my journal 2

[Yamato]

Yeah.

I'm gonna fall into lethargy for a couple of months.

I will rest my eyes.

I will give time to the investors to resume trading or drop me.

I will give time to my dad to ask me if I want any money, and then answer "yes".

I will give myself and everyone time. In the meanwhile, I will rest, and of course I will forward-test my systems as usual. No loans, no extra work, no nothing.

I am going to rest like I haven't done in over a year.

 

[counter_violent]

Will you still be monitoring the systems in play drawdown / equity performance from the 26/9/11 which was the day they were turned off ? I expect you managed to catch the absolute bottom in terms of drawdown. This may give the investors some confidence in the systems and that maybe you were all just unlucky on the timing front!

 

[Yamato]

Yeah, I'll forward test my systems, of course, and this will show they still make money (I am positive about this), but who knows if they'll use that confidence to trade them without me or not (they already told me they don't trust them anymore, but this is bull**** and I don't buy it). Either way, I am better off alone. I have been burned out by this thing, and i can't take it anymore anyway. For a year I felt like the doorman, it was quite humiliating to have to stand up on a daily basis for my honor, dignity, independence, and pride. I can't take it any longer.

I hope that my father, too, won't change his mind and that he will give me 30k or something like that. Then, with my money, and with my systems, I will finally be able to do things right. Provided that I lost my compulsive gambling habit for good, which I won't be sure of, until I'll be alone and trading my own account.

But before I do anything on my own or with others, I must:

1) forget the concept of max combined drawdown and replace it with the concept of combined max drawdowns (plural). I will need to look at the 10 highest combined drawdowns. I will also need to sum up all the drawdowns of all the systems traded and see what happens if they get hit all at the same time. HARD WORK

2) automate the selection of systems to be traded. HARD WORK

This year, with the tasks given and the assignments, I got so burned out that i lost the ability to think, to be creative. I was gradually turned into some executing accountant/programmer doing the menial work, being directed by others, who actually trusted me only where they should not have: as a money manager. So now it looks like it's my fault and failure. Once again, the systems have not failed. The money management has failed. And I have always sucked at that, given my 13 years of compulsive gambling. So there were big problems, big mistakes, but they weren't all mine. Having said this, i still consider it a good and stimulating experience, and I would not have been able to do better on my own. But now I am tired.

I am a long way from becoming the money manager I need to be to trade on my own. I should probably start with a small capital again, but by small this time it is obvious that I mean 30k. With my usual 5k I don't get anywhere. I just get to blowing out my account over and over again. So I am not starting unless i have 30k, and I am not giving my systems to anyone ever again.

I have finally realized that we have not lost because of the systems (as i knew all along), not because we have been unlucky, not because we weren't willing to take the drawdown (which is what I was thinking): we have lost because we have optimized the combined drawdowns, and have done it so much, that we were expecting an unrealistic max drawdown. We... I engaged into the curve-fitting of combined drawdowns.

I used a brute-force optimization software actually, Palisade's Risk Optimizer, the worst idea I ever had. I used it to come up with combinations showing the best back-tested return and lowest back-tested drawdown. Of course you'll find an optimal combination for the past. But the better job you will do at finding that optimal combination, the less reliable that value will be for the future. And guess what? I spent weeks and weeks optimizing that value, both with the software and without it. No wonder the actual combined drawdown was exceeded by so much.

There is no "bad luck" excuse anymore.

We ****ed up. I don't know enough about money management. This might be the right time to start reading some books on this subject, but i don't trust books. 99% of books are bull****. I will wait for people to give me their advice here. But if they criticize too much, then i'll ban them.

 

[megamuel]

Can't you just program the systems to only risk a certain % per trade? Why don't you do that? By the way, gotta say that I truly admire your honesty. Success or failure you are completely honest about it. You even criticise yourself and big yourself up when times call for it. Very admirable. Oh and I am very very drunk so apologies if that sounds a bit gay. But you are Italian so that is probably a normal heterosexual compliment. Lolz x

 

[Yamato]

Thanks for the feedback and compliments. Thanks for noticing that I am honest and transparent and that whatever happens or whatever I think, I write it here. I will keep this quality, because it's one of my best qualities, and by now it's become effortless. Of course, every once in a while, when people take advantage of my honesty to attack me, given that I have made myself vulnerable, then every once in a while I'll ban someone. But only at the rate of 1 person every 2 months. There's 12 people on my ignore list after 2 years of honest journal.

No, sorry, I cannot risk a % of money on a given system/trade. You have to let them run free. I don't use stoplosses nor takeprofits. Everything is ruled by time: time entries and time exits.

 

[Yamato]

Aw, this is just great. We have this convention for our compliance department, in 20 days. It's the usual bull****. Everyone told me to go, that they'll sit next to me to make it less boring and stressful. I don't like conventions, I don't like the bull**** side of it, and the crowded side of it.

But get this: they sent me an email asking me "do you confirm your participation or not?". I don't want to go, they give me the choice of saying that I don't confirm... could I click "yes, I confirm"? Nope. If I have to come, then tell me that I have to come. If you ask me if I want to come or not, or if you ask me if I will come or not, the hell with it: i will say that I am not coming, and this is what I did.

Then a colleague said that I destroyed my career, and I told him it's already destroyed anyway. This was the fun and enjoyable part of today at the office.

They thought they'd pretend to ask me an opinion, and get a submissive "yes", but I took advantage of their rhetorical question, and said "no".

Once again, I am not into bull****, and if they ask me a question, I am very likely to reply with what I think.

 

[pecas]

Hi Travis,
I never posted in this thread but I've been following for the last 50 pages or so with interest.

You started this with a small capital, then, as long as investors came aboard, you enabled new systems or increased the trading size.
You did this for sure, but in case you didn't... Did you test what would have happened if those investors didn't come aboard? Would you have had the same drop in equity, would you be idle as well?

Second comment:
I've been involved in a similar experiment on my own, with systems without tragets/stop losses but just time entries and time exits since May 2011.
I was in the green until August, and increased my trading size accordingly, based on 13 years of backtests and foretests (because, and this is the point, when you don't have stops/targets, that's the best, yet imperfect, way to estimate what the worst case scenario could be). When my system went long on ES the Monday after S&P downgraded US debt, and lost 33% of my capital in one day, I stopped trading.

I've been following my systems since, and they're doing good. But at the same time, I realized I couldn't risk this much, and worked hard to find stops/targets that didn't hurt my profit (and other metrics), but at the same time gave me the certainty to lose no more than a fixed % of my bankroll in just one trade.
In other words, I made a classic money management error, but I'm trying to correct it now.

Sincerely, enabling three systems (SI, GC, CL) that, by your admission, could make +/- 20K (combined) in one trade, with a bankroll of 160K, looks like a money management hazard.

 

[Yamato]

Replying as I read.

Thanks for following the thread with interest for the last 50 pages. You've been very patient and meticulous. Congratulations on your energy and effort.

First of all, I started by blowing out my small account for 13 years. Then, without any more money and desperate, I turned to outside investors. We started with 15k, then after 8 months increased to 30k, then 40k, then 160k.

Accordingly, we increased the traded systems/contracts.

Yes, I know exactly what would have happened if they didn't come aboard: I would not have traded or I would have kept losing money by compulsive gambling for another year and a half.

If I had traded my systems on my own, with capital, I would have probably made similar mistakes, or different mistakes, but probably I would have blown out my account, which is worse than just losing the profit, as we did.

Yes, and now I would be idle.

Mmh, interesting, "you stopped trading". If it was according to plan, it's ok. Otherwise there could be a debate on this, and whether it was good or not.

You're right, and I've been talking to megamuel about this. I cannot have that advantage that allows me to know exactly how much I will lose at the most. The first reason is that it's complex to code a stoploss on excel for TWS. I can't do it, and I can't do it efficiently. On the other hand, my advantage is that my trades always close at the same time, which turns out for me to be more orderly, simpler (as a way of exiting), and even more profitable.

Also, I don't think that, because of my time exits, my losses are out of control. You can only lose so much with 1 contract. But yes, I don't know how big the loss is.

I never said + /- 20k. I said that silver can make + /- 10k, but not the others (you need to read more carefully to find out about that detail). This means quoting me out of context, and misquoting me.

Other than this, your post was ok, so I am not going to say, like i did with bbmac, that you're misquoting me and misrepresenting what i said in order to discredit me. And I will not ban you.

Having pointed out that have misquoted me, it is true - and I have stressed it out before - that it was early to enable the silver systems. But the other systems were fine. At any rate, I've already written plenty about the mistakes we have made and might have made. I will not list this again here.

How about my hypothetical proposal (see below), given that you are italian as well.

How would you like it if I traded your account (let's say a 30k account) with my systems, but without giving you access to the server, and without showing you my code?

You would just make money or lose money, but without knowing why.

Of course you're not going to be interested, but I need to start formulating proposals to people, so this is my first attempt. Also, you never know if the investors might still resume their trading with me, in which case I have an exclusive agreement which doesn't allow me to do anything with other people's money. But i'd like to hear what the objections to such a proposal would be.

 

[pecas]

I never said + /- 20k. I said that silver can make + /- 10k, but not the others (you need to read more carefully to find out about that detail). This means quoting me out of context, and misquoting me.

Sorry if I misquoted you, I read this in your post #3067:

Bad balance of systems. Silver, gold and oil have too much weight in the present combination. So when they trade, this stuff happens. Silver trades cause you a plus 10k and minus 10k per trade. Gold and Oil cause plus 5k and minus 5k per trade. The other choice was not enabling silver, which seemed like a pity, because the systems were very good, and still are. But we weren't ready for it.

I thought it was equivalent to say you could make +/-20K combined with SI/GC/CL, as I reported. But English is not my first language as well, as you noticed.

As for the rest of your answer, I see your point: investors = discipline. One thing I'm not very good at.

I'm not interested at trading a blackbox system because I'm a coder myself. Exchanging experiences is always good though. If you don't mind, I can send you a pm with my trading experience, and you can answer if you want, with no obligation.

 

[Yamato]

Good job on keeping track of what I say and of my posts!

Well, I wouldn't combine trades the way you did (otherwise you'd also have to add up trades from the other 50 systems, and would come up with a potential loss of 100k per day), but if you want to combine them then you could also say that I lost about 25k in one day, not based on your calculation, but on reality (cfr. latest equity curve, the last day of trading).

Yes, there were mistakes and yes, let's work together.

But let's skip private messages, which are no good. We want to keep sharing things with everyone else, 1) for their benefit and 2) I don't want to write the same things twice or thrice (in 2 private messages and then again on the forum).

So let's work on this.

1) how do i automate completely systems' selection (portfolio selection)?

2) how do I do #1 in light of the fact that my 120 systems trade 16 futures which:
a) change price all the time
b) are traded with different contracts, requiring different margins, having different leverage, implicating different risks...

How the hell do i make the whole thing automated, and efficiently so? Throw quick short ideas at me. Don't write the usual essays many people write. I don't have to grade your thesis.

We have very similar problems and this will probably benefit you as well.

The previous semi-automation attempt was "let's not look at systems' performance but let's look at their combined performance". That could very well be looked at, but you must then use an average of the drawdowns rather than the worst possible drawdown, because that could lead you to curve-fit and find one lucky combination of systems that have a low drawdown, which did happen in the last 10 years, but might not happen in the future.

This is very tough. And I can't do this while there's people throwing assignments at me without doing their own homework, like it's been in the last year. This just leads you to being burned out. I guess the only advantage is that I didn't burn out my capital, which is why I was practically obeying orders in the first place.

 

[pecas]

I'll try to be short in describing what I do, also because I didn't solve this problem myself, but at least I automated it.

The setup in short:
1-My systems (just 11 at the moment vs your 120) are all enabled, although I don't have the bankroll to trade them all.

2-My systems know what's my bankroll directly from TWS, and what's the initial margin for the futures they're trading (this one is still a manual input at the moment, must work on it to retrieve it automatically).

3-When a trade setup happens, the system(s) compute the right number of contracts to trade based on these info and on the money management conditions.

Now how do I select:
The systems recalculate on every bar, on first tick of the new bar, so what happens If two systems should trade at the same time: the first one that "ticks" wins... the second one knows a piece of bankroll is busy and will trade for reduced contracts or not trade at all... and so on. So, all is based on the fact that any system, at any given moment, knows what's the free margin left to trade.
Basically, it's a random selection... but random>discretionary imo, so this is a step forward. What I hope is that it will even out in the long term, i.e. selecting sometimes the best, sometimes the worst. Futures that have more ticks will be selected more often, but I don't see a big problem in this: you want to trade the most liquid markets (less slippage).

In the past I tried selecting based on your same reasons: combined drawdown (but did have big dips neverthless, like you); so I can't help you with new ideas. I'm very interested in this matter though, and if I come out with some new idea, I will share it for sure.

Uh Oh, maybe too long anyway... I can be longer if you want to know my entire setup and what softwares/hardware I'm using for this. It's interesting, but I'll respect your request, as this is your thread.

 

[Yamato]

Replying as I read.

What are these "money managements conditions"?

Yes, that is how I used to do it, but this way we don't know what the maximum historical drawdown has been for that systems' combination, because you trade in different sizes according to which trade starts first.

We must get to a more refined and a stricter and more reliable money management, despite the fact that your method maximizes the use of capital available.

With fewer systems, your job is easier, but with a smaller capital, your job is harder, because it leads to doing what you are doing.

After all, we might not have enough of the same problems to be able to work on this together and both benefit from it. But then we could both just work on my problem, so I could benefit from it. It's a problem you will have in the future, so you will benefit from it as well. In other words, we should not work on your current method, because it is not as good in my opinion.

Maybe I know your setup. I use TWS as well, so are you using excel as well, or you created your own platform? Because in that case it means you're a real programmer, not like me.

Anyone, one question to start from (besides those above) is if you have a system on GBP and a system on a bigger contract, don't you think the GBP system should be allowed 2 contracts or more and how do you allocate that?

But forget all my questions as well, if you want, because we might be on different wavelength, so I'd be wasting your time. Well, if you answer, just answer shortly in a total of 5 minutes. Let's not waste time writing essays. I don't have to grade your thesis.

 

[pecas]

Money management conditions are the maxmum risk/trade I set up as % of my bankroll. Now that I have stops/targets I can do that.

Yes actually I didn't solve the problem as you said, I just automated the selection, and I'm very interested in solving it before I need it. Automating all and leaving it unattended is good for my discipline.

I'm not a real programmer, I'm using Tradestation 8 to run my strategies, data flow from TWS, and a software called TradeBullet that executes any Tradestation strategy order directly in TWS. So my coding ability is limited to Tradestation EasyLanguage.

In the past, I used Tradestation 2000i, created a program with AutoIt to read Tradestation orders and transmit them to Excel. With Excel API, I was executing them in TWS, so I have Excel+TWS experience too, but what I have now is much better.

 

[bedsit]

20110923 - YouTube

I thought you may like this scene (Visconti, Mahler)

Straordinaria scena finale di "Morte a Venezia" (Luchino Visconti- 1971) - YouTube

 

[Yamato]

Well, thanks for comparing my thing with visconti. I'll take it as a big compliment, even though of course you'll agree that mine is better.

 

[bedsit]

Let's say they are different.

Maybe you should pursue a filmmaking career - could be more enjoyable than trading.

 

[Yamato]

Yeah, that's why I did trading, to finance my movies. I wanted to be a director, well, actually first I wanted to be a writer, then an actor, then a philosopher, then a lot of other things. Everything required money, so I turned to trading. It might have been faster to start directly with film making. I got sidetracked because my dad pretty much forced me to study political science.

 

[Yamato]

Money management conditions are the maxmum risk/trade I set up as % of my bankroll. Now that I have stops/targets I can do that.

Yes actually I didn't solve the problem as you said, I just automated the selection, and I'm very interested in solving it before I need it. Automating all and leaving it unattended is good for my discipline.

I'm not a real programmer, I'm using Tradestation 8 to run my strategies, data flow from TWS, and a software called TradeBullet that executes any Tradestation strategy order directly in TWS. So my coding ability is limited to Tradestation EasyLanguage.

In the past, I used Tradestation 2000i, created a program with AutoIt to read Tradestation orders and transmit them to Excel. With Excel API, I was executing them in TWS, so I have Excel+TWS experience too, but what I have now is much better.

Replying as I read.

Yes, I see. Now that you have stops you can do that. Yes, it makes sense. But this is not addressing the drawdown, just how much you could lose in one day. It's not telling us for how long your (combined and cumulative) loss is likely to keep going. But then what is the difference between one day and one month? That is why we used the concept of max combined drawdown rather than the maximum loss per day. Because the trades are like a continuous list of events. It doesn't matter how long it takes you to lose everything. What matters is if it's going to happen or not.

Ok, then let's keep discussing on my problem, so it will be useful to you as well. Even if you don't help me, i will still try to solve it on my journal, as always.

Oh, I see: TradeBullet. Me too, I am using Tradestation, except I coded everything on excel. It gives me some more flexibility.

Yeah, I used Tradestation 2000i as well.

Yes, Excel + TWS is what I am using now. I am very happy with it.

Ok, i'll keep you updated in the future weeks on what I do the get back on my feet, and, before I'll resume trading, I'll need to have solved this portfolio automation thing.

 

[Yamato]

another automation attempt

Let's start from a practical point of view, otherwise I get stuck into impossible situations.

1) How do systems get enabled? They need a sharpe ratio above 2 and at least 10 forward-tested trades. How many of these do I have?

26 systems

2) How do systems get disabled? When they drop below a sharpe ratio of 1.

What was the worst forward-tested drawdown if I enabled 1 contract on all these 26 systems today? Roughly 5k.

This method of enabling and disabling seems univocal. Does anyone have anything to say against it?

Then more things will need to be measured, to know how much we are risking, such as the max combined historical relativized drawdown. Ok. I'll do that.

But the point is that we're now not choosing systems based on how well they fit together (which runs the risk of being curve-fitted and risks only working in the past), but based on how good they are individually, and right now.

So:

1) as soon as a systems goes above a sharpe ratio of 2 (with at least 10 trades), it will get traded.

2) as soon as it drops below 1, it will be removed.

bbmac will be happy, as there is no discretion in all this process.

This way we do not have to wait for a combination to fail before disabling systems. And yet we're not enabling and disabling at will, because it has to drop below 1, and it has to rise above 2, for either condition to be met.

Then we have one final problem, which has consequences. Does it make sense that we allocate one contract only for systems trading small future contracts? No, it doesn't.

How do we deal with this problem? I am lost.

I don't think margin but max loss of the system should rule how many contracts of it we are trading. But then max forward-tested loss is not good enough, because in some cases we're only looking at 10 trades.

So it should be max backtested relativized loss. That's what right now makes the most sense to me. All the academics are gone now. And no one will address this issue. Only good at giving general bull**** advice, spitting on my work, and discrediting me. Better gone than having to read two kilometers of posts that say nothing.

So let us look at these 26 systems and their max relativized losses.

system	Rel. Back max loss
CAD_ID_03	-3,922
CL_ID_05	-9,674
CL_ON_03	-7,226
ES_ID_05	-1,806
ES_ID_06	-4,074
ES_ID_07	-1,726
EUR_ID_05	-3,055
GBL_ID_02	-634
GBL_ON_01	-1,912
GC_ID_01	-4,878
GC_ON_01	-4,264
GC_ON_03	-4,264
JPY_ID_04	-2,433
NG_ID_01	-4,236
NG_ID_02	-3,565
NQ_ID_03	-2,198
NQ_ON_01	-2,046
SI_ID_01	-13,178
SI_ON_02	-18,456
YM_ID_03	-1,313
YM_ID_04	-1,467
YM_ID_05	-3,394
YM_ID_06	-1,537
YM_ON_01	-1,960
ZN_ID_03	-1,917
ZN_ID_06	-1,392

Well, anyway. There might be some mistakes because I just realized that my max dd named range is not covering all systems, but I'll fix it, and probably will come up with 30 systems on the list.

The idea here is to allocate 1 contract to that SI system (see above) only if we first have allocated 10 contracts to those showing max losses of 10% of what SI_ID_01 is showing.

So, in case we don't have much capital, things do not start based on how good a system is, but how big is its biggest loss. Does this make sense? I don't know.

Things would start by allocating GBL_ID_02 its contract. Then, if we have more money, we proceed to the GBP systems, but only once we have allocated 2 contracts to GBL_ID_02.

Now I have to stop because i am tired.

Let me stress to myself once again that the problem with the other method, choosing a portfolio of systems based on the max combined drawdown, has the flaw that you do not how lucky that combination has been, especially if you have optimized it several times, for weeks and weeks of tests. Or, alternatively, you could use that, but without expecting the drawdown to hold.

This is a mess. As I said a while ago to bbmac, this should be done, but its implications are huge and I have no idea if I am doing things right.

A slight mistake in the elements of this formula could mean blowing out your account.

Which reminds me it's time to post the failed experiment one more time:



This was not just bad luck, this was not at all a failure by the systems: this was instead all about bad money management. So no work is required on my systems. I need to focus entirely on money management.

I think my castle crumbled to the ground, and I am going to need to study some engineering before I can start building another castle all over again.

I am going to ban a couple of people, just momentarily, because I need to focus on money management and take notes on my journal. I cannot afford anyone discouraging me right now.

Just do not get offended, all negative posters, whom I will ban. I cannot afford to argue or debate right now. I have to think out loud on my own.

 

[Yamato]

I have found this book:
Wiley Money Management Strategies for Futures Traders.pdf

I won't read it though, because I have problems reading, from working too much in the past year.

I am just going to browse through it in the next few days...

**** this book.

I have browsed through it long enough.

I am going to read more on the web, hopefully I will find good summaries, on such places as wikipedia:
http://en.wikipedia.org/wiki/Modern_portfolio_theory

**** the wikipedia entry as well.

I am just going to need to think, think and think. I will do it all by myself. I just need to think and simplify, simplify and simplify.

Now let's say I have ten systems. And they all trade different futures. And they all have a sharpe ratio of 2 and deserve to be traded.

The GBP system only trades a small future, so it deserves to be made bigger. Yes, it does.

It has to equate the CL, GC, SI, HG and all the other mother ****ers, so this principle of equating all systems stands.

How do we equate them? Relativized max loss.

So it's not even about how big the future is, but about what risks each system takes.

Now, if we do not optimize by combined drawdown, then the drawdown is huge, and that is the way it should be.

BUT.

Probably that happened because there was one big systems in there, such as a silver system.

So, here's another principle for automating system selection.

The relativized max loss.

It has to be small for all of them equally. This will automatically reduce the drawdown. It will tend to do so.

Yes, because we still need to look at it, at the relativized max drawdown, and make sure that we are ready to see it doubled in the future.

So, the systems added are going to be enabled from sharpe ratio 2 to higher, but also based on the capital available, they're going to be enabled from relativized max loss of -1000 to higher.

Let's say that, out of those 10 systems, only 3 of them have rel.max losses of -1000. Then we will trade those 3.

Then we move on to the next step and we trade a system with a rel.max loss of -2000. This means we have to enable 2 contracts on the other systems that have a rel.max loss of -1000.

This keeps going up until the system trading silver.

But then comes into play the combined relativized max drawdown. If we cannot bear this (twice that amount), then we have to reduce the systems, and we start reducing from the bigger ones.

So basically these principles are regulating the enabling of systems:

1) sharpe ratio above 2
2) contracts allocated based on rel.max loss
3) need capital to cover the rel.max drawdown times two: if I don't, then the equated contracts get cut down.

TO BE CONTINUED... (in the future days)