Supernova System Journal

LOL... and a real good one there!!

Bye-bye!! http://www.jubailforums.com/vb35/ima...ilies/bye1.gif

Closing out for the night as well.

See you 2 morrowwwwwww

Hiho... back after quite little sleep!

I have been looking after the Trendchaser idea:

http://www.forexfactory.com/showthre...t=56248&page=5

It is in posts #64-66 and has been mentioned earlier in the thread, but there he shows that the quality does not matter with entries & exits on opening or closing candles! Maybe tests will also be faster with less details.

(The grain pot must be ruined , but I hope you are fine again!)

Greetz
GMatts

Good Morning!

Looks like you guys covered a fair bit of ground yesterday and possibly ruined a couple keyboards - LOL!

Well, Qpip - I am certainly sorry to hear of your misadventures at work. Sounds like a rough situation - just dealing with the stress caused by the foolishness of others. That is exactly why I quit my job almost 3 years ago now. Only it was my #1 boss causing the problem and he was owner of the company - not much room to appeal up the chain of command there. My #2 boss was actually pretty decent (and even a GIRL). I would have been followed by about 5 others from the department had it not been for the fact that they had families to feed and nowhere else to go. I was just in a good situation at home which allowed me to leave.

I did try testing the settings you emailed to all of us but did not report my results to you as I was just learning how to use the EA for the first time and was not at all sure that I was even doing that right. I was trying to make sure that I understood how to use it properly before I started passing results around - LOL! Then I ran into that snag with the PSAR/CCI thing and sort of put the EA on hold until that was sorted out.

It looks like Gnat and I will have time to explore that a bit more in the next couple weeks and perhaps during/after the holidays we can all get together on it again.

I am also looking at ways to improve my manual trading (at least for myself). I have gotten myself in a bit of a jam sleepwise by trading the whole London session so am looking at possibly cutting back on that a bit and using my alerts more. I don't want my health to suffer and I am also a cranky daddy when I am sleep deprived. I will sort that out this week hopefully.

I should be around all day - at the computer off and on.

Have a good one!

NOFX

Hey Nofx, thanks for the moral support .

I'm so happy for you that your manual trading is sufficient to support you (and a bit jealous lol). However, to the detriment of your health/life schedule, yeah that's not nice. If I ever come close to finishing my bot families, and things forward look like backwards, then I believe you may be able to sleep better/more . I might concentrate my efforts on only GBPJPY or something to speed things up.

But currently I'm at a sort of semi-complete stage and am manually merging data in order to have a feel for the overall potential drawdown (I can live with drawdown, just not drawdown that crosses the red line heheheh (zero)).

Have a good one too.

I'm "online" but working at merging. Will come peek now and then.

C-I-A-O !

and HI GNATT too

Hiho, Qpipsy!

Merging sounds like a spreadsheet of table-like dimensions again! Hope that'll work out nicely. Is your 116 hours test finished with results now? 116h is quite massive - poor PC! :
Testing currency after currency may really be a good idea - will add results after results and then in the end the overall picture can be drawn, while maybe with the clearer/easier pairs there is already some inflow of money.

Hope you are fine with your cooking... did you have to start all over again, or did you just change the recepie?

Greetz
Gmatt

Just found a famous cook to help you! http://www.jubailforums.com/vb35/ima...ies/chefsp.gif And tomorrow I will have an early egg day out of two again. So it is getting up at 5.30. "So good night - well goodnight - good night train is gonna carry me home..." http://www.jubailforums.com/vb35/ima...ies/harhar.gif (from Gerry Rafferty)

Sorry I missed you there, had to take care of holiday SHOPPING.

As for the soup mess, I HAVE (not done yet) to start over. Today was a total rush day: testing, merging, restarting the bots, home budgetting, holiday gift shopping (), send my time sheet, do some laudry, emeil 2-3 people (MIG, amongst others), we had guests over, my b-friend had to go to 3 places to find food (long story) and he's coming down with the sneezers, and finally all that with my brain constantly bring up the "boss no. 2 horror show".

Left on list (I would personally go to bed) is make 4 different batches of soup (8-10 cups each), cool, put into bags (we do this together) then freeze. Then maybe a bath, prepare stuff for tomorrow........... I'm also supposed to make some chocolate chip coookies, but doubt I'll have the stregth.

My 116 hours test ......... this morning I got up and my puter had rebooted (on its own) during the night. Very very nice grrrrrrrrr.

Anyway, I was and still am inspired by you and Nofx who bravely afront this world by believing in themselves and saying "I'm a trader" and going ahead with the plan. This thought is with me at all times.

Amen (lol)

Hi Pipney!

Apart from the shopping stress (each year - what a mess) and the cooking to be redone http://www.jubailforums.com/vb35/ima.../fireman_2.gif (btw.: Don't you have an extra arm to fill the soup into the bags by yourself?? LOL), I hope you had a good solution for your boss#2 treatment today. I would hate him to spoil your day or week cause of his selfimportance.

If you step back a few paces, you will easily find that you do not need us to know that there is a way now to say "I am a trader!". Maybe you just do not get the right feeling for it as boss#2 and the working schedule as such are overshadowing your set of mind right now. As long as your recent tests do not show that the whole thing is worthless (ROTFL - what a thought), then there is no need for that.

Still I know that it is always good to get some input from your "brothers in arms" and you can surely count on us to give you that input. I only wanted to make you see that actually you already know it and would probably not need us! http://www.jubailforums.com/vb35/ima...es/smilie3.gif

And for getting sick: I am today a constant user of the procellain-part of the houses furniture as my stomach is trying to move out and find a nicer place to live! So all the guys are having their pre-christmas sickness right now. Nofx-ter took his sneezers through the weekend - wise move there!!
http://www.jubailforums.com/vb35/ima.../stretcher.gif

Hope you are fine when you return from work!
Greetz and huaaalp!
GMatt

Ya - still got the hacking cough but the headache which was unbearable is now just annoying. Trying to talk myself into trading again tonight in a limited capacity.

NOFX

Hi Pipney!

Just wanted to hint at: http://www.jubailforums.com/vb35/ima...lies/mail2.gif


Hope you are alright with boss#2 or maybe he is... http://www.jubailforums.com/vb35/ima...lies/death.gifby now! LOL

Ciao
Matt

Can someone explain why the ea suddenly stopped making good trades? I'm not talking about the expected losses, but why did it miss the last two big moves, and why the most recent long??? Thanks, Wolfer

edit: I think it may have to do with a set file I created at 12/10 5:00GMT. I was pleased with the returns to that point, saved the file and now this....

Hi Qpip!

Hope the boss #2 did not swallow you alive this week - just checking in with you.

Talk to you soon - still doin it doggystyle this week. Need some time to get the EA going. My head is somewhere else right now with personal shite as you can see in the hive.

Hope you are well,

Corey

Hey all.

As you can see it wasnt too pretty of a week. I think if I had been around to close the open positions before the market close that I would be at the same balance as last week. Hope every one has a good weekend and are enjoying the holidays.

Adam

Hey Adam,

from looking at you trades you may actually have spared you some drawdown on some trades, but especially taking profit a few hours earlier would have made a "lala" outcome into one hell of a week!

Still on the finding journey, but getting closer!

Good weekend!
Matt

Hi Adam,

What was your initial deposit?
I found that the last few weeks were quite difficult and my "curves" (not MINE, the bots ) were slightly down overall. This week was a bit more positive, however, and accounts are oscillating around an average of just about 100$ per day.

Hi GnMatts : (and all),

A good saturday morning and afternoon !!!

I was having a look at some reports and I'm not totally clear on the meanings of two notions, the "profit factor" and the "relative drawdown". For GBPJPY, for instance, I get a PF of 2.18 and a RD of 14.32%.

Could you explain where these figures are from, what they represent, what they're telling me exactly? Does 2.18 mean doubling the capital (plus a few puffs) in the period in question? How is relative DD calculated? Thankarooooooooooos

Hiho, Qpipney!

Here is the first result of the explanatios seeked:

profit ratio = winrate/(1-winrate) * (total won/total loss)

<1.10 you probably will loose your money or be flat
1.10 bare minimum - 90% chance to double capital in 11,000 trades
1.20 slow system - 90% CTD in 1,200 trades
1.30 usable system - 90% CTD in 500 trades
1.40 average system - 90% CTD in 250 trades
1.50 good system - 90% CTD in 190 trades
1.60 very good system - 90% CTD in 125 trades
1.70 excellent system - 90% CTD in 100 trades
1.80 amazing system - 90% ctd in 80 trades
>2.00 holy grail - 90% CTD in 50 trades

This is all based on Risk of Ruin of 0.10%. also, how many times you can trade in a year really determines how holy your grail is.

Break.... and now something different:

Any expert can be tested on history data. After the expert has been tested, the summarized expert testing results and some key characteristics are displayed in the "Report" tab. Reports allow comparing to each other both various experts and the working results of the same expert with different inputs working. This article explains how to read such reports and to interpret the obtained results properly.
Exemplary Report about Testing Results

Let us study the following report about testing results as an example:

http://www.metaquotes.ru/images/expe...ort/report.png

1. 'Bars in test' field displays the depth of the history on which the modeling was based.
2. 'Ticks modelled' field displays the size of the modeled sequence. Every record of the sequence represents the bar state (OHLCV) at one or another moment. Different bar states can be modeled depending on the timeframe, modeling method, and presence of history data from smaller timeframes within a bar.
3. 'Modelling quality' is calculated according to the following formula:
   ModellingQuality = ((0.25*(StartGen-StartBar) + 0.5*(StartGenM1-StartGen) + 0.9*(HistoryTotal-StartGenM1)) / (HistoryTotal-StartBar))*100%;where:
   1. HistoryTotal - the total amount of bars in history;
   2. StartBar - the number of bar with which the testing was started. Modeling starts at at least 101st bar or the bar corresponding with the initial date of test limits;
   3. StartGen - the number of bar with which the modeling on the nearest timeframe started;
   4. StartGenM1 - the number of bar with which the modeling on minutes started;

   at that:

   1. The distance between the beginning of modeling of databases for the nearest timeframe and the beginning of modeling on the nearest timeframe data has a weighting factor of 0.25;
   2. The distance between the beginning of modeling on the nearest timeframe data and the beginning of modeling on minutes has a weighting factor of 0.5;
   3. The distance between the beginning of modeling on minutes and the end of history data has a weighting factor of 0.9;

4. Gross profit, the summed up profit for all profitable transactions;
5. Gross loss, the summed up loss for all unprofitable trasnactions;
6. 
   Total net profit, shows the difference between gross profit and gross loss:
   TotalNetProfit = GrossProfit - GrossLoss
7. 
   Profit factor, shows the ratio between gross profit and gross loss:
   ProfitFactor = GrossProfit / GrossLoss
8. Expected payoff, to be calculated as follows:
   Expected Payoff = (ProfitTrades / TotalTrades) * (GrossProfit / ProfitTrades) - (LossTrades / TotalTrades) * (GrossLoss / LossTrades)where:
   1. TotalTrades - total amount of trades;
   2. ProfitTrades - the amount of profitable transactions;
   3. LossTrades - the amount of losing transactions;
   4. GrossProfit - the summed up profit;
   5. GrossLoss - the summed up loss.

9. Absolute drawdown is the difference between the initial deposit and the smalles value of balance within testing:
   AbsoluteDrawDown = InitialDeposit - MinimalBalance
10. Maximal drawdown is the highest difference between one of local upper extremums of the balance graph and the following lower extremums:
    MaximalDrawDown = Max of (Maximal Peak - next Minimal Peak)The basic stages of changing the maximal drawdown value within testing are given in the picture below. The total maximal drawdown value is in the thick arrows.
    
    http://www.metaquotes.ru/images/expe...alDrawDown.png
    The maximal drawdown percentage shows the ratio between the maximal drawdown and the value of respective local upper extremum:
    MaxDrawDown % = MaxDrawDown / its MaxPeak * 100%

Other results shown in the "Report" tab are obtained using the simplest mathematical calculations.

1. Total trades - total amount of trades made by the expert within testing;
2. Short positions (won %) - total amount of short positions and the precentage of profitable ones among them (profitable short positions / total amount of short positions * 100%);
3. Long positions (won %) - total amount of long positions and the precentage of profitable ones among them (profitable long positions / total amount of long positions * 100%);
4. Profit trades (% of total) - total amount of profitable transactions and percentage of the total amount of transactions (ProfitTrades / TotalTrades * 100%);
5. Loss trades (% of total) - total amount of losing transactions and percentage of the total amount of transactions (LossTrades / TotalTrades * 100%);
6. Largest profit trade - the largest profitable trade among profitable trades;
7. Largest loss trade - the largest losing trade among losing trades;
8. Average profit trade - averaged size of profit of profit trades (GrossProfit / ProfitTrades);
9. Average loss trade - averaged size of loss among losing trades (GrossLoss / LossTrades);
10. Maximum consecutive wins (profit in money) - maximum consecutive amount of wins among profitable series of trades and the summed up profit within this series;
11. Maximum consecutive losses (loss in money) - maximum consecutive amount of losses among losing series of trades and the summed up loss within this series;
12. Maximal consecutive profit (count of wins) - maximum profit of a consecutive series of profitable trades and the amount of trades in this series;
13. Maximal consecutive loss (count of losses) - maximum loss of a consecutive series of losing trades and the amount of trades in this series;
14. Average consecutive wins - averaged amount of trades in consecutive profitable series.
15. Average consecutive losses - averaged amount of trades in consecutive losing series.

Colors used for modeling quality diagram


The following colors are used in the color diagram:

1. <LI style="COLOR: #00ff00">Lime - modeling on minutes, marked with 7 in the picture below.
   <LI style="COLOR: #006000">Deeper green colors show modeling on large timeframes, from M5 to H4.
   <LI style="COLOR: #ffa0a0">Pink color - pure fractal modeling of the databases of smaller timeframe, marked with 2 in the picture.
2. Gray color - modeling limitation by date, marked with 1 in the picture.

http://www.metaquotes.ru/images/expe...ng_quality.pngThe color diagram in the above picture is drawn according to the following initial data of the modeling quality calculation:

1. Bars in test = 4190;
2. StartBar = 2371;
3. StartGen (H4) = 3042 (marked with 3 in the above picture);
4. Start H1 = 3355 (marked with 4);
5. Start M30 = 3841 (marked with 5);
6. Start M15 = 3891 (marked with 6);
7. Start M5 = 0 (no marking in the picture since data on minutes started earlier);
8. Start M1 = 3917.

Having substituted these values into the formula of modeling quality calculation, obtain:
((0.25*(3042-2371) + 0.5*(3917-3042) + 0.9*(4190-3917)) / (4190-2371))*100% =((0.25*671 + 0.5*875 + 0.9*273) / 1819)*100% = 46.78%

Be with you in a while!!

GnMatt

I love maths ... when it is done by others and leads to results!

So here is the max. relative drawdown in a very catchy definition, but it is fine for the scientific standards we apply to our work here!

Be again back in a few timespans!
3 words!
Matt

Hey Qpipney... a little article on how to understand the ratios we use:

Unfortunately no credit can be given to the unknown author... :

Most everyone intuitively knows that it’s harder to bankrupt bigger accounts than it is to bankrupt smaller ones (assuming all else approximately equal). But, for any two account sizes, just how much more or less danger is there? Twice as much? Ten times? Can you put a number on it? In this article, I’ll show how you can numerically judge the relative danger of bankrupting accounts, and then show how you can adjust risk to trade a small account with the same loss profile as bigger accounts.

Definitions

First, lets all get on the same page about what I’m calling “danger.” I talked about the probability that someone using a trading system will sustain enough consecutive losses to empty the trading account. I considered an account to be empty when it drops below the $25,000 freeze point for pattern day traders in the USA. I called the chances that an account would end up empty the “Probability of Financial Ruin.” If that’s not a scary enough name, maybe pick “Probability of Having to Go Back to Your Day Job.”
In this article, I use the same concept, but I drop the probability and trading system, and think purely in terms of the number of losses an account can sustain before reaching the freeze point. This is one way of characterizing the relative danger of trading two account sizes, if you assume most other factors are about equal. For instance, if account A can sustain 40 losses, and account B can sustain 80, then I’ll say account B is twice as “safe” as account A. I’ll also say account A is twice as “scary” as account B. I use “scary” because terms like “risky” have too many meanings already.
We’ll use three example accounts throughout: a $50k account, a $200k account, and a $1Million account.

Risking a Fixed Value per Trade

One approach to trading is to risk a fixed dollar amount per trade. The dollar amount chosen is typically a percentage of the initial trading stake. I don’t recommend this approach for any account size, and the numbers below will show you why.
The number of losses an account can sustain, under this approach is:
http://www.movethemarkets.com/blog/w.../equation1.png
Where “n” is the number of consecutive losses an account can sustain, “stake” is the amount of money a trader starts with, and “pct” is the percent of the initial stake that a trader chooses to use as the amount to risk per trade.
Here’s how our accounts fare under this approach, assuming they risk 2% of their initial stake on each trade:
$50k account can lose 25 consecutive times
$200k account can lose 43 consecutive times.
$1Mil account can lose 48 consecutive times.
So, from this perspective, a $50k account is 1.72 times and 1.95 times more scary than the respective larger accounts. Notice that the larger accounts look a lot closer, with the $200k only 1.11 times as scary as the $1Mil account. This makes sense, because if it weren’t for the $25,000 freeze point, all accounts would have a worst losing streak of 50 trades, at 2% each. Even a $2Billion account could only lose 50 times in a row this way. Surely there is something better?

Risking Fixed Pct of Current Equity per Trade

This is the approach I recommend, and the approach most traders I know use. Instead of risking a fixed value, you risk a fixed percentage of whatever your current account value is. The advantages of this approach, such as risking smaller amounts during losing streaks, have been discussed in detail elsewhere. Here, I’m only concerned with the practical effects related to financial ruin.
With this approach, the number of losses an account can sustain is:
http://www.movethemarkets.com/blog/w.../equation2.png
Again, “n” is the number of sustainable consecutive losses, “stake” is the amount of money the trader starts with, and this time “pct” is the percent of current account value risked per trade.
Here’s how our accounts fare under this approach, assuming 2% risk per trade:
$50k account can lose 34 consecutive times
$200k account can lose 103 consecutive times.
$1Mil account can lose 182 consecutive times.
First, note that this is a less scary picture all around than the previous approach. Also, this approach scales better with account size. As such, the $50k is 3.2 times and 5.6 times more scary than the respective bigger accounts. The $1Mil account owner should be happy to spot that the $200k account is 1.77 times more scary with this approach. At least he or she gets something for all that wealth this time!
As it seems better in every way, this is the trading approach assumed for the rest of this article.

Reducing Small Account Danger

So, now that we have a numerical way to judge the relative danger level of different account sizes, we can also numerically even out the playing field by adjusting the percent risked. In plainer english, we know that a small account will always be scarier than a large account when they both risk the same percent of current equity. We want to know how much less a small account should risk than a larger account, if it wants to be just as safe as the larger account.
This rather ugly equation will tell you just that:
http://www.movethemarkets.com/blog/w.../equation3.png
Here, “pct” and “stake” have the same meanings as in the previous section. They are subscripted “little” and “big” to differentiate the values for the two account sizes in question.
This tells us that if a $200k account wants to be as safe as a $1Mil 2% account, it should only risk 1.13%. To double check, we can use the formula from the previous section to calculate that at 1.13%, it can lose 182 consecutive times. That is exactly as safe as the $1Mil account at 2%, so this equation works.
As you might expect, our smaller account must cut risk much more sharply to be as safe as the bigger players. For the $50k account to be as safe as the $200k 2% account, it should only risk 0.67%. To be as safe as the $1Mil 2% account, it should only risk 0.38%. So, initially, that’d be a risk of $335 and $190 per trade, respectively.

A Slightly Different Spin

Instead of comparing account sizes, we can also just ask, “what percent can I risk if I want to be able to sustain n consecutive losses?” This equation will tell us that:
http://www.movethemarkets.com/blog/w.../equation4.png
Summary

We’ve seen a way to judge the relative safety of different accounts and risk amounts, in terms of the number of losses it would take to freeze the account. We’ve also seen a couple of ways to tailor the percent risked so that an account will have the desired level of safety.
As always with these articles, I hope there was some food for thought for you in here. There’s a lot of general discussion about percent risked, but I don’t see anyone spelling the consequences out numerically. And, for some reason, it seems like nobody takes the $25,000 pattern day trader threshold into account, even though it’s a very important number for stock traders.

Good saturday to you, too!

I hope you came out of the working week without the dreaded final showdown with Agent Asshole the Second! The weather seems to be quite christmaslike already in your region - well here it is only rather cold without any snow, but stiff winds, too!

Well, I guess I gave you enough to read for the first installment, but I am standing by to "listen and talk" as you return from the daily95 never got them pips!

Greetz!
GnMatt http://www.jubailforums.com/vb35/ima...1ughbounce.gif