Tech/a Techtrader - my questions

[TjamesX]

I have some questions I would like to ask Tech/A regarding the Techtrader system, and rather than just PM'ing him I thought it may be worthwhile (if he is kind enough) to provide some answers for all to see.

I have read numerous threads about the techtrader system and no doubt there has been many many many man hours developing and experience gone into it. I myself am not cut out for pure technicals and the amount of time required to develop the skills required for sort of analysis. (Having said that I'm not adverse to leveraging of other people that have)

I have a number of questions and I'll keep them all in this thread, but I'll start with only a couple. Some may have been answered before - apologies, but maybe we can consolidate them in this thread.

Tech/A - whether you answer the questions is obviously up to you. If you dont want to answer any thats fine.

TT = techtrader

Q1) How far back (data wise) have you tested TT, and what is the earliest time for which you have meaningful data to test the system?

Q2) What is the longest period of time that a portfolio trading TT has been negative (in loss - using zero leverage)? On second thoughts this is incredibly data intensive. I suppose you could measure it with portfolio entries at yearly intervals starting from the earliest period for which you have meaningful data (ie answer to Q1). If you wanted to increse the resolution of results, look at portfolio entries at every half year period or monthly period. If you insane, resolution of portfolio entries for every day!!........... You don't have to answer this question - this is more of a project

Q3) As a substitute for Q2 - What was the portfolio return using TT with entry on 01 Jan 2000 and finish at the low for the XAO (around Arpil 03) using no leverage?

Cheers
TJ

PS more questions to come..............................

 

[TjamesX]

With reference to Q3 - here's a graph of the XAO for the last 10 years.

 

[tech/a]

Tjames.

Thanks for the questions.

(1) I have data back 8 yrs and yes this is a problem for those testing methods.
Infact I never clean my data---havent done so for 3 yrs and as a result have stocks that are no longer in the data downloads as they are delisted.This gives some more accuracy to testing.But clean accurate data is a huge issue.Not so much for portfolio traders but definately for Futures or Forex traders using tick data format.
So 8 yrs.

(2) Its not just a matter of comparing equity graph's v the All Ords.
For there are 2
1------closed trade equity this graph is less than impressive as it only records CLOSED trades.---When trading very longterm---years in T/T's case in some trades we can have a great deal of consecutive losses while a full 10 share portfolio is being purchased.So this graph looks terrible.---MEANWHILE

2------Whats not being recorded on the graph is OPEN Equity---the on going unslod profit that remains in the method until its sold out or a trade triggers an exit.This graph is MUCH smoother.I havent run these over the last 8 yrs for ages but off the top of my head I think the biggest DRAWDOWN on CLOSED TRADES was around 20%---however open trades had the method in profit continually.---The initial Drawdown was around 12%---thats the figure needed according to testing to get the method up and running.
Ill run it again and post both graphs--both closed and open equity,at latest Saturday.

3----This is a common question asked---How does the method perform in less than ideal conditions.
Firstly testing over short timeframes is pretty impossible as the AVERAGE trade time length for winning trades is just under a year (330 days) and losers is average around 40 days.---However we can test open equity as against closed trades so while no conclusive can give an idea.

Secondly the method is a LONG---UPTRENDING trading method designed to catch and ride trends---its not designed to out perform markets in downturns.
In downturns we do get drawdowns.However I have been working on reducing even these times of drawdown---more to the point reducing the drawdown.This can be done buy trading the equity curve or the index curve as a 'Portfolio" exit methodology.The Amibroker experts have been kind enough to lend a hand testing this idea (As Metastock/Tradesim combo cant do this). Their results so far in early days are that impressive I sold the total of my portfolio about 6 weeks ago when the index fell below the 180 day EMA.As doing this increased profit 3 fold---in tests.

Ive just returned to the market with my long term portfolio---Time will tell if my personal result is as good as testing---I left my re entry later than the testing model---but Im up on holding the portfolio---except for bloody tax.!

I can and will also run these periods in answer to your question for you,but place all results in context of the methods character.

tech

 

[TjamesX]

Tech, with regards to Q1 and Q2, yes I would be evaluating the system on closed trades plus equity position at any one point in time. So if we entered the market at any point in time with say 100k cash, I would monitor TT performance at any point in time by how much cash is left in the trading account plus the value of all equity positions (ie total assets). So any comparison percentage wise of TT I would look at the % rise/fall of total assets compared to original assets (100k).

I assume (if I rememeber correctly) that in backtesting TT there is no account for dividends/franking credits recieved - but this is offset by the non inclusion of brokerage on trades.... ie we'll assume they cancel each other out.

The reason for Q2 was to basically analyse a whole basket of portfolios using TT, but entering the market at different intervals ie 8 scenarios, each with their start date 1 year apart (for 8 year data set) - or if you wanted to get more detailed analysis 8x12 = 96 portfoios with their start date being 1 month apart. Then for each portfolio we'd analyse the results over the life of the portfolio monitoring time at which point the return was worst (open + closed trades), what was the longest period of time the total return (open + closed trades) was negative. I think this would give an accurate picture (over the 8 years of data) of the periods where TT did its best and worst relative the the market - but obviously is a lot of work, which is why I eyeballed the chart and chose the period to test for Q3.

The reason i chose this period was to incorporate analysis of TT in a downmarket, but also give it enough time (> 3 years) to provide meaningful analysis. I'm not expecting TT to return positive, but am very interested in the results.

OK..... sneaking in another question

Q4) A day in the life of TT - using TT you have mentioned that you only use a small amount of your time out of your day implementing the system. So on an average day what would you do...... ie

When do you run TT, each day? Do you download daily EOD data and Run?
How do you implement buys/sells TT has flagged (ie online broker, full service broker, automatically?)
How do you determine position size when entering a trade?
Do you implement all buy/sell signals or some? If only some how do you chose which ones? Does TT rate the strength of buy/sell signals relative to each other?

ooooohhhhh i'm getting greedy now - was that 1 question or 10!!

Cheers
TJ

 

[tech/a]

The reason for Q2 was to basically analyse a whole basket of portfolios using TT, but entering the market at different intervals ie 8 scenarios, each with their start date 1 year apart (for 8 year data set) - or if you wanted to get more detailed analysis 8x12 = 96 portfoios with their start date being 1 month apart. Then for each portfolio we'd analyse the results over the life of the portfolio monitoring time at which point the return was worst (open + closed trades), what was the longest period of time the total return (open + closed trades) was negative. I think this would give an accurate picture (over the 8 years of data) of the periods where TT did its best and worst relative the the market - but obviously is a lot of work, which is why I eyeballed the chart and chose the period to test for Q3.


Cheers
TJ

I have the benifit of MonteCarlo analysis.
Hence the method has been tested over 20000 portfolio's.There is a 100% success rate---IE not one of the 20000 returned a loss.I will also run the Monte Carlo analysis and post the results.--Chart form seems to be the best way for people to understand or at least visualise.
For those who dont understand Monte Carlo analysis its simplest explaination is --

Give 20000 traders each $100,000 and your method and tell them at different intervals to go trade your method and bring back the detailed results in 8 yrs.

Fortunately I can run the test in a few minutes.
Much was gleened by this analysis.
Standard deviation of wins and losses and return was small under 10% if I recall.Not EVERY trade needs to be taken to be profitable infact there are more buy signals than capital could possibly take.

OK..... sneaking in another question

Q4) A day in the life of TT - using TT you have mentioned that you only use a small amount of your time out of your day implementing the system. So on an average day what would you do...... ie

When do you run TT, each day? Do you download daily EOD data and Run?
How do you implement buys/sells TT has flagged (ie online broker, full service broker, automatically?)
How do you determine position size when entering a trade?
Do you implement all buy/sell signals or some? If only some how do you chose which ones? Does TT rate the strength of buy/sell signals relative to each other?

TJ

On an average day I'd just check price which is on my desk in the form of live data everyday. Unless Im closing something or buying something my time is negligable.When starting up it could take 30 mins a day for a month or so until all is up and going.

Just out of curiosity I run a search most days just to see whats flagged.
My veiw is I should be running more than 1 portfolio using this method but I'm not going to until Ive finished testing ideas with the Amibroker people.I want fixed rules not wishy washy ones. I use a full srvice broker.Only because of the ease and the low number of trades I do. Total around 40 a year over 3 methods.

Position size is a flat 10% of capital as it increase so does the parcel size.
This is a little different to testing which is $10000 a trade.

I only buy some and I do eyeball the charts picked and UNLIKE the systems test I wont trade some charts.I have 2 eyeball criteria which I have made known from implementation----this could be one reason why the realtime results outstripped the test results---but I cant test it so cant be sure.

(1) The chart must be in an OBVIOUS uptrend or in an OBVIOUS break of a downtrend.
(2) The chart cannot have been ranging like between $1 and 1.60 over 5 yrs it must have upside potential.

So as Excal said there is a very small element of discretion in my trading.

As for Questions.
I welcome them. This gives peoplereading a chance to follow how a method was developed and the questions are often those that people either would love to know and werent confident to ask or they dont know what to ask.
Every now and then i get a question that stumps me or I didnt consider---these questions are the ones that help me learn more and I want to learn all I can as I'm putting my hard earned on the line---so it better be right!!

 

[RodC]

Hi tech,

Why do you think you should be running more than 1 portfolio? Are these slightly different versions?

I'm actually running a slight variant of TT myself and as you've said it certainly doesn't take much time once it's up and running. I usually do a scan daily as well (Amibroker) and there are certainly more buy signals than the capital allows.

I'm also interested in your exit from all positions when the index dropped below the 180EMA. This seems a little discretionary as I imagine most stocks in the portfolio didn't signal an exit.

regards,

Rod.

 

[TjamesX]

Thanks Tech

I will also run the Monte Carlo analysis and post the results.--Chart form seems to be the best way for people to understand or at least visualise.

Looking forward to it

I only buy some and I do eyeball the charts picked and UNLIKE the systems test I wont trade some charts.I have 2 eyeball criteria which I have made known from implementation----this could be one reason why the realtime results outstripped the test results---but I cant test it so cant be sure.

(1) The chart must be in an OBVIOUS uptrend or in an OBVIOUS break of a downtrend.
(2) The chart cannot have been ranging like between $1 and 1.60 over 5 yrs it must have upside potential.

So it looks like we've covered the buy side. TT promted with some discretioanry T/A applied to individual stock charts and an entry into some. It sounds as if the discretionary part is optional, but you have been able to outperform the purely mechanical TT, but you're unsure whether this is because the discretion part is actually better or you've been lucky so far.

So onto exiting positions;

Q5) I am aware that you use a stop loss. Is this stop loss standard for all trades (in % terms)? Does the stop loss get revised up as the price increases? How does it get revised up (ie instructions to broker)?

I have never used stop losses, so i'm assuming that if the broker is instructed on a stop loss the instruction is entered into a system in some sort of automatic way such that it requires no human interaction? From there the stop loss order can be modified/adjusted by the broker?

Q6) Does TT signal other sell orders (ie that are not standard stop loss triggers)? If so do you act on those signals? Do you use your own discretion in exiting sometimes?

Q7) In the event of a sell trigger from any of the above - is it a total exit of position at all times? Or is there sometimes a partial sale or profit taking where some of the position is still held onto?

Cheers,
TJ

 

[tech/a]

Basically yes thats right,

Q5 Its the same 10% of purchase price for all trades.
IE buy$1 stop $90c.

No thats where it stays.Its is not raised.

I watch the stop myself with 10 trades going its no problem particularly when I have a ticker on the desk.

Q6 The exit is cross of close below a 180 day exponential M/A of the low.No I dont use my own discretion.
You could say I did receiently on evidence that selling when the index is below the 180EMA gave better results.Sadly I'm worse off that those that followed the method( my own!!) to the letter due to TAX.

Q7 Its a total exit.

I'm also investigating holding the profit when exiting EG sell the initial cost of the trade---and then just hold profit ---but this is a work in progress.

tech

 

[TjamesX]

To avoid duplication that may occur, I found Tech's post of the actual formula behind TT.

Mark there are 300 + posts on TechTrader here and a great grounding for the method.

http://www.reefcap.com/ubb/Forum8/HTML/000091.html

Im happy to use the Method as a live case example as its been released for public scrutiny for 2.5yrs.I do have another method I trade which will remain proprietry.For those with Metastock here is the code.

ENTRY

Cross(H,Ref(HHV(H,10),-1)) AND H > Mov(C,40,E) AND HHVBars(H,70)=0 AND C < 10.00 AND C > O AND Fml("Liquidity") > 500000;

[In English,Todays high is greater than the highest high of the last 10 periods,AND the High is greater than the moving average of the close over the last 40 periods,AND todays Bar is the Highest High value of the last 70 periods,AND C is Less than $10,AND Close is Greater than the open,AND the average trading turnover for the last 21 periods is $500000]

The liquidity formula which should be placed in the INDICATOR BUILDER is

Mov(V * C,21,S) [call it Liquidity]

The system has a INITIAL SET STOP of

If(Ref(C,-1)>0.90*EntryPrice,0.90*EntryPrice,Ref(C,-1));

This is code for TRADESIM and should not be entered in Metastock.
However in English.[If the last close falls below 90% of the entry price then exit]

EXIT

Cross(Ref(Mov(L,180,E),-1),C);

[English. If close Crosses the 180 day moving average of the low on the previous bar then sell this bar.]

ALL BUY,STOP and EXIT signals are traded NEXT DAY on OPEN

tech

This is the thread that the post comes from and covers a lot of stuff about TT

https://www.aussiestockforums.com/forums/showthread.php?t=717&page=1&pp=10&highlight=Techtrader

 

[TjamesX]

Ok after beefing up on a bit more reading...... I have some more Q's

Q8) I realise that the number of stocks flagged to buy using TT is quite large. Do you have actual statistics on the number of flagged buy signals over the 8 year testing period? What is the average num buy signals per year and per month during that period? Is that easy to work out.....?

Q9) You have mentioned that your stock universe is limited by the BT margin trading list (this is also the extent of fundamental analysis ). What universe of stocks do you use in your back testing? Is it all listed companies on the ASX? If its different from the universe you actually trade - have you noticed any differences in performance as a result?

With reference to Q8, as a result of the large Num buy signals generated by TT, actual trades entered involves some form of discretion, whether that be random or eye balling charts for further T/A, So....

Q10) Have you considered any other form of discretionary analysis (Fundamental or technical) on the output of TT to see if it may/may not increase the overall return? If yes, is that part of your proprietory method you use?

Q11) Do you think overall performance can be improved by applying further analysis on the output of TT? Or do you think it will take away from the simplicity of the system?

Cheers
TJ

 

[tech/a]

Ok after beefing up on a bit more reading...... I have some more Q's

Q8) I realise that the number of stocks flagged to buy using TT is quite large. Do you have actual statistics on the number of flagged buy signals over the 8 year testing period? What is the average num buy signals per year and per month during that period? Is that easy to work out.....?

Yes the testing programme tells me that.

Q9) You have mentioned that your stock universe is limited by the BT margin trading list (this is also the extent of fundamental analysis ). What universe of stocks do you use in your back testing? Is it all listed companies on the ASX? If its different from the universe you actually trade - have you noticed any differences in performance as a result?

No the list used is the BT list.
Yes Ive tested many lists including the ASX full list.The performance on the whole list is less than impressive.(Profitable though).
I have also tested many Bourses due to Overseas interest,the best performing bourse of stocks was Hong Kong.
The NYSE list not far behind
DAX and FTSE lists also did well.

With reference to Q8, as a result of the large Num buy signals generated by TT, actual trades entered involves some form of discretion, whether that be random or eye balling charts for further T/A, So....

No thats not so to complete a full portfolio may take a few weeks ALL have been generated by the method so the only "Discretion" is which one.

Q10) Have you considered any other form of discretionary analysis (Fundamental or technical) on the output of TT to see if it may/may not increase the overall return? If yes, is that part of your proprietory method you use?

I have tested 100s of entry tweekings---entry is NOT the most important aspect---which most people cant get a head around---its purely a start point and not far above a random entry---
The proprietry methods are WEEKLY and not remotely the same as T/T.

Q11) Do you think overall performance can be improved by applying further analysis on the output of TT? Or do you think it will take away from the simplicity of the system?

I'm not convinced that T/T has been tweeked to its maximum potential.Infact my own view is that its about a 7.5/10 rating--purely because I have some tweeking and other methods to compare it with.A few I know who use it think its the Ducks Guts.I encourage all to have a play with it--I'd appreciate the feed back--particularly if there is improvement.Some trade it as it is and others tweek it to their style.Fine thats what its for---a foundation---a learning curve. Or simply a method that works---most people doubt that one actually exists--well here is one and FREE!

 

[tech/a]

OK some Charts.

Firstly the RAW as in not tweeked original Techtrader.
The report below shows a lot of answers to TJ's questions so let me know if its not clear.

Chart 1 is a comparison of OPEN EQUITY to CLOSED EQUITY the top line is ofcoarse the open equity.

Chart 2 is a MonteCarlo simulation of 5000 portfolios I can do whatever number you want but chose 5000 this run.

Chart 3 is Yearly Profit or loss of all CLOSED trades

Table 2 is the Monte Carlo stats.

Detailed Report
(TTrader BT margin Master 1)

Simulation Summary
Simulation Date: 6/23/2005
Simulation Time: 8:37:45 PM
Simulation Duration: 0.54 seconds

Trade Summary
Earliest Entry Date in the Trade Database: 1/2/1997
Latest Entry Date in the Trade Database: 6/22/2005
Earliest Exit Date in the Trade Database: 1/17/1997
Latest Exit Date in the Trade Database: 6/23/2005

Start Trade Entry Date: 1/2/1997
Stop Trade Entry Date: 6/22/2005
First Entry Date: 1/2/1997
Last Entry Date: 6/3/2005
First Exit Date: 1/17/1997
Last Exit Date: 6/23/2005

Total Trading duration: 3094 days

Profit Summary
Profit Status: PROFITABLE
Starting Capital: $100,000.00
Finishing Capital: $352,578.29
Maximum Equity/(Date): $252,578.29 (6/23/2005)
Minimum Equity/(Date): ($1,319.47) (3/3/1997)
Gross Trade Profit: $412,766.40 (412.77%)
Gross Trade Loss: ($160,188.11) (-160.19%)
Total Net Profit: $252,578.29 (252.58%)
Average Profit per Trade: $1,180.27
Profit Factor: 2.5768
Profit Index: 61.19%
Total Transaction Cost: $12,840.00
Total Slippage: $0.00
Daily Compound Interest Rate: 0.0407%
Annualized Compound Interest Rate: 16.0272%

Trade Statistics
Trades Processed: 2739
Trades Taken: 214
Partial Trades Taken: 0
Trades Rejected: 817
Winning Trades: 69 (32.24%)
Losing Trades: 145 (67.76%)
Breakeven Trades: 0 (0.00%)

Normal Exit Trades: 149 (69.63%)
Delayed Normal Exit Trades: 0 (0.00%)
Open Trades: 11 (5.14%)
Protective Stop Exit Trades: 54 (25.23%)
Time Stop Exit Trades: 0 (0.00%)
Profit Stop Exit Trades: 0 (0.00%)

Largest Winning Trade/(Date): $60,816.00 (12/15/2004)
Largest Losing Trade/(Date): ($4,167.50) (4/6/2005)
Average Winning Trade: $5,982.12
Average Losing Trade: ($1,104.75)
Average Win/Average Loss: 5.4149

Trade Duration Statistics
(All Trades)
Maximum Trade Duration: 1226 (days)
Minimum Trade Duration: 1 (days)
Average Trade Duration: 135 (days)
(Winning Trades)
Maximum Trade Duration: 1226 (days)
Minimum Trade Duration: 5 (days)
Average Trade Duration: 316 (days)
(Losing Trades)
Maximum Trade Duration: 222 (days)
Minimum Trade Duration: 1 (days)
Average Trade Duration: 48 (days)

Consecutive Trade Statistics
Maximum consecutive winning trades: 7
Maximum consecutive losing trades: 15
Average consecutive winning trades: 1.73
Average consecutive losing trades: 3.63

Trade Expectation Statistics
Normalized Expectation per dollar risked: $0.8800
Maximum Reward/Risk ratio: 47.79
Minimum Reward/Risk ratio: -1.51
Average Positive Reward/Risk ratio: 4.07
Average Negative Reward/Risk ratio: -0.64

Relative Drawdown
Maximum Dollar Drawdown/(Date): $16,762.99 (10/1/2002)
Maximum Percentage Drawdown/(Date): 8.6120% (2/25/2000)

Absolute (Peak-to-Valley) Dollar Drawdown
Maximum Dollar Drawdown: $22,161.19 (18.2600%)
Capital Peak/(Date): $121,392.44 (8/7/1998)
Capital Valley/(Date): $99,231.25 (2/25/2000)

Absolute (Peak-to-Valley) Percent Drawdown
Maximum Percentage Drawdown: 18.2600% ($22,161.19)
Capital Peak/(Date): $121,392.44 (8/7/1998)
Capital Valley/(Date): $99,231.25 (2/25/2000)

Monte Carlo Report
(TTrader BT margin Master 1)

Simulation Summary
Simulation Date: 6/23/2005
Simulation Time: 9:01:18 PM
Simulation Duration: 26.77 seconds

Trade Parameters
Initial Capital: $100,000.00
Portfolio Limit: 100.00%
Maximum number of open positions: 100
Position Size Model: Fixed Percent Risk
Percentage of capital risked per trade: 2.00%
Position size limit: 10.00%
Portfolio Heat: 30.00%
Pyramid profits: Yes
Transaction cost (Trade Entry): $30.00
Transaction cost (Trade Exit): $30.00
Margin Requirement: 100.00%

Trade Preferences
Trading Instrument: Stocks
Break Even Trades: Process separately
Trade Position Type: Process long trades only
Entry Order Type: Default Order
Exit Order Type: Default Order
Minimum Trade Size: $0.00
Accept Partial Trades: No
Volume Filter: Ignore Volume Information
Pyramid Trades: No
Use Level Zero trades only: Yes

Simulation Stats
Number of trade simulations: 5000
Trades processed per simulation: 2739
Maximum Number of Trades Executed: 216
Average Number of Trades Executed: 194
Minimum Number of Trades Executed: 175
Standard Deviation: 5.94

Profit Stats
Maximum Profit: $553,152.78 (553.15%)
Average Profit: $314,285.62 (314.29%)
Minimum Profit: $193,884.65 (193.88%)
Standard Deviation: $57,687.96 (57.69%)
Probability of Profit: 100.00%
Probability of Loss: 0.00%

Percent Winning Trade Stats
Maximum percentage of winning trades: 39.78%
Average percentage of winning trades: 35.35%
Minimum percentage of winning trades: 30.56%
Standard Deviation: 1.37%

Percent Losing Trade Stats
Maximum percentage of losing trades: 69.44%
Average percentage of losing Trades: 64.65%
Minimum percentage of losing trades: 60.22%
Standard Deviation: 1.37%

Average Relative Dollar Drawdown Stats
Maximum of the Average Relative Dollar Drawdown: $3,967.01
Average of the Average Relative Dollar Drawdown: $2,367.15
Minimum of the Average Relative Dollar Drawdown: $1,691.01
Standard Deviation: $278.56

Average Relative Percent Drawdown Stats
Maximum of the Average Relative Percent Drawdown: 2.0524%
Average of the Average Relative Percent Drawdown: 1.2679%
Minimum of the Average Relative Percent Drawdown: 0.9202%
Standard Deviation: 0.1485%

Maximum Peak-to-Valley Dollar Drawdown Stats
Maximum Absolute Dollar Drawdown: $47,454.57
Average Absolute Dollar Drawdown: $25,021.31
Minimum Absolute Dollar Drawdown: $14,831.98
Standard Deviation: $5,233.97

Maximum Peak-to-Valley Percent Drawdown Stats
Maximum Absolute Percent Drawdown: 16.7127%
Average Absolute Percent Drawdown: 11.1289%
Minimum Absolute Percent Drawdown: 6.0137%
Standard Deviation: 2.1266%

 

[TjamesX]

Thanks alot for the simulations and charts tech!!

I have tested 100s of entry tweekings---entry is NOT the most important aspect---which most people cant get a head around---its purely a start point and not far above a random entry---

I think I've had a bit of an aha!! moment.

I have a couple of questions regarding the simulations....

For simulation 1 - that was a single portfolio with $100,000 starting capital. I am assuming that obviously the simulation did not trade every buy signal from TT otherwise it would have run out of $$. So did it randomly pick which trades TT signaled to enter, or did it trade every signal until it had no money, and then wait for capital to be released?

For simulation 2 (monte carlo) - I am assuming that all potfolios started at Jan 1997 and ended Jul 2005, the only difference being which stocks they actually picked to trade?

I believe that both simulations were using no margin on any trades?

OK, so some analysis on the results;

I assuming that the best data to analyse TT performance is the average profit realised from Monte Carlo sims. This being $314,285.62, which represents an increase in total assets from $100,000 to $414,285.

Over the trade period (approx 8.5 years) this represents an annual compound return of 18.20% . Eyeballing a chart of the XAO, Jan 1997 was about 2350 and yesterday it closed at 4225... which equates to a compound return of 7.14%. Biiig difference.... and where does it come from. Random entry to stocks at the start of 1997 and holding, should return you about 7.14% compound over this period. Somehow during the trading period TT has been able to extract more value out of the rise in the All Ords..... and if TT entry signal is not much better than random entry, then its not here. So the outperformance of the system could be mostly attributed to working out when to exit a stock. In essence, TT has caused the portfolio to (in an uptrending overall market);

Hold stocks that will outperform relative to All Ords
Sell stocks that will underperform relative to All ords

All in two simple exit criteria....

Is my analysis wrong... or is that just weird.

TJ

 

[tech/a]

I think I've had a bit of an aha!! moment.

Excellent!

I have a couple of questions regarding the simulations....

For simulation 1 - that was a single portfolio with $100,000 starting capital. I am assuming that obviously the simulation did not trade every buy signal from TT otherwise it would have run out of $$. So did it randomly pick which trades TT signaled to enter, or did it trade every signal until it had no money, and then wait for capital to be released?

The latter

For simulation 2 (monte carlo) - I am assuming that all potfolios started at Jan 1997 and ended Jul 2005, the only difference being which stocks they actually picked to trade?

The portfolios are picked at random around the date of 1/1/97 buys go out as far as is needed to simulate x number of portfolio's.All close on the last date specified.

I believe that both simulations were using no margin on any trades?

Well thats a matter of perspective!!
$100,000 initial capital OR $30,000 on Margin which would equate to $100,000 on Margin---approx.Of course I'd prefer to look at it on Margin as the $250,000 profit is 800% on initial capital where as on $100,000 its 250%.

OK, so some analysis on the results;

I assuming that the best data to analyse TT performance is the average profit realised from Monte Carlo sims. This being $314,285.62, which represents an increase in total assets from $100,000 to $414,285.

Over the trade period (approx 8.5 years) this represents an annual compound return of 18.20% . Eyeballing a chart of the XAO, Jan 1997 was about 2350 and yesterday it closed at 4225... which equates to a compound return of 7.14%. Biiig difference.... and where does it come from. Random entry to stocks at the start of 1997 and holding, should return you about 7.14% compound over this period. Somehow during the trading period TT has been able to extract more value out of the rise in the All Ords..... and if TT entry signal is not much better than random entry, then its not here. So the outperformance of the system could be mostly attributed to working out when to exit a stock. In essence, TT has caused the portfolio to (in an uptrending overall market);

No the out performance comes from being able to find and stay with trades which have outperformed the ORDS.This is where our discussion will get very interesting as most people just cant understand this fact of trading--(Which fact)--the above.

Hold stocks that will outperform relative to All Ords
Sell stocks that will underperform relative to All ords

Id love to be able to use the word WILL. But we cant as this is NOT predictive---its reactive.All we can do is set parameters or a framework to trade in and test that if that framework is adhered to then we can profit and as can be seen that can vary between 100% over 8 yrs to 550% even that we dont know---hence the answer to RODS question of why I think I should be trading multiple portfolio's of the same method.

All in two simple exit criteria....

One is a stop and one an exit.
I'm yet to find the perfect exit and have tried 100s.
Exit is a much more complex issue than entry and although it seems very simple it isnt---why did I choose 180 day EMA of the low? (Thats a seperate discussion).

Is my analysis wrong... or is that just weird.

No its inquisitive,analytical and I'm sure enlightening.All traits necessary to move forward in any persuit of understanding.I dont know all the answers/truths but those I have found I'm happy to share.

 

[RodC]

tech.

Maybe it's just me. But I'm having still trouble understanding what you gain by trading multiple portfolios (of the same method) compared to say, 1 large portfolio (with more stocks in it). eg 5 portfolios of 10 stocks or 1 portfolio of 50 stocks.

Isn't it going to average out the same (roughly)?

Rod.

 

[tech/a]

Rod.

Yeh well its possible to look at it from a number of veiw points.

Here is mine.
I know that from 5000 portfolio tests that not ALL will trade the same infact the difference is dramatic.150% to 550% return.according to testing.

Now I maybe lucky and get on the 550% one or Unlucky and get on the 150%. ( IE The portfolio I end up filling and consequently trading.)

I could do as you say but I'm not going to be sure which one I'm on until years of trading has elapsed.
I feel trading different portfolio's of the same AND/OR different portfolio's of a different method/s (which will also have their high return and low return portfolio's) is mitigating risk----that I wont get in the lowest one---rather than maximising profit that I can get in the highest one---Thats just the way I think---defensive---and if I get that right the rest will follow.


Maybe its good thinking,maybe some think its not.
If I had data back 50 yrs it may show that the deviation is non existant over that time and If I was 20 then it wouldnt matter.(This can NEVER happen (Have a tested result) as its impossible to have a margin list accurate after 6 mths---yes another topic re accuracy of results from present day data----we will be here months discussing this---no problem---gets people thinking out of the square).

At 50 I need it right for the next 10 yrs so I can enjoy it!!

 

[RodC]

thanks tech,

I think I know what you mean.

Rod.

 

[TjamesX]

OK,

I'm going to pursue what you mentioned about TT entry signals not being much above random entry in improving performance.

Have you actually quantified how much of a difference there is between

a) Random entry and TT exit/stop loss.........and
b) TT Entry and TT exit/stop loss

Can you post comparitive results of a Monte Carlo sim of say 5000 portfolios using each method with the same trading period as before?

I'm still trying to work out exactly where the difference between 18.20% and 7.14% comes from, and I would like to know exactly how much TT entry has a part in this (every % point counts!).

If most of the difference can be attributed to the stop loss and 180 EMA - wow, for me thats pretty significant...

Why is it significant for me? Because;

My original thoughts were that I could possibly apply some of my portfolio to using the TT method, while my normal mostly fundamental part technical analysis would be used for my other half..... and see what happens. But I think there are significant ways I think the two could be intermixed.....

1) Normally a decision on entry for me would be a combination of economic/cyclical/industry conditions as well as some fundamentals on the particular company, maybe a little TA - but rarely. With TT entry signals there is no reason why i couldn't overly my analysis on top to chose/rate which if any to enter. This is because economic/cyclical/industry conditions change very slowly and would not be hard to apply, and there are certain fundamental analysis techniques that can be applied across the board to rate companies very quickly on their financial position. One of these methods has been used very successfully discussed and applied on another forum - will discuss further down the track......

and

2) Depending on how much difference tech's results come back between random entry and TT entry - there seems to be no reason why I couldn't apply my normal fundamental analysis on entry (ie no TT entry), but then apply TT exit/stop loss criteria once I have entered. The reason is becuase I have always struggled when to exit stocks to maximise my returns.

The question of whether any value can be added through my own methods 1 & 2 is obviously a big fat ???? and there is really little way of back testing easily any fundamental entry...........

so it continues

TJ

 

[RodC]

TJ,

I reckon dome of the difference between the 18.2% and the 7.14% would be due to the "universe of stocks" that TT is working with (the BT Margin List).

tech/a mentioned in one of his earlier posts that TT didn't perform as well when tested on the whole All Ords.

Rod.

 

[TjamesX]

I reckon some of the difference between the 18.2% and the 7.14% would be due to the "universe of stocks" that TT is working with (the BT Margin List).

Good point Rod.... it would be interesting to see the index return for the BT margin trading list over that time - the XJO would possibly be a better index to compare TT with, but that only has data for the last 5 years.

Maybe we should all limit our of stock universe to the BT Margin List

TJ