be the sole filter to determine if I will take the next trade. I use an 8-minute chart to determine direction very simply, and find entries on either the same 8-minute chart, or an 8-range chart. So let us say that I have an entry determined at 1010 EST. The price projected by the TimeSliceRange will be my target. As I know the entry price, I can use that target to determine if I will take the trade. Once it has been determined that the trade meets the initial criteria, other factors will kick in, such as, for example, am I going long at a natural resistance? In which case, I may pass. etc., IOW, TimeSliceRange is a targeting system.

[renkoben]

If there is no correlation of future results with past performance ie automated trading is random then why bother with optimizations and BT'ing
I suppose the answer is that there is some sliver of having a bit of that performance repeat and that is where you will make your sliver of profit if your lucky enough to find it.?

[koganam]

If there is no correlation of future results with past performance ie automated trading is random then why bother with optimizations and BT'ing
I suppose the answer is that there is some sliver of having a bit of that performance repeat and that is where you will make your sliver of profit if your lucky enough to find it.?

A market is not a determinate entity: it is probabilistic; the result of the actions of its participants. If there were an exact analytic formula to predetermine outcomes, then it would cease to be a market, as all one would need to do would be to plug into the formula, and all participants would and could only make profits. Such a system is often called the "Holy Grail". The search for it is often identified as the second stage of development of the trader.

While the Holy Grail does exist, it is not some magical combination of indicators, which guarantee profits. Then again, that discussion should probably be on a forum more dedicated to trading, rather than support of a trading platform.

There are quite a few threads over on BMT: not quite so many on here. Here is one pretty long thread, at BMT, that contains a post that I made about 3 years ago on that topic: http://www.bigmiketrading.com/psycho...html#post21212