my journal 2

I see your points. They are all valid ones. As I said, I strongly believe the cause is the Dollar and how it interacts with my LONG-only currencies systems. But the systems somehow sense ahead of time that the dollar is about to reverse, as they did three weeks ago (and the dollar only reversed a couple of days ago, at least as measured by the 40-day moving average).

In other words, if I base the reversing of signals on the Dollar Index (by how I measure it, with the 40-day moving average), then my systems make less money than if I base it on the fact that the previous week was negative. This has been the case for the past 8 months. I also would have preferred your approach... and now you're confirming that it was a good point, and you're making me wonder whether that approach may actually be the best one, and that the "reverse if past week negative" approach only worked better because it was lucky.

By the way, when I say Dollar Index, I really mean the EURO (which is what the Dollar Index is largely based on), because yes my systems are also on the JPY and GBP, but it just gets too complicated to reverse them all based on each own's index (that is, IF I used the Dollar/Euro reversal of signals approach, which at the moment I don't).

The rationale of all this is simple: my systems go long on the EUR and the others. So, if the EUR falls (as the dollar rises), it is quite reasonable to expect that they won't make as much money with their LONG-only strategies. Or rather: I find it reasonable now that I noticed it does happen, and I never thought about it, until when, a few months ago, my systems, unexpectedly, started losing everything they had made until then.

This will work until the EUR (the dollar and the others) moves in trends, because were it to start zigzagging aimlessly... then I'd probably be screwed, but this has never been the case in the past 10 years.

This is really tricky and unpredictable stuff. That is why I hope to repay my debt as soon as possible and at least not worry about that. Sure, I'd like to make money from arbitrage trades and be sure to produce money every day. But I've only got what I've got, and it's already a miracle.

-----

Tonight, as I'll get home, I'll come up with my 5 best systems, to be used with the new capital (coming from the loan). I want to select the best of the best, and precisely:

1) CL: one intraday and one overnight system
2) the best EUR intraday system I've got
3) the intraday GBL system to be traded with 2 contracts
4) one GBP system, for diversification's sake
5) 2 more intraday systems, the best ones
6) 2 more overnight systems, the best ones

For a total of about 10 systems.

Strange but your journal lacks the 'quick reply' formfield. I had to hit the reply button on your last entry to write this.

Anyway, I see where you're coming from. So in your terminology, each system is one algorithm with one instrument? Have you written them up yet as robo-traders / EAs (hate that term)?

The reply button is different because I set it up (see my first post) so that the few people on my ignore list cannot post here, so no "quick reply" because not everyone can post.

Yes, I have a different algorithm per system, even though as I said before, some algorithms are similar and some are even identical (especially those on EUR and GBP).

Yes, they are all back-tested and entirely automated. All I have to do is turn my excel sheet on every day.

Ok, let's start with this job, above-described. Here's all the systems, once again.



My objectives were to pick:

1) CL: one intraday and one overnight system
CL_ID and CL_ON_3 (no doubts here)

2) the best EUR intraday system I've got
EUR_ID_2, EUR_ID_3, EUR_ID_6 (they're all good)

3) the intraday GBL system to be traded with 2 contracts
GBL_ID

4) one GBP system, for diversification's sake
NO GBPs - they're not good enough (would just use up margin)

5) 2 more intraday systems, the best ones
Nope, there's none that are good enough.

6) 2 more overnight systems, the best ones
YM_ON and ZN_ON_2

Let's now see what we've come up with and if there's any margin being wasted:

Intraday
CL_ID, EUR_ID_2, EUR_ID_3, EUR_ID_6, GBL_ID

Overnight
CL_ON_3, YM_ON, ZN_ON_2

The overnight can use two more good systems, because otherwise those systems will trade twice a week.

I will add CL_ON_2 and JPY_ON. I'll also add one ZN_ID and one ZN_ID_2, because they have a low drawdown and do well.

Overall I'll then trade these futures:

Intraday
CL_ID, EUR_ID_2, EUR_ID_3, EUR_ID_6, GBL_ID, ZN_ID, ZN_ID_2

Overnight
CL_ON_2, CL_ON_3, JPY_ON, YM_ON, ZN_ON_2

These 12 systems (2 contracts for the GBL_ID) will bring me from 14k to 50k. Then I will look into changing things, after having repaid my debt. Until 50k I will use the same systems, without increasing contracts, because I might sooner or later encounter a big drawdown.

Here is how they look if I traded them for the past 8 months:



Maximum drawdown in these past 8 months of 5k, which is pretty much all I can handle if I am at 30k, but which would destroy me if it happened when I start at 14k, bringing me to 9k to a lot of sleepless nights.

Adamus, you got me thinking, and I tested again and I tried again (as in the past) the idea of trying to reverse signals based on the "cause" (the EUR going down) rather than the "consequence" (the past week being unprofitable), but it doesn't work at all. I can't seem to make it work properly no matter how many moving averages I try.

Good luck Travis.

That's a nice equity curve by anybody's standards. If it trades a lot, the period is also pretty good.

When you select the models to include, try looking at a correlation matrix. It is not uncommon to find that adding an oddball system that is nothing special alone could add to the overall picture (if uncorrelated, or negatively correlated).

Thanks for the feedback and advice. I like what you say a lot. It makes me feel like I've really achieved something. And at a time when the money is not there to prove it yet / any more (because it did many times), encouragement and a pat on my back is what I need. Yesterday I felt like an idiot, going for my third loan in one year, after repaying the two previous loans, and each time thinking I wasn't going to have any money problems ever again in my life - the proof is that I had already started wasting thousands on restaurants, treating colleagues to everything, and so on.

As far as the "oddball system", unless I already have it/them (which I might), it will be almost impossible for me to include yet another system right now, because I am exhausted, since I've created 40 systems in 5 years, and pretty much most of them in the past 2 years, because I wasted 3 years on useless systems.

As far as your question about how often they trade: they made over 1300 trades in these 8 months, which is almost 10 trades per day, and an average of 32 trades per system. But some started trading only a few months later, so I'd say they make exactly (on average) 10 trades per day.

About their correlation and diversification I can say that there's about 6 groups of systems, which all trade according to different principles, which means that (you have to remember that I am trading 9 different futures), there's about 7 different systems based on the same principle on average. A given principle sometimes doesn't work on all futures, or doesn't work well enough to be worth using.

Travis, I forgot to ask. What sort of forward testing did you do? Have you been running them in simulated trading?

With the automation, how did you do it in excel? Was it with the TWS API? Did you do it all yourself or are you using something from a code library?