Cost of Production

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
SL
Senior Member
Senior Member
Posts: 30
Joined: Tue Feb 10, 2004 7:02 pm
Location: New Zealand

Cost of Production

Post by SL » Sun Aug 22, 2004 10:38 pm

The best LTTF systems it is often said are ones that trade the same rules for long and short markets and the same rules for each market and keep the rules for entry and exits to a minimum say 5 at most.

My testing agrees with this logic. The overall conclusion is, simple is good.

However there is one issue I have with this simple logic.

When talking about commodities particularily grains, metals etc. there is a base cost of production or bottom price at which you would no longer want to sell your goods for. Each market has a minimum cost of production below that figure you pack up and go find something else to do with your money. I am not going to get into the fundamentals of what happens about supply and demand. My interest is LTTF system rules.

Many LTTF systems tend to trade on a breakout or use a moving average of some sort. The upside on long trades can be in theory, infinite. The short side of the market is a different story. On the short side if the system gets a new low it fires and places a short trade. It doesn't concern itself about the 'price' as such of the commodity just the relative price to some figure generated by a previous low or a moving average. So the systems fires all the way down looking for the elusive trend if it already isn't in a trade.

The problem I see is that below a certain price point the risk vs the reward on the short side is very limited in each commodity. That's becuse the price will only go to the cost of production. The downside potential of taking a short trade is limited depending on where the system takes a trade. The risk vs the reward on the long side is a different story.

But my systems are dumb, they do as they are told and hunt for new lows or whatever until the market turns and long trades are presented. I know myself that a quick scan of the screen over 20 years would not have me trying to enter a trade at certain prices on the short side for a LTTF system. I know the trade won't last long enough to provide a suffcient reward to overcome losses on average before the market is forced to reverse as it gets close to the COP which about 80% of the time will be short lived by a suddenly rally busting out your stop.

My system has to generate x number of points profit to overcome average losses. LTTF systems need a good dose of profit to overocme the losses so it requires a good size trend to do that. Theres a good chance that once the price gets below a certain point near the cost of production it has a much greater chance of going north or simply whip sawing along the bottom looking for an upside breakout.

Now, if I were to take all these special rules for each market into consideration I would be breaking the system 'design code' i.e don't have to many rules. This could extend the system rules to easily above a dozen or more. Actually I am not even sure I could put these per market rules in my platform not easily anyway.

To look at this objectively I would assume I take my own advice and work out from a certain price point what the reward would be on the downside and decide not to take a short signal i.e overide them. The other alternative is to have a short term system of counter trend system kick in below a certain price point again requiring rules on a market by market basis.

Do we just live with this shortcoming? Have I got this wrong? What do our panel of experts say?

Stephen

damian
Roundtable Knight
Roundtable Knight
Posts: 815
Joined: Tue Apr 15, 2003 8:43 pm
Location: dusseldorf

Re: Cost of Production

Post by damian » Mon Aug 23, 2004 10:16 am

SL wrote:The problem I see is that below a certain price point the risk vs the reward on the short side is very limited in each commodity. That's becuse the price will only go to the cost of production. The downside potential of taking a short trade is limited depending on where the system takes a trade. The risk vs the reward on the long side is a different story.
You might also need to consider the influence of using certain styles of back adjusted data. I agree that a minimum price may well need to be respected as "not worth being short below this level". However, if you define that price as being historically static within your rule set, ie for Crude Oil lets say $8 per barrel, then you will encounter the problem that your historical back adjusted data are correct as a ratio to current levels but are potentially very incorrect as a ratio to the prices that actually traded on any given day in the past. This is a problem when you say, for example, in your rule set: "ignore short signals if Close<8 dollars". A problem especially considering that a great deal of Crude Oil back adjusted prices in fact have Close<0 dollars.

This is not a fatal issue, it can perhaps be overcome. Different historical contract chaining techniques might overcome this.... and subsequently introduce unique issues of their own (usually depending on the type of indicators/mathematical price modifications your system employs). The joys of back testing.

On a slightly related issue, quite some time ago I raised the point that with ED so close to 100.00, there is perhaps little point taking a long term trend following buy side trade with entry at 98.50 and position sizing initial exit at 97.80, unless you believe in negative future interest rates in the exchange traded market for discounted simple interest products.*

damian

(* ps: I have actually dealt many times at negative yen interest rates, but that is usually an "implied" rate. A few times I have seen a negative bid being given in the standard cash market as well!! Being paid to borrow money was a strange concept at first but another story).

SL
Senior Member
Senior Member
Posts: 30
Joined: Tue Feb 10, 2004 7:02 pm
Location: New Zealand

Cost of Production

Post by SL » Mon Aug 23, 2004 9:36 pm

Damian,

All valid points.

The backadjusting levels already occured to me. I use CSI backadjusted data with forward adjustments, raised etc.etc.

I have no doubt that trying to make a rule for each market would be an exercise and a half and beyond what my platform might accept.

Either way it may pay as Hiramhon suggested to trade the two sides as different systems. This may not be the turtle way, but I wouldn't discount it on that basis.

Sometimes it is very hard to seperate sound advice from logical thinking when the majority have done it one way for so long everyone comes to believe that there is only one right way. Logic says the short side has some very distinctive traits that should be taken into consideration when designing systems. After a while you become limited only by your own beliefs which have been superimposed by others.

The answer is to approach the issue scientifically as if you didn't have a clue either way as to what may occur and examine the outcome based on the facts only.

Cheers

Stephen

TC
Roundtable Knight
Roundtable Knight
Posts: 101
Joined: Thu Jul 08, 2004 2:12 pm
Location: Toronto, Canada

Post by TC » Tue Aug 24, 2004 1:08 am

For both Long & Short positions my LTTF parameter values are the same in all the markets I trade with the exception of individual stocks and stock index futures. There is such a strong bias to the Long side that I simply do not Sell Short equities and their derivatives with my LTTF system.

I revisit this issue periodically but fail to resolve this dilemma satisfactorily. The fear of curve-fitting is partly responsible for my reluctance to run different parameter values for Long & Short trades and also the fact that it is considerably more difficult to make as much money shorting stocks cf only trading Long.

I'd be interested to know if other stock traders use different systems to trade Long & Short or different parameter values in the same system or have found some other way to sell short with the same level of profitability as they trade long

Nathan
Roundtable Knight
Roundtable Knight
Posts: 127
Joined: Sun Jul 13, 2003 3:52 am

this

Post by Nathan » Tue Aug 24, 2004 2:50 pm

reminded me of bruce kovner's interview in mkt wizard. He talks about how he had to adjust to the characteristics of bear markets and shorting.
not doing this resulted in his one losing year.

Perhaps some wisdom from his discretionary method/style could be of use.

richard
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Tue May 18, 2004 10:17 pm

Post by richard » Tue Aug 24, 2004 5:37 pm

TC wrote:There is such a strong bias to the Long side that I simply do not Sell Short equities and their derivatives with my LTTF system.
This may be true over the long run from 1982 - the present, but have you tested for periods prior to 1982? That is the start of the Great Bull Market.

If you tested between a period like 1966 - 1982, you would probably test well doing shorts as well as longs, or better with shorts even. But I don't know, I'm just basing that on what the indexes did.

budonk
Roundtable Fellow
Roundtable Fellow
Posts: 62
Joined: Tue Dec 30, 2003 7:44 pm
Location: Lake Tahoe, NV

Post by budonk » Wed Aug 25, 2004 10:32 am

hi gary, great results. depending on how small the stop is, the Veritrader testing engine might not treat those stops correctly(No inside bar lookback). Not to say those results are not correct, just want to warn you.

TK
Roundtable Knight
Roundtable Knight
Posts: 167
Joined: Tue Apr 15, 2003 5:45 pm

How does VeriTrader handle stops?

Post by TK » Wed Aug 25, 2004 10:41 am

budonk wrote:Veritrader testing engine might not treat those stops correctly(No inside bar lookback).
If on a given bar both the profit target and stop loss stops may have theoretically been hit, how does VeriTrader handle this situation? The worst-case (and safest) scenario would be to assume that on such days the stop loss stop is always hit first. Is it the case with VeriTrader?

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt » Wed Aug 25, 2004 11:44 am

I've warned about this many times elsewhere on the site.

You can't do reliable testing with only daily bars if you are using small stops and intraday entries.

See:

viewtopic.php?t=101
viewtopic.php?t=35

The following are in the VeriTrader support section, so you won't be able to read them if you are not a customer, sorry:

viewtopic.php?t=445
viewtopic.php?t=906

It doesn't matter whether you make optimistic or pessimistic assumptions, there is a very large potential for error in either case. My recomendation is simply not to rely on any tests with intraday stops of less than 1 ATR and preferably 1.2 to 1.5 ATR. Even with stops this large, there will be the occasional day that rips through the entry, breaks through the stop and then makes new highs that a test will miss.

Pessimistic assumptions will pretty much rule out any small stop with a breakout-type system as most breakouts would result in your stop being hit.

You really need intraday data to solve this problem.

- Forum Mgmnt

TK
Roundtable Knight
Roundtable Knight
Posts: 167
Joined: Tue Apr 15, 2003 5:45 pm

Post by TK » Wed Aug 25, 2004 5:16 pm

Forum Mgmnt wrote:It doesn't matter whether you make optimistic or pessimistic assumptions, there is a very large potential for error in either case.
If I insisted on being able to test "pessimistic assumptions" just to be on the safe side, will it be possible in VeriTrader 2.0?

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt » Wed Aug 25, 2004 5:40 pm

Yes, you will be able to do that in 2.0 if you want.

In 2.0, you will also be able to test using the associated intraday data and come up with accurate tests even for fairly small stops.

- Forum Mgmnt

Post Reply