Morning success, Afternoon problems.

[MBender]

First, I apologize for my short absence. I had to do my day job to keep cash flow while I work all this stuff out.

Did VERY well in the morning, then lost it all in the afternoon. I don't know what it is about these e-mini's -- but the mornings and afternoons are just different. If I make money in the morning, I lose it in the afternoon, and vise-versa.

Then again, maybe it's all the instruments. But, e-mini's are my current selection, so I'm sticking with them, at least for now. If anyone knows of any other instruments that are more stable, I'm all ears.

My problem was the stops. I didn't need them in the morning, so I got sloppy with them in the afternoon.

Amature's mistake.

Merv, What's $TICK? I tried a search on it and couldn't find it. It sounds like you've had some success with it, so I'm interested.

Brad,, Thomas and Merv, I'll be mulling over your interchange this evening -- thanks for the input.

Cheers, all -- and Welcome all Lurkers.

Michael

[Doji3333]

My general rule with stops is they should be out of the way (initially) and then come into play if the position is moving against you or when you are starting to make profits.

My adaptive strategy allows the market price to determine a stop level initially, then there are three trailing mechanisms for trailing the stops based on different conditions.

I was very interested to see that the ES gave no triggers to trade yesterday, but today it caught a short almost all day. But the NQ traded pretty well yesterday and not so well today. I believe we're in one of those "transitional periods" where some days are great and others are tough.

My main robot is doing well this week (still). Two trades in the ES and two trades in the NQ - All winners.

With the ES this week:
Mon : 2 trades, both winners +4 pts
Tues : one trade, one winner +2 pts
Wed : One trade, one winner +2 pts
Thurs: two trades, both winners +4 pts

So far not too bad - eh? Let's hope tomorrow is like today.

[arbifox]

Sounds all to familiar

Brad - the results sound great, but how does the robot perform week over week? For instance, in July I lost maybe 13 points on the Russell, while I made 54 points in August. So I am always interested to hear how it does over the long haul.

Regarding Michael's point about mornings and afternoons and how they are different, this is such a pain in my ass in terms of programming. I have had days, and days, where I allow for 2 bad trades and I get them back to back in the morning, and then the program goes into paper mode tojust track additional trades, and all those wind up huge winners. I wish I would tell you how to get around this, and all I have done in my work is to check the bad trades that were taken, see what criteria might have prevented the trade, and fire up the backtester and see if a change would have provided better results. In most cases any new condition proves worst results.

In my travels, I have just chalked it up to "Win some, lose some".

BTW, $TICK is a measure of the difference between issues advancing through an uptick, versus those dropping in a downtick, and there are flavors for each market. In my experience, it is purely a scalping tool and I have found it has little use for medium to long term trading since there are so many whipsaws.

More ideas to follow . . . I have REALLY enjoyed this conversation.

Just an idea . . . any interest in forming a Yahoo group for folks like us? I already have a few Fund Managers following my work and I am more than happy to make connections for anyone that has a proven track record. You need about 1 year of history that can be proven through bank records.

Thomas

[merkro]

Ok guys, put me straight on terminology here please. Thomas, what do you consider a scalp as opposed to medium to long term trade? I consider a scalp being from 3 to 6 ticks of the Russell, taking from seconds to maybe a few minutes depending on conditions. I would call a day trade anything that takes longer then a few minutes up to several hours, and medium to long term I would count in days. Are we talking about the same thing?

If you had a good month with 54 points, that is some 2.75 points per day for 20 trading days, which may qualify for a couple of scalps per day. So I can talk the same language in future posts, please put me straight.

I'm not sure I agree with ceasing after 2 losses. I think auto trading is simply a matter of averages. If over a period of weeks you are reasonably sure of getting sufficient wins to outweigh the losses, then it doesn't really matter whether the losses are all in a row, or spaced out. (drawdown restraints excepted). A bad day loss of $500 is more then acceptable if the following 2 days return $500 each. If a certain level of losses are expected, then they are simply a part of the cost of doing business.

Though the market definately changes depending on the time of day, and a certain amount of control over trading can be maintained by using time periods to regulate trading sessions, the real answer to this problem has to be volatility. If we can successfully measure the volatility in time to catch the movements (rather then after the event), then we have it made. I firmly believe the most important ingredient of what we do is forcasting trend direction, or lack of same. I have not discovered the secret, however believe the final answer will involve the principals of the Tick.

I would not qualify for your Yahoo group, however would like to see the 3 odd threads we have going here combined. Perhaps I'll post a couple of charts with results one day.

Cheers

Merv

[arbifox]

I would agree with your terminology on scalping. Most of my trades last several minutes. I would also say the a "scalp" trade in my books usually has a profit target defined. Tomorrow, I will attach my trades for 2010 to date, and you can see how most of my trades unfold.

August was just one of those months where there were some massive swings in the market, and the program managed to capture quite a few good ones.

In terms of shutting the system of after a couple of bad trades, a few thoughts on that. First up, I have custom designed back tester, and through testing the system on data back to February 2009, no other setting proved as pofitable as shutting the system off if you have several bad trades back to back.

This brings into play volatility, and I agree 100% that it is vital to understand. Looking at the days where I have two trades that **** me down for the day, more often than not the rest of the day was equally bad. I examine such days and see if I find any patterns or anything that might allow me to filter for volatility. To date, I have not found anything that can help.

At least for my purposes, I have found that during times of high volatility the agreement of my key indicators across several time frames does not happen, and as such, trades are not taken. So all I have to figure out are the times that trades are taken and the conditions are volatile.

Strangely enough, I have a volatility monitor built into my system and I have not sed it as much. Perhaps it is time to revisit that code. In this monitor I use the magnitude of a 3M candlestick to trigger the volatility condition, so long as I am not in a trade. I usually wait for the market to reverse or another event to occur to get me out of that condition.

I welcome the use of the $TICK, but ultimately I think that can only help in true scalping of a market, or high frequency trading as the cycles of the TICK are very cyclical and have a very short amplitudes.

Thomas

[merkro]

I'm enjoying your posts, Thomas. Two comments. First the Tick. I agree to actually trade the tick, you are scalping fast time. However the tick has more use then simple scalping. As I see it, the Tick is the difference between the number of stocks that went up on thier last trade, and the number that went down. So, if the tick is positive, then there must be more stocks rising then falling, and the trend should be up. As the tick is the sum difference of the market moves, it is instantaneous with the moves, making it a proactive indicator. The eminis on the other hand, mirror the equiveleny market, trailing to a certain degree, meaning thier movements are slightly slower, and following the tick.

Now, if we forget the fast fluctuations of the tick, and take an average of it in relation to zero, then perhaps we have a slight indication of where the market is heading. For instance, an average of 400 over the last 30 bars should mean bullish conditions. If we further refine this and compare the 30 period average on bar(0) to say bar(-15), then we should be able to tell if the tick is holding or losing momentum, again giving us an inkling of what may happen to the market. Now if you think this theory crazy, please tell me, and save me some heartache in the future.

I now understand your reasons for ceasing after a couple of losses. Perhaps I should do more research into similar instead of striving for a long term average. I would be interested in any ideas you have re a method to guage tradability at any given time. If you revisit that code, let us know what you find.

Cheers

Merv

[merkro]

Of course, Brad's 'Adaptive Strategy' is exactly what we are talking about here. He is determining market tradeability as he alters his indicator parameters to suit the current conditions. Perhaps he'll crack it for us all?

Cheers

[MBender]

Thanks and follow-up

I really appreciate all your (collective) comments.

Stops: I was good at these not long ago. Before I enter the trade, I try to pre-determine the point where I'm absolutely sure the trade has gone against me. For example, if I'm going long at the bottom of a channel, I put the stop a few ticks below the channel's floor. I just got sloppy having had a string to successes.

For algorithm trading, I put the hard stop a few ticks below the computer's calculated stop (in case of loss if internet, etc.)

When the trade is sufficiently in-the-money, I move the hard stop to a few ticks in the money, then leave it there as the algorithm adjusts the exit point.

Trade filtering using time of day.

I've thought about this a lot, but I've decided against it. I've only been doing this about 5 weeks full time (well, almost full), my experience is that the NQ trades poorly in the AM and well in the PM, but yesterday it was reversed. I suspect as the markets change, so will the timing, so -- back to adaptive strategies.

Adaptive Strategies:

I have two I'm playing with. First, I'm working on "mud" indicators. These try to determine when the market is muddy (just not going anywhere). My algorithms won't live-trade when the market is "in the mud". They work well sometimes, I lose profitable trades others, but they do improve the overall success.

Second: Fuzz (term comes from "fuzzy algorithms"). Fuzz adjusts my signaling indicators based on volatility (I primarily use ATR). So, let's say the S/W determines that the market is channeling. I want to detect when the a breakout occurs. I can't just use the trades that go above the channel ceiling as this could simply be the channel expanding. So, I add a fudge factor (Fuzz) as a buffer. When the trade goes above the ceiling + fuzz, then I assume the market's broken out of the channel.

This is where I got into trouble yesterday. The market broke through the ceiling + fuzz, when up a few ticks, then fell back down. In hindsight, it was beginning to form a cup w/handle, and it never broke below where my s/w set the channel's floor. Had I put my stop where I should have, I could have ridden it out.

Too many losses:

I understand both perspectives. I've noticed my gains and losses run in spurts. My hypothesis is that my strategies do well during certain market conditions, and poorly in others. Continuing to trade in poor market conditions is unwise. For me, right now, I' going to with with a three strikes and out approach.

Thank you all for the comments.

Cheers,
Michael.

[MBender]

Opps. forgot.

Multiple time frames:

I've found some days trade well at 1-minutes, some at 2-minutes, and some at 3-minutes (and some don't trade well at all). I haven't tried 150T or 450T yet -- It's on my to-do list for next week.

I like to confirm with 5x chart; so, if I'm trading a 1-min chart, I'll confirm with a 5 min. If I'm trading a 2-min chart, I'll confirm with a 10-min, etc.

The markets are fractile, but fractile is not continuously divisible. 5x just seems to work for me.

Cheers,
Michael.

[Doji3333]

Well, I worked out all my stop issues today. The ES chart took it's first full loss on two contracts this AM (with a short), But climbed back with a mid-day long and is currently long going into the close of the day. We'll see how this works out. I have the loss limiter set to 2 right now for testing. I would probably leave it at one for live mode.

The funny thing is the NQ did well today on two winning trades. Not huge gains, but still helped to offset the losses on the ES.

I'm thinking I'll need to run this code in pairs on ETFs or on the e-Minis. I've been looking at the FAS/FAZ and TNA/TZA ETF symbols as they are inverse pairs. That may play out really well as one goes up, the other goes down.

I'll post more charts in the Adaptive thread after the close.

Glad many of you are enjoying these threads.

[merkro]

A snapshot of some 18 hours of my scalping system, bearing out previous comments about morning profit/afternoon loss. Forgetting the first 3 trades, there were 6 winning trades between 3am and 11:52 am, followed by 4 consecutive losses from lunch time until approaching close. Another winner right on close, for a net gain of 6 points from 14 trades. I consider this a good result, providing there is consistancy over a period. Perhaps trading only between 3 am and midday would further improve the result. I have found the 3am to 7am period normally very rewarding.

Cheers

Merv

[arbifox]

I am starting to follow the posts on Active Traders a little closer, and I am going to start to discuss details on my system there if anyone wants to know. I think MBender set this up . . .

I will still follow these threads as well . . . and post when I can.

T

[merkro]

I have not been able to find the Active Traders thread. Can anyone kindly supply a link?

Cheers

Merv

[arbifox]

Active Traders Link

http://thebean326.blogspot.com/

Try that out . . . and let me know if you can get in.

T

[MBender]

Glad to hear from you, Merv

Merv:

It's an invitation-only blog. We're still using this forum, but on the blog, we can be a bit more open.

Send me an e-mail at theBean <at> allybusiness.com and I'll forward you an invitation.