Futures Journal

Hi Kosmo,

Actual trading results match testing results almost identically, with the notable exception that there has been more slippage than anticipated with the MD and RL contracts. I am investigating ways to resolve this, and may go to the e-minis in an attempt to do so. I'd hate to drop MD and RL and go with only SP and ND, but may be forced to do so eventually anyway as size overtakes volume.

I have been trading the longterm system for three months and added the short term system about three weeks ago.

Doctor Nevo
 
Hi Daniel M,

I've been trading the longterm system for 3 months. I tested it in conjunction with the shortterm system over 6 3/4 years because the SP began trading on Globex in 19960101, which is when it became appropriate to begin testing the short term system.

In developing the short term system I tested it in isolation from the long term system using 9 different US indices over this 6 3/4 year period -- this gave a sample size of over 4000 trades. 5 of these indices had insufficient volume to trade real time, but provided extra sample size for testing purposes.

In developing the long term system I also tested it in isolation from the short term system, using a separate 44 market portfolio and testing over a 20 year period. I don't recall exactly how many trades resulted from this in-sample period, but I was satisfied that it was sufficient to establish for my purposes that it was robust and not overly curve-fit.

I enter all trades in front month contracts and roll, generally when the next month contract OI exceeds the front month, but I also listen to my broker's advice here. Things can get a little dicey with some contracts, like the Bund, where there is literally no volume in the next month contract right up until the delivery date.

Doctor Nevo
 
Doc, you tested on some 40 markets, yet the number of markets you'll be trading is substantially less than that. Did you decide on the markets to trade solely on the basis of their past performance?

Also, what's the most number of markets your system had you in at the one time? Is there a limit to how many positions your equity level will allow you to be in at any one time?

Are you varying the size of your positions any during the course of each trade (based on volatility for example, or other conditions), or are you maintaing a constant size throughout?

Thanks. (Just ignore if I'm being too intrusive)
 
Hi Daniel M,

I selected my 19 market portfolio to be diversified and contain liquid markets that generally trend well. Markets that didn't meet these criteria were not included in the 19 market portfolio. In achieving a "diversified" portfolio, I chose to avoid having more than 3 markets in any sector (although I currently have 4 in the currencies -- I'm thinking of getting rid of the Dollar Index). Liquid markets needed to meet my somewhat subject standard -- at least several thousand daily volume. As for trendiness, I just looked at backtesting results. Obviously, markets like gold, meats and cocoa didn't make the cut.

In general it has me in about 2/3 of the markets at any time. I currently have open positions in 13 markets

I do make infrequent adjustments over time. Although I don't want to elaborate too much, I do not pyramid, scale out in drawdowns, or make adjustments based on volatility.

I would like to investigate the impact of "rebalancing", whereby periodically appreciated positions are partially liquidated and other positions are increased to bring general risk level to an even level among positions. Unfortunately I haven't found any software that permits such testing, although the new version of Trading Recipes is rumored to be able to do so.

Doctor Nevo
 
@Doctor Nevo

In your testing of the commodity markets did you take into account the effects of limit days? Can this be done in Trading Recipes?

Many Thanks

Bernd Kuerbs
 
Hi BKuerbs,

Obviously the data already takes into account limit days with respect to historical testing.

Trading Recipes does take into account limit days with its current trading mode insofar as it filters out stop orders that would be more than a limit move away from the previous day's close. That is, when running TR as a tool to determine what working orders you should have in place the following day, TR will not tell you to have a working stop order that could not be reached because it is more than a limit move away.

Doctor Nevo
 
Originally posted by Doctor Nevo
..Obviously the data already takes into account limit days with respect to historical testing.

...
More...

Sorry, I do not understand that: the information, that a limit move has taken place, is not contained directly in the data. You have to know the size of a limit move, which may change historically. Otherwise I know only the simple and maybe sometimes wrong method of checking whether High = Low for the day.

Originally posted by Doctor Nevo
...
Trading Recipes does take into account limit days with its current trading mode insofar as it filters out stop orders that would be more than a limit move away from the previous day's close. That is, when running TR as a tool to determine what working orders you should have in place the following day, TR will not tell you to have a working stop order that could not be reached because it is more than a limit move away.

Doc..
More...

Imagine the following scenario: you are short in a commodity. Next day a limit move up takes place. This may happen for several days in a row. Most software will tell you that you exited your short position with a small loss and made a killing entering a long position while in fact you had a substantial loss being cought in that short position. I ignore the possibilities of buying puts, offsetting in backmonths (which again may not work for crop-related commodities): these techniques may at most reduce your loss.

How ist that/can that be handled in TR?

Many Thanks

Bernd Kuerbs
 
OK, I understand your point.

TR does not have a way to account for getting caught in an adverse limit market move. It would be extraordinarily difficult for any software package to account for this since limit rules can be quite complex and change over time. I suppose you could code it into the data, but CSI and normal ASCII data don't contain such information -- it's only O,H,L,C,V,OI.
 
I'll make a quick journal entry for last night.

Last night, I received my daily statement -- new equity high made yesterday.

Downloaded daily data and ran it on my software. Long-term system gave a sell order for Yen. I looked at my betsizing algorithm and it called for 5 lots.

I logged onto my broker's software and placed an on-line order to sell 5 lots Dec Yen MOO.

Then I ran my short term system on the indices. Observed what the reversal orders should be and sent an e-mail off to my broker effecting working stop orders for the following day.

This morning, woke up and saw first my broker's acknowledgement of the index orders, then saw my fill notice for the Yen order.

None of the index reversal orders were triggered today, so we continued to be short all 4 indices. This, combined with adverse moves in the debt instruments and London metals, has led to what will probably be about a 5% drawdown today (quite a substantial one day drawdown).

Doctor Nevo
 
You must log in or register to reply here.
Back
Top