I wrote a bot and gave it 25k to trade

[_terminus_]

mr _terminus_, something tells me that you have cogent documentation to go with your system. techno dub reco's? Thx
-
_3 contracts; when would be good to go to 3+?
if that's of interest,
chapters 4, 5 & 6 might be worth a browse...
https://elitetrader.com/et/threads/fixed-ratio-position-sizing-money-management.303441/#post-5012086
-
then there's this,

-

Thank you. Go to YouTube and search for "groove dub techno". The first results, a dozen or so, are all good. You may also want to look for "deep house", especially those by Johnny M. But the real secret is a good audio system and crank up the volume (no headphone allowed).

3 contracts because I'm using 2:1 leverage. If and when the equity will get to 38/40k I'll probably add one more.

My system can be set up to use fixed percentage, fixed risk, fixed contracts and fixed $ position sizing. I choose to use fixed contracts and then change it so as to maintain the leverage roughly constant.

 

[IAS_LLC]

​

To verify that, I ran two Montecarlo simulations, the first on the above backtest, the second on a test system that opened the same number of trades, with the same distribution of duration, but at random times. The upper cloud refers to the backtested system, the lower is the random entry:

How did you monte-carlo your own backtest? What was randomized? ...Or, did you just fit a distribution to your returns and sample from it?

 

[_terminus_]

For the 1.400 trades produced by the backtest, the Montecarlo test was done by scrambling their order at random, repeated 1.000 times. So the percentile rank refers to 1.000 data points.

For the random entry test, I first wrote a routine that generated 35.000 trades with the same distribution of duration as the backtested trades. Meaning that if for example 65% of backtested trades lasted 15 min, the test routine also created 65% of trades with the same duration. The difference was that the opening time was chosen at random. Out of those 35.000 trades, an equal number as the backtested trades (1.400) was chosen at random, then executed in random order. This process was also repeated 1.000 times.

I did the same two tests for the three months since when I went live and the results were as follows.

Actual trades (N=104), same processing as the backtested ones, that is, their order was re-arranged at random:

Random entry trades, also N=104:

Percentile rank of the above plots, red = actual trades executed, green = random entry:

To me, this test is even more telling than that on the backtest, because one recurrent comment that I'm receiving - not only here on ET - is: "Yes, but the last months have been very directional and trending". Which is true, but as this test shows, the probability of opening a bunch of short duration intraday trades at random, and produce a profit, are very low even when the market is in a strong trend.

Unless, of course, you buy and hold by knowing (or betting) in advance that there will be such a strong trend, but that doesn't fit my style.

Please note that these tests prove nothing. The market conditions could change at any time and the system break and start losing money. In fact, I expect that to happen.

 

[IamaMars]

Despite the fact that I am quite skeptical about trading with robots or any other passive systems, it seems to me that you have managed to find an approach that can provide you with income, I would like to clarify with you how you managed to set up the algorithm and did you have situations when your system failed or caused a drawdown because the robot failed to catch a certain price movement? I'm just saying that people are experimenting and looking for new ways to make money, but in fact, I still believe that everyone has to personally control their money so that capital is always safe and you can act confidently in the world of trading.

 

[Zithiel89]

It seems to me that trading with a passive system is both a good and a bad solution. On the one hand, if you don't have enough time or knowledge and you realize that you can't do it yourself, you should definitely look for something like that, but don't allow blind work, everything should be tested in practice, with small amounts and charts, where you can always see the potential of the movement. On the other hand, I wouldn't abuse this approach, because it can relax you and you will basically stop trying to find something that will help you to get into the market yourself. Or you can potentially combine active and passive trading to significantly increase your chances of winning.

 

[_terminus_]

it seems to me that you have managed to find an approach that can provide you with income

I guess it's too early to state that, even with the benefit of the doubt. Let's see what happens in a couple years and then talk again about it. In the meantime I will cross fingers and everything I have.

how you managed to set up the algorithm

What do you mean exactly?

did you have situations when your system failed or caused a drawdown because the robot failed to catch a certain price movement?

No. First, because I tested it for several months using the IB paper trading account, which is an invaluable tool for developers. I wrote test algos that bought and sold like crazy, for debugging. Second, because the system went live only in November, so there is still time for discovering other bugs.

No software is immune from errors, but on this side I'm quite confident there will be no major disaster due to the trading framework. I guess it's more likely that the trading algos will stop producing profit before any such disaster occurs.

I still believe that everyone has to personally control their money so that capital is always safe

And you are right. I think exactly the same. What differs is the confidence in the tools and skills that one has to achieve it. As I said I don't consider myself a good trader, if by "trader" you mean trading by hand, following an instinctive/discretionary method. But I can code.

It's similar to driving a car. Driving is one of the most dangerous and risky activity that a person can do, yet pretty much everyone feels confident when they are the driver. People hate to delegate to other entities what they deem important. Self-driving cars are another glaring example. But the feeling of safety can be deceptive...

I do control my money, the program sends me a message every time it opens/closes a trade and I monitor its operation via web. Should any anomaly occur, I can connect to the trading computer and take the necessary actions. And in case the computer crashes I can still use the IB app from my phone and connect directly to my account.

 

[_terminus_]

everything should be tested in practice, with small amounts and charts, where you can always see the potential of the movement

Yes, I monitor the charts and started with smaller positions.

I wouldn't abuse this approach, because it can relax you and you will basically stop trying to find something that will help you to get into the market yourself

There are more options, for example to let the algo work and in the meantime keep developing and testing more algos.

 

[d08]

Despite the fact that I am quite skeptical about trading with robots or any other passive systems, it seems to me that you have managed to find an approach that can provide you with income, I would like to clarify with you how you managed to set up the algorithm and did you have situations when your system failed or caused a drawdown because the robot failed to catch a certain price movement? I'm just saying that people are experimenting and looking for new ways to make money, but in fact, I still believe that everyone has to personally control their money so that capital is always safe and you can act confidently in the world of trading.

Incremental improvement. You eliminate problems until things run without issues and regarding missed fills, those are part of the model. It's not like it's any better putting the trade on manually. If a fill was missed, you don't chase the price.

 

[TickSeeker]

Hi guys thank you for your answers.
Concurrent programming...

How did you implement the concurrency with IB API libraries ?

For example did you use callbacks, state machine or co-routines ?

 

[TommyR]

did it do okay in the 11pcter? remeber the next 11 percenter is much less in dollars so things are getting brighter