my journal 2

[Yamato]

I can't help noticing and noting that you're using the term "forward-testing" in a different way than I use it.

For you it seems to be a synonym for "out-of-sample". And in some cases you use it even as a synonym for "in-sample."

I can't help but point out that this is a relevant detail, causing potential misunderstandings.

For me instead "in-sample" is the first part of the data, the one you can optimize your systems on. The "out-of-sample" is the second part of your data, which you've kept hidden, and you use to verify if your system works.

The "forward-testing" means running your systems in real-time on TWS, with real data and real prices, regardless of whether you execute those trades or not (the results are almost identical).

I can't help arguing with you. I can't help but argue with you.

 

[meanreversion]

How do you quantify whether performance on OOS (out of sample) is acceptable? Presumably you have a couple of metrics that you use to determine the usefulness of a system - how does the system need to perform in OOS compared to IS, in order that you would consider taking it live?

 

[Yamato]

Yes, good question.

I don't automate any systems where the OOS doesn't at least make some money. That is the only constant rule of thumb metric I can state clearly and with confidence. The rest varies from system to system, and I don't remember exactly what other rule-of-thumb metrics I applied (I can't go too far with univocal formulas, but I suck at formulas). But that requirement is always valid. On average, though, I can tell you that almost all of the systems I automated have delivered in the out of sample at least half of the money they delivered in the in-sample (considering in the measurement the amount of trades as well).

That is the requirement for automating.

The requirement for trading live is a sharpe ratio above 1.5 in forward-testing and at least 10 trades. But sometimes I wait longer.

 

[Yamato]

major news from work: panick/anger attack

I need to vent out some stress from work. Today I really lost it.

The bitch from the "efficiency" team was in our office again, and we had our daily meeting with her, at 9.30 am. The usual farce I can hardly bear to take part in. After warning my boss for weeks about this potential danger of me bursting out in anger (I had asked exemption for the bull**** meetings), today I finally could not be quiet about my frustration with their changes. Everything they're changing is a setback and worsening of our work. I kept quiet for the first five minutes when the boss was talking bull**** and pretending there was progress - he was pretending, because the bitch was there, and she was sent to us by the management. So basically everyone around her acts as if she's always right. Except me.

Then enough was enough: I heard one lie too many and I exploded, yelled my reasons ("I worked my ass off for 3 years here and I can't accept to see all my orderly work fall apart because of these absurd changes") against this project for less than 2 minutes, then stormed out of the room, and before doing that I asked to be exonerated from any further bull**** meetings with the ACE team, and I got my way later on during the day, when the boss told me that when she'll be there, I'll be allowed to not show up. I was shaking with anger for 15 minutes after leaving that meeting. It was close to being a panic attack.

Then later I told my boss: "look, I don't have to do what I am doing (statistics and all that). I am not attached to this job. I am only attached to doing things properly. I can teach it to someone else and let them do it, and he will do it any way you want to have it, and screw everything, just the way ACE wants it. But I cannot accept to do my work badly, because these clowns are asking me to. If I don't believe their changes are a good idea, you can't force me to implement them. You can't expect me to mess up my own work". He agreed that I was right. But he's a goddamn idiot, too, because it didn't have to take my anger explosion to stop these stupid changes. He was supposed to stop them in the first place. He's the idiot, along with those other two ACE guys.

 

[Adamus]

I can't help noticing and noting that you're using the term "forward-testing" in a different way than I use it.

For you it seems to be a synonym for "out-of-sample". And in some cases you use it even as a synonym for "in-sample."

I can't help but point out that this is a relevant detail, causing potential misunderstandings.

For me instead "in-sample" is the first part of the data, the one you can optimize your systems on. The "out-of-sample" is the second part of your data, which you've kept hidden, and you use to verify if your system works.

The "forward-testing" means running your systems in real-time on TWS, with real data and real prices, regardless of whether you execute those trades or not (the results are almost identical).

I can't help arguing with you. I can't help but argue with you.

You are right again, it is a poor terminology but one that I learnt from Pardo - Design, Testing and Optimization of Trading Systems. I've been using the term for a couple of decades so it's a case of old habits dying hard.

in-sample and out-of-sample does have the sound of something more scientific but again is ambiguous if you don't know the context.

I'll try to remember to use your terminology when I post here. When in Rome, as they say.

 

[Yamato]

Yeah, I am in Rome, not in San Diego, like Ron Burgundy.

 

[Yamato]

more systems to "fix"

My next little project, for today, is to find a few more systems to work on. Not necessarily systems that have exceeded their drawdown, but systems that have failed to deliver and that also trade a lot: above 150 trades per year.

I cannot accept from such systems to have delivered anything less than 1000 dollars. They can't stay at break-even for 300 trades. That's unacceptable.

Those will be the systems I will fix (most likely) on top of the list I've established here:
http://www.trade2win.com/boards/trading-journals/85510-my-journal-2-a-230.html#post1546004

CL_ID
CL_ON_4
EUR_ID_3
EUR_ID_6
GC_ON_2
JPY_ID
JPY_ID_2
JPY_ON
ZN_ID_2


Once I'll find these new systems to fix, I will get to work on all of them at once, so finally I'll get to the final automation part.

[...]

So, ok. Back from watching tv.

I need to select systems that have traded at least 50 times. That is the first thing. There's 22 of them.

Then I need to focus only on those that have made less than 2000 dollars of profit. There's 13 of them.

Ok, let's see what they are:

CL_ID
CL_ID_2
ES_ID
ES_ON
EUR_ID_2
EUR_ID_3
GBP_ID_4
JPY_ID
JPY_ID_2
JPY_ON
YM_ID
ZN_ID
ZN_ON

Of these, some were already included in previous selection of systems to be fixed. I am going to get rid of them, and produce a new shorter list to be analyzed:

CL_ID_2
ES_ID
ES_ON
EUR_ID_2
GBP_ID_4
YM_ID
ZN_ID
ZN_ON

There's 8 of them. Let's see their trades and profit:

system	Profit	trades made
CL_ID_2	1665.5	59
ES_ID	-1465.4	123
ES_ON	-229.6	152
EUR_ID_2	-1563	90
GBP_ID_4	1325.45	69
YM_ID	-1014.2	170
ZN_ID	-240.415	74
ZN_ON	825.605	362

Each one of them has to be studied in detail to see if it can and should be fixed. By the end of today I will study them and post my thoughts here. The rest of the work will be done in the next few days.

Let us remind ourselves once more that these systems have not failed yet. They have not exceeded their previous maximum historical drawdown and in some cases they're even profitable. So what we first have to really make sure that, after so many trades, we're fixing them because they have failed to deliver and failed in performance compared to what we expected of them.

Not surprisingly, all 8 disapponting systems are from before I learned to use the out-of-sample testing methodology. This is in part because I didn't allow the recent systems to be examined (and the out-of-sample was only used since 10 months ago). And it is also because the new systems didn't place enough forward-tested trades yet. But my estimate is also that the systems created with the new method (the out-of-sample) will fail less than the ones I created without it.

[...]

Back.

I am going to study the situation of those 8 systems now. What do I need? I need the report and the forward-tested trades, to be compared together.

Let's get started. This is more fun than watching tv anyway.

CL_ID_2

Holy cow. This is a perfect example of a system that has not exceeded drawdown in forward-testing (only because we started in July 2009) and yet it has exceeded it by far in the out-of-sample - which I did not use in my testing. Look (the red line divides the end of my all-sample testing):



This is crap. I either fix it, or I'll keep it as it is. But I hope to fix it. It was crap to begin with, which doesn't make things great.


ES_ID and YM_ID (same system on similar futures)

These are all right, but I want to work on them to make them excellent. WIth the knowledge I have now, I can do it. What I mean to say here is that any time we have a frequently trading system like this one that makes little money, we can turn it, with an extra filter, into an excellent system that trades half as many times.




ES_ON

This is the twin of YM_ON which is excellent. I need to keep it as it is to monitor the not yet explained differences in performance between the two systems trading very similar futures. The one on YM made 3600 and the one on ES made 200 dollars. Of course we're trading the one on YM, but I need to keep them as they are to monitor this phenomenon.


EUR_ID_2

Here's another one that trades a lot and that has not failed and can be improved.



But hold on... if I fix these systems that still have not exceeded their past drawdown by 1.5 times, then I would even get busy fixing many other systems in the future before making sure they have failed.

You know what? I am going to keep all these systems except those that have blatantly failed. Also because I have plenty of work on my hands already.

SO, REVIEWING ALL MY CHOICES OF THIS POST:

CL_ID_2 has blatantly failed and so it has to go. The others have not failed yet, so they stay.

Basically it was simpler than I thought. I am merely looking at the out-of-sample which results of the period I didn't have available when I built these systems by using up the whole sample (to optimize them on).

Three more to go:

GBP_ID_4	1325	69
ZN_ID	-240	74
ZN_ON	826	362

After all this work, we need a music break:

GBP_ID_4

This is a special case in that it has not failed but it has had a big drop in 2008 which could be avoided by increasing the range requirement, and raise it as high as the one used for the twin systems EUR_ID_4 (the same code but with a higher daily range requirement). So I will fix it just that way, and keep the trades it made in forward-testing but I will add a note to it.

This is how it looks like with the original code:




This is the way it looks like with a small change in range:




Definitely worth changing, all the more so since that way it will be identical to EUR_ID_4. The trades decrease by 60% and profit decreases by 20%. I'll take it.


Now the two ZN. I am tired.

ZN_ID

I am still stressed out about work, and still upset and offended that i've had to deal with such crap and demeaning bull****. Why am I in a situation where I have to fight for such basic things as good work and things being done orderly? Why am i in a place where people, in order to kiss up to the bosses, get so low as pretending they appreciate the work of a stupid retarded lazy bitch sent to us by management? How can i still be in a place like this? Why did I achieve so little in life to have to be around such stupid lazy despicable individuals? Answer: compulsive gambling. Without the compulsive gambling of the past two years I would not be at the bank now. I would not be around such despicable people. Stupidity is their number one problem I despise, second one is laziness, and third one is a low sense of duty. You should try to get promoted by hard work, not by kissing up and lying and pretending something is good and well done when it is not, and yet things at my place work just the other way: people who pretend and kiss us and lie get promoted, and I get treated like I have some problems. What problems? I can't lie and pretend?

Anyway, back to ZN_ID

Here's another system I cannot touch, because it hasn't failed at all. It did exceed the max historical drawdown but by less than 1.5, so if it hasn't failed on the chart, it hasn't failed on the drawdown point of view... and there's no immediate fixing it, then I have to keep it. Look:



Actually, if anything, dude, this is an excellent one, despite it having been unprofitable so far. No way that i am changing it. We had even selected it for real trading for a few months.

And now, finally:

ZN_ON

Ok, this dude has nothing wrong at all, but it simply has not made money in the past two years, and it's not looking bad either on its equity.



However, since it trades almost every day, I can easily find something that fixes it and makes it trade once a week and make a lot of money instead of trading every day and breaking even for two years in a row. This is the real pearl because it provides a lot of trades for free basically and every day. And if I can add a magic filter to it that adds a little bit of an edge - this is going to be excellent. But I have to be very careful here to not abuse the out-of-sample because by now I now it quite a bit. I must not find a solution that works only because it knows what the out-of-sample is like. I mean, I have to find something on the in-sample and only get to try once if it works or not, preferably. I can't just try adding to it 10 different filters that work on the in-sample and then pick the one that also works on the out-of-sample, because that would be totally cheating.

So... the summary of all this writing and thinking (it helps me to work harder if I post it here), is that I will now have to fix precisely 9 failed systems (appraised yesterday) and 3 other non-failed systems but which could greatly benefit from my fixing. Any system that won't get fixed will keep being traded just as it is. I will now if the fixing works from the out-of-sample verification.

Total systems to ultimately automate or re-automate is now going from 40 to potentially 53. Yes, we can. I was at merely 40 systems when i started this journal, and now I am going to add 53 systems in just a matter of two months. Nothing can stop me now. Just a possible killing rampage at the office (by me).

Both for working on the systems and to avoid a killing rampage which would land me in jail, I need to do this:

1) leave at 3 pm according to my part-time schedule. This is something i have not done in a long time thanks to the "efficency" team, which has slowed me down by an average of 3 hours per day.

2) **** the boss. Basically if he asks me to make any changes, I have to make him understand that he has to remove me from my position, and I won't make any worsening "improvements" just because he's kissing up to the management and listening to the efficiency team. Today I told him he can give my responsibilities to someone else if he wants to screw my work of 4 years. I won't be the one destroying my own work. If he is so stupid and so butt-kissing that he would rather screw the efficiency than tell these idiots to get lost, then he can put someone else in charge of my work and I'll mop the floor instead. There's no way I am undoing the good work I have done in the past years to please some retarded guys who don't know what they are doing and saying. He's an idiot. I've realized my boss is an idiot, today. I am leaving at 3 pm every day from now on.

 

[Yamato]

http://www.youtube.com/watch?v=qivOTbxXRF4

http://www.letmewatchthis.ch/watch-7472-The-Goods-Live-Hard-Sell-Hard

 

[Yamato]

weekly update

The week is still going on, because there's 3 overnight trades opened last night, but up to yesterday's closed trades the situation looks like this (real money):



Comment: quite happy. The drawdown, as usual has gone all the way to 5000, before turning around. It seems to be happening each time (just a coincidence, because the systems traded have been changing and increasing). But this time it lasted much less (hopefully it's over).


The forward-tested systems are shown in the picture below. I have circled in green the 15 showing out of the 16 that are being traded (the CL_ID_3 doesn't appear because its sharpe ratio has fallen below 1) and I have circled in red the 2 that should be included. The YM_ON_2 is a duplicate of ES_ON_2 (which works better, so i trade that one):

 

[Yamato]

All right, let's get started.

I will start from the systems that can improve but that have not failed. No, actually, even better, I will change the systems by symbol, so I don't have to set it up on tradestation each time.

CL_ID	failed
CL_ID_2	not failed
CL_ON_4	failed
EUR_ID_3	failed
EUR_ID_6	failed
GBP_ID_4	not failed
GC_ON_2	failed
JPY_ID	failed
JPY_ID_2	failed
JPY_ON	failed
ZN_ID_2	failed
ZN_ON	not failed

In alphabetical order, I should start from the 3 CL systems.

Wait, a second... I just realized that I am exhausted.

The advantage with economies of scale is that I'll build 53 systems at once. The problem is that I am almost puking by the amount of reasoning required of me all at once. Yeah, because the creative stage is all concentrated in one station, and I am the one doing it, so I am going to take another break now, because given all the stress from work (efficiency team), from housing (there's going to be some heavy works here starting on Monday), and now from these systems... I can't handle it right now.

I have already flipped out at work yesterday due to my phobia for bull**** farces (efficiency team). I have to be very careful right now with my mind. It might not be able to handle the stress coming my way: a few more weeks of the efficiency clowns, a few weeks of works by plumbers and all that (and I am afraid they'll steal stuff from the house)... I have to make sure i don't flip out and do something drastic like killing someone.

I am tired, and stressed out.

However, i have to do this. If I postpone it, I will be stuck at this stage forever, because of how heavy it is. I need to finish the back-testing creation phase and move on to the automation phase. And I need to finish this phase this weekend.

But you know what the biggest source of frustration is? What is going on at work. This "efficiency team" is undoing all the hard work I have done during the last 4 years. It's a terrible feeling. I have organized, created statistics, kept everything in order, made continuous improvements for 4 years, and now, these clowns and my boss (who's guilty of letting them have their way), just to pretend they have performed some miracle with their intervention, because it's sponsored by the management, are asking/forcing me to fix numbers (with fake deceiving new ones), to change procedures, the order of things... it's all a worsening and a setback of all the work I have been doing for 4 years, from Spring 2007 to Spring 2011. All my work is going down the toilet, for the wrong reasons (making some clowns look good) and I cannot bear this sight. At the same time, I am expected and almost required to be quiet, smiling and accommodating (which is what everyone else is doing), and I try to do it, but it's so against my nature that I end up bursting out in anger at meetings about once a week. Let alone that I am complaining endlessly, when I am not in the presence of the ACE clowns. The last time was Friday when I told the boss they are "clowns" and that they're making him lose his mind and idea of what is right and what is wrong.

 

[Yamato]

Before resuming I need some music videos to cheer me up.

 

[Yamato]

fixing CL_ID (failed system)

This system has been back-tested and optimized throughout its sample, which was from 2000 to April 2008. Then it failed, pretty badly:



Can we fix it? Yes, we can. It trades every other day. So we can add another filter, which will reduce its trades by 50% and hopefully avoid the drawdown.

How do we fix it? This is a very simple system. It goes long/short if price is above/below average at a given time of the day, and it stays that way until the end of the day.

Now, i have noticed in these two years that the system fails when SP500 and EUR/USD go down. But EUR/USD is enough. But then I do not want to make a CL system depend on EUR/USD because it has all sorts of wrong and problematic implications in terms of back-testing and automation. I want a system to just depend on its future and nothing else.

So I will try a filter that filters trend. How do we filter trend? We see where price was a week earlier. If we're above it, then it's a long trend, and viceversa.

I am going to first test this on the in-sample (which I didn't use when I created the failed system).

The trend filter is not working.

Getting there. I am throwing away the moving average filter and replacing it with c > yesterday's close (and viceversa for short). This reduces the drawdown.

Ok, I am almost done. It takes me longer than 1 hour per system, maybe 2 good hours, including breaks. I will be done with this work by next weekend. I can't rush it. I either do things well, or i don't them. I am not like the ACE team of clowns.

In this new version of CL_ID we got rid of the moving average requirement and replaced it with c > closed(1).
We cut trades from 1000 to 220. The drawdown almost disappeared entirely. We moved the entry time and the exit time. We added two trend requirements: c > closed(5) and c > closed(6).

Done. Finished. This is a good one. One system to be proud of. I should do the rest of them like this, but now I am dead tired. I will only be able to do one of them per day. Or two at the most. Which means two more weeks of this stuff.

Here's the new equity curve:



For the record, I did allow myself to optimize the entry and exit time on the entire sample. This I can do safely, from my experience and rule-of-thumb guesstimating.

If I'll get home at 3 pm every day, I'll be able to do one per day, and two per day on the weekend, and therefore I'll finish this thing by next weekend. There's no way I can finish this restructuring work by this weekend.

 

[Yamato]

fixing CL_ID_2 (not failed but improvable system... actually failed, too)

All right, let's do another one. Taking advantage of the fact that today tradestation is not giving me the usual error when saving reports.

One thing is for sure. After I'll be done with the automation of the 114, I will not create systems for a long time. If I'll have any energies left, I will restructure the ones that fail.

This CL_ID_2 system was created in the Spring of 2008 (without using the out-of-sample, like all the ones I'll be fixing) and this is what happened afterwards:



Yes, big flop. However, in forward-testing it did not exceed its previous back-tested drawdown, but just because the forward-testing was shorter than the out-of-sample. But it is indeed equivalent to one of the failed systems, so I should call it that way.

The fixing will be easier, because it's very simple. The problem however is that it trades much less, so adding filters will reduce trades to almost unreliable levels and bring almost unacceptable risks of overoptimization and curve fitting.

Ok, what does it do and how do we fix it?

It goes only long, it has an entry period of one hour and it enters if price is above yesterday's close and the close of yesterday is less than the close of the day before (oversold situation) and if price is more than 15 minutes ago plus some portion of the daily range. It is a volatility breakout system. Then it stays long until the end of the day.

In fact it works for the in-sample (which was all the data I had available, since I didn't use any out-of-sample back then), but then it blatantly fails, as the picture above shows.

First I will try to tweak the range coefficient and then if that doesn't work, I will add a new filter, maybe the same filter i used for the previous system, a trend filter.

It's working. I got rid of the daily range coefficient (which was also causing me differences between back-testing and forward-testing) and I added the trend filter.

This is good... I am getting there. Even the short version works.

Ok, here's another one bites the dust... another bites the dust... another one bites the dust:



What i've done is optimize the entry hours across the whole sample. By only working on the in-sample, I had added the range filters (price has to be more than the closes of 5 and 6 days ago for long trades - viceversa for short trades). And finally I have gotten rid of the daily range coefficient. Trades get triggered by price merely being above the previous close. This doesn't really act like a volatility breakout anymore, since there's no daily range required. Maybe I'll put it back now.

Nope, that was useless. I'll keep it as it is.

The magic filters here are the trend filters (c > closed(5) and c > closed(6) for long trades and viceversa for short, with the added closed(4))
Another magic major ingredient is c > c[1]. Maybe I should consider using it in all of my systems. Too hard to review them all right now.

Maybe all systems should have the c > c[1] ingredient. But as i said right now I don't have the energy. I'll keep it in my pool of genes and pass it to further generations of systems by including it in this and other systems.

Now I'll take a break before restructuring the third and last CL system.

And also I need post this song, that was written for me basically:

 

[Yamato]

fixing CL_ON_4 (failed system... actually healthy one)

Ok, let's start the post for later, so I can place the bookmark in the system's folder.

This is another failed system. Here's what it looks like:



I really have to make sure that this system has something wrong with it, because at first sight it doesn't look bad at all. On top of it, it is a system that works perfectly on ES and YM, so I really have to think twice before changing it. Those systems that work across many futures are the most reliable ones. I should not change it just because it is incurring a bad period.

You see, the more CL rises, the more it is bound to fall, so the fact that it fell recently for two months in 2009 and it exceeded its previous fall of two months in 2005, does not mean this bigger drawdown is a sign of failure. Obviously if it rises more, it will fall more. In percentage terms the fall is the same.

Screw this, I will do it right away. I am too pissed at my office and the more i work on this the faster I'll get out of the office. So let's do it.

You know what? I have appraised things and this should be kept as it is. I will remove it from my restructuring hospital. I will send it home as a healthy system.

The same will probably happen to GC_ON_2, which belongs to the same family. I'll do it next.

 

[Yamato]

fixing GC_ON_2 (sick mother ****er)

This one, too, and even more so, is one system that is probably healthy because gold has risen so much that of course when it falls it falls more (in absolute terms, not in percentage terms), and by doing so, it exceeds the previous drawdown, without it meaning that it's become unhealthy.



Nope, wrong. This is a sick one. We can't take this crap from anyone. Not even from a system that has twin systems that are profitable on other futures.

All right, so I am just going to keep going.

Ok, the trend filters work but not well enough, so I can't keep them. I can't just sum up a bunch of filters and come up with one trade per year.

It is not out of laziness. I have done a lot of testing and I am not prepared/knowledgeable enough right now to fix this system - I can't do it. It doesn't work obviously, but I'll keep it as it is, for the sake of monitoring it. Maybe it will resume working. At any rate, I could not find anything worth fixing. It's not good, but it cannot be fixed properly.

I will resume tomorrow. 4 systems taken care of. Only 8 more systems to go (to fix or dismiss). Then finally I can start automating stuff and finally rest.

 

[Yamato]

fixing GBP_ID_4

can't sleep. i'll fix another one.



fixed by increasing the daily range coefficient required to trigger a volatility breakout trade.



Just three more symbols consisting of seven more systems to fix.

For some reason the EUR_ID_4 has the same exact code as GBP_ID_4 has now and yet it doesn't work (big drawdown in the last two years). I'll have to keep it that way because i tried everything but could not fix it.

Anyway, only 7 more systems on 3 symbols to go. It might take as little as 3 days to ge done with this.

I can do 3 systems tomorrow, then I'll have left 4 to do. Easy. The less I have the more energetic I feel.

 

[Yamato]

Can't sleep still. Took all those pills. But can't sleep.

My eyes are closing but I forcefully keep them open. Something inside my mind and spirit makes me do that. I don't know what. It's always been my problem ever since i was a teenager.

Just like for compulsive gambling, it's easy to state some causes and they sound credible, but there's no way I can really prove that I've identified the correct cause. All I can say is that something keeps me awake. Most likely some worries. But I can't be sure about that either.

This is not math, how our mind works is not like a math formula.

For example, I say that at work I am fighting a crusade against laziness of other people. Maybe that is not the truth. Maybe I am just fighting because I am upset about something else and I use people's faults as an excuse to fight them. You know what I mean? It's so easy to deceive ourselves with words. Maybe speaking confuses us more than it clarifies what is actually going on. Maybe being quiet would be better than spitting out wrong appraisals of situations.

Psychology is much more complex and less precise than other sciences. The force of gravity is one thing, and it can explain a lot of events, but our actions are not as easily explained as something caused by the force of gravity. You know what I mean, right?

One of these actions is my insomnia. I can't explain it. I could list ten different plausible explanations and they're all different and they would all sound reasonable - this shows why they're not reliable. And why I can't still understand clearly what the cause is.

I could say I don't do enough exercise, but then would lack of exercise be the cause of it? Maybe both the fact that I don't exercise and the fact that I don't sleep are caused by a common factor, such as the fact that I work, then get into a bad mood, then this takes from me the ability to exercise and to sleep, but the lack of exercise is not causing insomnia - it just coexists with it.

You see, after all the office is like a cage, jail. I am like forced to go to jail and stay there for several hours a day. I am also forced to go into one particular room and do one particular thing... that is obviously a big source of frustration for all of us. Let alone when we had to go to school and sit there, around those jerks, in front of those assholes of teachers for several hours each day. Sounds reasonable, right? But maybe this is not it.

Maybe it all stems from childhood and how my dad made me anxious with things he said to me, constant worrying and criticizing. That sounds very likely, too.

But then maybe it's something else. I could go on for hours with likely causes.

 

[Yamato]

fixing ZN_ON

I might as well do or start another one of those systems to fix.

This trades several times a week, almost every day. This could be fixed to become a gold mine.

This is the present situation. It doesn't look bad. It's just that I've been unlucky on the forward-testing period and it's almost made no money.



I've never really felt that my dad was on my side. This has always burdened me, mentally. All my life. It's made insecure.

After 2 or 3 hours trying, I finally give up. I've checked this system thoroughly from all points of view. It cannot be improved. This actually is a good system - we're just waiting for it to end the drawdown and start working again.

 

[Yamato]

fixing ZN_ID_2

Dude, I am drinking wine because I woke after only 6 hours of sleep, as usual in the weekends. This time I really want to nail it and sleep 9 hours and the only sure way is this: drinking.

While I get drunk I will fix another system.



Dude, this needs no fixing obviously. It opens my eyes as to how forward-tested drawdown exceeding back-tested drawdown sometimes is not enough to say a system has failed. This is a healthy one. It is not one of the top 20 we have, by any means, but it is a healthy one that needs no fixing. I don't care that drawdown was exceeded. This will start making money soon, maybe in a year, but it will resume being profitable.

I now only have 5 more systems to evaluate for fixing and then I will be done. I will do them during next week.

 

[Yamato]

http://www.youtube.com/watch?v=qErGcTGg6eE

http://www.youtube.com/watch?v=0qIzXa-EWkQ

http://www.youtube.com/watch?v=nW6agN1KH8o

http://www.youtube.com/watch?v=EZ2SyLch95U

http://www.youtube.com/watch?v=rC9TeJWS3T0