How Useful is Paper Trading

Recently I did a paper trading using TurtleFarm on Capitaland, and a friend of mine had questioned the usefulness of such practice. I share here some of my thoughts.

I understand that since no money in involved, there is less emotion burden on me. But I tried to model as closely as possible to real trading, including going through all the preparation, ready for buy/sell etc - just short of keying in the real order and bet with real money. I hope that I can get out the most from this exercise by doing so.

This paper trading exercise did a few things, for me at least: it tested not only the software, but also the turtle rules and system, and finally as a trader myself.

First, let looks at the easy part: the software. TurlteFarm is reasonably stable, though I did experience a few minor bugs and conflicts with my other programs (eg: my screen capturing software created a drop down box at the upper left hand corner), but those I can live with it and don't bother to spend time solving it. TurtleFarm did a good job in generating the signals in advance thus preparing trader for the action the next day. Since it rely on EOD data, so intraday trading using this software (at least the current version) is out of question. Currently it don't have a screening capability, so I have to rely on my workhorse Metastock to do the short listing first, and use TurtleFarm to do all the position sizing & entry/exit signals.

Next the turtle trading system. This is more difficult part. Since this system requires intraday entry/exit as per signal generated, I find that at times when was not in front of the trading screen, price moved and trigger the buy/sell, I may not be above to enter/exit at the exact price (our broker don't have stop buy/sell order). Slippage is an issue need to be addressed by the trader. Once the price has moved away from the planned price point, later entry will incurred additional risk and affect the profitability, while no entry you risk not riding on a long trend without full 4 positions and affect the profitability again.

The turtle entries are in batches, meaning that for every counter, we have a potential of max 4 buy price everyday (that is for my BUY ONLY setting). The entry price and stop price are also updated everyday. Imagine what it looks like when we are tracking 6 potential counters at the same time. SGX also has this lot size requirement that does not fit nicely with the turtle position sizing, where we as trader had to do some adaptation (eg: round down).

Finally, the most difficult part: we the trader who is using this system to trade. I don't know what other will do, so what stated below is based on my personal experience. There are times when the entry signal is generated, I have a different opinion about the market (eg: window dressing). There are times when my discretionary chart reading skill is telling me it is bad sign already, but turtle trailing exit still quite far away. Seeing the possibility of a profitable trade (base on current closing price) that may be exit at a loss is painful. Trying to control myself not to interfere with the trading signals generated need some effort. I think one way that can lessen the pain and make me adhere to the trading signal is NOT to trade all my trading capital using this system. Using only part of my trading capital on this method of trading, will leave me part of my trading capital on other trading method.

I believe the paper trading exercise is useful, at least the stimulation has uncover some of the rough spots without costing me a cents. I would rather notice these now and hopefully think of some kind of preparation/work around then discovering them when my money is in the line and pressure is building up.

Hc,

In order not to clutter this thread, I wrote about the combobox at
the Turtle Trading System (link) thread.

I also very briefly touched on the bug caused by the unusual
min tick > hi-lo range bug you reported.


You have correctly pointed out the problem of the SGX-ST system
no able to accept stop orders. This is a design limitation on SGX
part. There is no such problem with US stock exchanges and US
futures (even the electronic).

Singapore like to say we have world class this and world class that.
For example, we say we have world class transport system but
if you observe at the Raffles Place MRT station, when the E-W train
arrives, the EMPTY N-S train is usually waiting with open doors.
Passengers will run out of the E-W train the moment the doors
open but before they could reach centre platform, the N-S train
would close its doors and say bye-bye.

ALL my feedbacks to the world-class MRT staff have fallen on deaf
ears. The N-S station there is only one station from the start point
of the line. The train is usually EMPTY. Why can't they just co-ordinate
in a world class way.

Let us hope the new SGX trading platform can accomodate at
least the standard universal orders - limit, stop, stop limit, market,
market if touch.

There are things I like to touch on trading the Turtle system but
maybe at a later time.

Please share your thoughts. Thanks.

I found this on the Internet. I think it is relevant for a person
learning the Turtle Trading System. It is supposedly a newsletter
that Russell Sands emailed out to his students, people who
attended his seminar.

I know there have been a fair bit of criticism about Russell Sands.
Some people like to practise the aggressive "kill-your-competitor"
type of tactics in the market-place.

Much like what we saw between MediaCorps and MediaWorks a
few years ago here in Singapore. Up till today, I am still wondering
what is the reason behind TCS' change in name to MediaCorps the
same year that MediaWorks was borned.

And do you still remember the Ren Ci Hospital Charity show incident
before the merger, when it was hosted by MediaWorks?

I can only say this: Right now, the most complete original
rules are from Russell Sands and nowhere else, no matter what
claims there may be in the world. He painstakeningly took notes
during the training.

If you have even that bit of anti-Russell feeling then you will read
this with a bias. But if you really understand the Turtle Trading
System, if you have gone through it over and over again, then you
will understand what he was saying. He may sound a bit harsh but
remember this is a dialogue between a teacher and his students.

Taken from:
http://www.trade2win.com/boards/archive/in...php/t-5634.html

QUOTE

Hey guys, if December was a nightmare for the Turtle equity curve,
then January must be the night of the living dead. A drop of over
100% of equity in less than two months on the computer model. This
is bad stuff. Not unheard of though, it has happened eight times in the
past fifteen years. well, thank god it is just a computer model. But
some of you seem to have lost the thought of that concept.

Do any of you remember that we discussed this scenario in the
seminar, under a concept called "risk of ruin"? Specifically on page 93
of the Turtle Manual. There was some stuff there that had to do with
the idea of cutting your position size if/when you start losing money.
Remember ? These were not secret trading rules guys, this is bloody
common sense !!

The computer models don't know Jack. They trade a million bucks.
It takes all the trades in all the markets, and sizes positions as though
its account was always a million bucks. The computer model doesn't
pyramid on winners, and it doesn't cut back on losers, and it most
certainly doesn't know how much money any one of us have in our
accounts in real life.

The nightly orders are just a guide, not the god damn holy grail. The
whole point of the Turtle system experiment was to teach people how
to think about and trade the markets. People with brains and common
sense. Any monkey can read and blindly follow a computer printout.
The Turtle is supposedly higher up the evolutionary scale than the
Monkey.

Cut back !! Cut back !! Fewer trades. Less markets. Smaller positions.
Was this whole discussion lost on you guys ? Even the brokers, whose
not their job it is, tell me they have been suggesting to people who are
losing to cut back their trading. But I hear that many of you are not
listening. Why ????

I don't watch your accounts day to day, I have my own problems to
worry about. And the brokers, well most of them are also sort of busy
at times like this, and some of them even want you to keep trading
more and keep those commissions rolling in.

I taught you guys how to trade, but I am not the babysitter. You all
need to learn how to think for yourself. The one weakness of the
Turtle system is that it sometimes goes through long and tough losing
periods. You have to put up the gloves and play defense here. If you
start losing money, then you start reducing your trading, it's as simple
as that.

My account has dropped from $130,000 to $106,000 so far this year.
I am not real ****ing happy about that. But I do occasionally look
at other account statements on the same brokerage run as me. And
when I see a guy who has dropped from $150,000 down to $ 50,000,
and is still trading multiple contracts per positions, well now that just
isn't right.

If you're trading a larger (say six figures) account, you need to reduce
the number of contracts in each of your trades, and maybe even reduce
the number of markets you're trading. And not just once, but again and
again, if your equity keeps slipping lower. Even if you're trading a small
account and can't reduce this way, you can still reduce the number of
trades, i.e. switch to turtle system 2 instead of turtle system 1.

The markets have been bad, but professionals like us should know
better than to make it worse than it already is. I am not going to tell
each of you what to do, and your broker is certainly not going to tell
you what to do. This is one that you have to step up and show some
responsibility and make your own decisions. Play defense and reduce !!

UNQUOTE

There are gold nuggets in there. You will recognise them if you when
you have "realization" of the Turtle methods. You will then know its
strength and weakness.

For example, do you detect that there is still discretion involved? Why?

If so, how about other trading system. It is so common to hear that
such and such a trading system is 100% mechanical. What do they
mean? Must you tak ALL signals? Can you skip signals? If yes, why?
If no, why?

The clue is diversification and risk exposure. Not a simple answer if
you don't understand risk and money management.

Having a trading system and a trading plan is very different from
the plan technical analysis type of thing.

Paper trading can teach you some of the things. It can teach you the
behaviour and characteristics of the system. It can show you in real
time how trades can turn out. The thing paper trading cannot teach
you is your emotion. Emotion go with real money involvement.

Paper trading can provide you the expectation before you put in real
money. But when you put in real money, your emotion can take over
and give you a totally different experience from the prior paper
trading. In paper trading it is easier to follows rules. When you have
real money in, you have become a different person. Not the same
one that has done the paper trade.

The clue now is therefore how you do the paper trading. If you want
to be the same person then you have to approach paper trading as
the money-in-the-market personality.

Realtime paper trading forces you to have a day where the market is
opened and then is closed. The mind can behave uncontrolled during
the day.

Use paper trading to also find out yourself, to experience yourself, to
find out what type of person you are. It is not just to find out the
trading system you are testing.

Remember discretion is still required in ALL trading system, unless
you can print money and therefore have unlimited supply and do not
have to worry about risk and money management. No one in the
world can do this. Even the world richest man will go broke without
this discipline because he still has a limited supply of money.

1. Emotion

Agreed that emotion is an important and significant part of trading, which paper trading cannot stimulate.


2. Discretion in System Trading

This is a tricky part, and I have some thought on it.

I understand that Turtle Rules do has a provision of reducing the notional trading account size by 20% when the trading account was reduced by 10%. And repeat that if trading account size is down for another 10%. This in a way is trying to trade smaller size when faced with a string of losses.

Question here is: is this sufficient to diverse risk?

Let just jump out from Turtle trading and look at it from the outside. Turtle trading is essentially a trend following trading system. That means it works when there is (strong) trend, and fail if there isn't. If the market undergoes prolong sideway movement, the drawdown of turtle trading can be severe. But if we take and skip the signals as we deem fit, we risk missing a trend. To me taking the signals as it come is non-negotiable.

So it become quite obvious now that if we have something that compliment this turtle trading, preferably some kind of trading strategies that perform better during non-trending situation, the risk will be capped. If we go along this line of thought, than dividing the total trading capital into various trading systems will become the natural conclusion.

As to the question of how to divide the trading capital, I am not thinking an uniform equal amount for each trading system. Just like Richard Dennis will allocate trading capital to each turtle differently based on their performance, I am more incline to allocate more to those that is more profitable within a predetermined period and review regularly. The exact method of division of capital may be subjective and discretionary, but I think consistency is the key.

Critics and comments are welcome.

This is a piece of news I read a few days ago: "Computer scientists working on machines that can match Wall Street traders"

http://www.iht.com/articles/2007/05/03/business/bxdata.php

I quote here:

"AI is very effective when there's a specific solution," Hamilton said. "The real challenge is where judgment is required, and that's where AI has largely failed."

If system trading is mechanically implemented (however comprehensive the trading system is), it is just like AI. And if you agree to the above comment, it will fail when judgement is required.

On other other hand, is human judgement more reliable? I don't know about other traders' judgement. What I know is that I have mood swing, I have issues in life other than trading to worry about. All these affect my judgement and my trading.

So how?

姜太公钓鱼, 愿者上钩.

As the story goes,
in ancient China,
there's an old man who spent all his time fishing.
he would never catch anything because he did not have a hook

Everybody laughed at him.
They did not understand, he was not fishing for fish but for a great man.

Try looking things at another angle.

---

28 Feb 2007,
did the charts look bad ? Yes
did things sound bad ? Of course.

However I had explained, if someone want to give u alot of money.
You just have to take it and buy.

But it is very difficult to change one's perception.

April 2000,
Nikkei did not "crash"
but for that particular week,
many made a fortune shorting Nikkei.

Why ?


This is trading.
No play money.
As much as your broker allow.

Why ?

What I am trying to say,
don't let the little things in trading distract you.

Paper trade, because you are not believer.

Discretion

Here it does not really mean picking and choosing the signals to
take and ignoring the others. The Turtles have strict risk control.
For any one instrument, the maximun position is 4 units. For a
closely correlated group of instruments, the total position size in
all of them is 6 units.

Let us take the currency group and the 4 popular ones - EUR, JPY,
GBP, CHF. It is very common and in fact usual that when there is
a trend, all 4 would likely move together. Individually, total position
in them would be 16 units (4 x 4 units). This is against the Turtle
rules. We have to forego 10 units since the maximun allowed is
6 units. The discretion part is which one to take and how many.
There is always the question of are we taking the correct ones
since the extend of individual price moves in term of P/L will not
be the same.

Then there is the maximum portfolio limit of 12 units long and 12
units short. Once this limit is hit, bascially no more signals can be
taken.

This is what discretion means in a trading system. Obviously there
will be other form of rules in other trading system.

In light of this, the idea of an automated mechanical system would
infer all signals are taken regardless of what is happening in other
instruments. Take the 4 currencies above in the currency group.
An automated mechanical system running on all 4 currencies will
initiate individual maximum postions in all 4. This will create an
unusually high risk exposure no matter how strong the trend may
be. A sudden unexpected development will affect all 4 currencies.
This is bad management. This is why a fully mechanical system
do not work out.

There are many people attracted to exactly this idea of mechanical
trading that they get hooked into programmnig their trading systems
into TradeStation, run it and have it generate orders unattended in
the expectation that it will print money for them. TradeStation can
handle orders generated and enter them automatically to the
brokers electronically an report back to the program.

They are still trying to find their dreams.

Hi,

Correlation is just a number. It can be programmed.

One of the problem that stumps many clever men, is that in times of trouble,
everything becomes highly correlated and liquidity dries up.

Worst, many clever men assume trouble will only strike in a million year. Trouble is they happen quite frequently.

Unfortunately, troubles do not happen "too frequently". Therefore a money manager with a few years of luck will get rich and the poor clients suffer.

Hi Undead,

"In light of this, the idea of an automated mechanical system would
infer all signals are taken regardless of what is happening in other
instruments."

This statement was in the discussion of discretion in trading system
where the impression of fully mechnical trading systems is to
take all signals which run counter to risk and money management.


"Correlation is just a number. It can be programmed."

Correlation does not even need to be programmed. It can just
simply be assigned to the Symbol.

If Symbol.Group = Currency then
If Currency.Position < Maximum then
Symbol Blah Blah Blah
End If
End If

Repeat this 4 times for the 4 symbols,

What happen when Currency.Position = 5, meaning there is
room for only 1 more unit, and there are signals in EUR and GBP
to add more units? The rules don't permit both to be taken but
just one. So which one and how will the decision come about?
By using discretion. Whether the discretion is programmed in
codes or made manually, a signal has to be dropped.

I would like to see something like that in Ver 2.
Eg a user-defined file

i.e
Symbol.Group = Currency
Currency.Position < 6
Symbol EUR
Symbol GBP

Symbol.Group = StkIndex
Currency.Position < 5
Symbol Dax
Symbol Stoxx

...

maybe

Year 2000
Symbol.Group = Currency
Currency.Position < 6
Symbol EUR
Symbol GBP

Symbol.Group = StkIndex
Currency.Position < 5
Symbol Dax
Symbol Stoxx

Year 2001
Symbol.Group = Currency
Currency.Position < 3
Symbol EUR
Symbol GBP

Symbol.Group = StkIndex
Currency.Position < 10
Symbol Dax
Symbol Stoxx

or whatever.
As it will be tedious to key in such information,
best for Ver 2 to read an external file where the format is well defined.

(While we are on Ver 2, perhaps you like to look at the printing codes. I do print out charts and paste to the wall.
I would say the printouts from V1 are poor.)

Ahh we are getting something new here,
we can solve this problem in many ways.
But let me dig further,
how does a Turtle make a decision ?

I'm always using a automated system.
More for market making.
But someone will always be there to monitor the positions.
Therefore I think even for a black box automated system,
a clerk is needed to check ie that the connection to the exchange remain sound.

Ver 2 can be integrated to Excel.
Many OMS has API to automate via Excel.

"I would like to see something like that in Ver 2.
Eg a user-defined file"

It is implemented in similar way when the portfolio is constructed.
Users will have to define the individual component Symbols'
properties.

Group Maximum Position has to be fixed at 6 to follow the rules.
It should not be variable from year to year.


"As it will be tedious to key in such information"

It will need to done once when constructing the component
symbols for the portfolio. Once for every portfolio.


"the printouts from V1 are poor"

It is being improved to produce printout like TradeSation 2000i.
The codes will print directly to the printer. Background will
default to white while OHLC and fonts will default to black. It
is more tricky to do than TradeStation because while TradeStation
has a limited 16 colors to chose from, Analyst's TurtleFarm do
not have such limit.


"But let me dig further,
how does a Turtle make a decision ?"

They made the decision by discretion. That was why all the
Turtles did not have the same exact positions. They could still
take both positions by taking 1/2 unit for EUR and 1/2 unit for
GBP for a total of 1 unit therefore still not exceeding 6 units.
Any further signals in that group, whether for a new position
in another symbol or to pyramid existing symbols will have to
be skipped.

Needless to say, therefore all the Turtles did not have identical
results. In actual fact, the Turtles were not doing well in the
first half of 1984. Richard Dennis and William Eckhardt then
taught them System 2.

So if you hear from one original Turtle bragging that he was
making money while other Turtles were not during that period,
saying he was following rules and insinuating the others were
not, it is better not to believe him. Take that as self-promoting
for self interest.


"Ver 2 can be integrated to Excel.
Many OMS has API to automate via Excel."

There is no plan to do this. The reason is it is not practical.
Every single symbol in the portfolio will always have orders
for the next day.

For a symbol that do not have a position, there will be 8 entry
orders: - 4 Buy and 4 Sell. Then there will be 8 stop losses
to correspond with the 8 entry orders, the appropriate stop
loss to be entered for the corresponding entry, cancelling
the ones entered earlier for the earlier pyramid entries.

There will also be orders for symbols in position if the max
of 4 is not reached. Also corresponding stop loss for pyramid
entries.

Automating this will mean entering all orders. Assuming there
is no positions and there are 30 symbols in the portfolio, there
will be 240 entry orders and 240 corresponding stop loss order
that will need to be changed when entries are made.

This brings us back to the discussion of discretion and why
automated mechanical system when taking all signals go
against risk and money management rules.

Hi csk,

The turtles are not having identical results. This is because some of the decisions are made by discretion.

Therefore understand the principles of the turtles, not just the rules.

Position should not be fixed. Traders should do their own research.

It may or may not be variable from year to year.

For example, consider Nikkei 225 and JPY. Sometime JPY goes up, Nikkei goes up. Sometime JPY goes down, Nikkei goes up.
In this case, will the results be affected by a constant 6 ?

I'm not sure if I follow the argument.

Let's consider a human trader. He has to note down 480 signals somewhere.

He will then have to monitor the markets constantly and trade accordingly.

He will note the appropriate stops and cancel signals that are now invalid.

Then of course he will ask,
what are the things best left to a computer ?

If Ver 2 move into portfolio management,
then it must have a screen that keep track of such data.

If there's such a screen, integrating into Excel is not a big problem.

If a firm decides to go automated, then the firm has to do add their own logic on the generic data provided by V2.

I am not sure what Undead meant by "integrating into Excel" , but I think a way to export the orders generated to Excel would be helpful.

Please consider. Thanks.

Hi Undead,


"Position should not be fixed. Traders should do their own research.

It may or may not be variable from year to year.

For example, consider Nikkei 225 and JPY. Sometime JPY goes up,
Nikkei goes up. Sometime JPY goes down, Nikkei goes up.
In this case, will the results be affected by a constant 6 ?"


Maybe, some part of the Turtle rules is not clear. I try to explain.

Nikkei 225 should be in the Indices Group while JPY in the Currency
Group. For exmaple:

Indices Group
Nikkei 225
S&P 500
Dow Jones Indus
NASDAQ
FTSE
HSI
MSCI Taiwan
KOSPI
etc.

Currency Group
EUR
JPY
GBP
CHF
AUD
CAD
etc.

Grains Group
S
SM
BO
C
W
O

Interest Rate Group
T Bonds - 20yr
T Notes - 10yr
T Notes - 5yr
T Notes - 2yr
Eurodolloar
Long Gilt
Short Stirling
Bunds
JGB
EuroYen
etc.

and so on for other Groups


Each Group has a Maximum Position of 6 units. The Maximum
Porfortlio Position is 12 units long and 12 units short.



"If Ver 2 move into portfolio management,
then it must have a screen that keep track of such data.

If there's such a screen, integrating into Excel is not a big
problem.

If a firm decides to go automated, then the firm has to do add
their own logic on the generic data provided by V2."

I had understood your earlier suggestion to mean controlling the
orders and execution in realtime since OMS automation was mentioned.
OMS, as I understand the abbreviation, means Order Management
System.

As you can see, handling the OMS automation requires keeping track
of positions - per Symbol, per Group and per Portfolio. It is not
practical for the program to do this.

However, if it means to export the orders to an Excel spreadsheet
then this is easily implemented. This is similar to exporting
the Performance Report to Excel workbook in the current version.
The same way that Analyst's DataServer exports EOD Snapshot to
Excel worksheet.

All orders generated for the next day go to a Tracking Sheet. This
Tracking Sheet can be exported to Excel. This is in Analyst's
TurtleFarm ver 2.

As you can see, the program goes through the Symbols in the Portfolio
to generate the orders. Similarly, outside of any Portfolio, the
program can go through the Symbols in a data directory to generate
orders, aka, scanning. This will go into a Tracking Sheet too.


"If a firm decides to go automated, then the firm has to do add their
own logic on the generic data provided by V2. "

I now understand that OMS automation is outside the control of the
program.


Hi Hc,

"but I think a way to export the orders generated to Excel would be
helpful."

This is in Analyst's TurtleFarm ver 2.

Hi csk,

I think if we follow the rules strictly,
then your solution is good.

But suppose I had some research and perhaps I believe,
last year and this year we should concentrate more on Fx for example,
next year less on FX,
then i need a little more flexibility on portfolio management.

I need to know whether turtle trading will add value to my research etc.

Sometime I wonder,
if we do away with 20 period channel and replace with 20 period moving average of high and low, what will be the outcome.

And so on

I need to clarify. Expoerting the orders from the New Orders Alert
Window will be in an upgrade. Need not be ver 2.

This is in response to a post at the The Speculator, Hedger and Arbitrator (link).

What does this means Analyst's TurtleFarm New Orders Alert Windows:

"No New Entry Orders - Account hit ruin on yyyymmdd"
where yyyymmdd is a date.

The Turtle money management rules stipulate ruin as when the account
equity hit 50% of initial account equity. For eample, if the initial $ equity
of the account or fund at inception is say $100,000, then when the
account takes lossess and equity drop to $50,000 then the account has
hit ruin. All trading must stop at this point.

This is proper money management for trading system. Sadly, a lot of
peope do not practice this. This is why I have been saying trading
system is a lot of different from plain technical analysis looking at
indicators or software churning out buy here sell there type of things
without consideration of the trading proper.

For the Turtles, ruin is 50%. For funds, ruin can be anything from 25%
to 35%. To repeat, once the account hit ruin, no more trading is allowed.

As you can see, this prevent the situation where one suffer a total loss.
Of course there may be abnormal market situation or condition where
losses can still be more than ruin. But in normal trading this is the
protection from proper money management rules.

Just to share that Dr Elder Alexandra has the following money management rules that applies to discretionary trading:

1. 2% rule: limit the losses of any trade to 2% of equity in trading account. This is similar to what the turtle trading position sizing.

2. Whenever the value of trading account dips 6% below its closing value at the end of last month, stop trading for the rest of this month. Close all open positions and stay side line, continue to monitor the market. Review the trades done and explore the reason for such loses. This rule stop traders from trading when facing a string of losses, be the reason be he is losing his mind or his trading style does not suit the current market.

(Dear new forum members: I have posted here a discussion on position sizing that is related to the risk management, check it out if you are interested:
http://z7.invisionfree.com/ChartistsUnited...post&p=1820681)

2% rule: limit the losses of any trade to 2% of equity in trading account.

Trading method A
Let's say I have a RSI divergence, how much should I bet ?
I guess 2% is a good starting point

Trading method B
Suppose someone offers DBS for $0.30,
how much will you bet ?

As much as possible.
(although sgx will bust the trade)

So trading method A gets 2 marks and B gets 100 marks.

---
If sgx bust a trade,
then the 100 marks will be voided.

But what happened to Nikkei on April 2000 ?
Too long ago ?

What happened to MSCI taiwan on 3 March 2007 ?