I wrote a bot and gave it 25k to trade

[_terminus_]

Thank you for your appreciation. Initially, I used Montecarlo analysis (see beginning of this thread), later I started to rely solely on backtests. I tested my last strategy over 13 years with good results, so I'm still using it. You can find the test results in the thread.

To avoid overfitting, you have to make sure that the number of trades produced by the backtest is meaningful (that is, high enough) relative to the whole period and the decision points/degrees of freedom of your system.

 

[_terminus_]

+++ Monthly Update +++

February could have been a very good month, ended up being disappointing. The system was up almost 5k, then got whipsawed and lost everything. This is one of the moments when I feel I'm missing a complementing short strategy.

Anyway, let's hold on and move on.

 

[_terminus_]

+++ Monthly update +++

I paused the bot, as the general index trend is sideways.
Therefore, no change relative to the previous month.

In the meantime, that is for two years now since I started this journey, my experiments with neural nets and statistical learning have continued. I think I'm slowly getting onto something interesting, but it will take a few weeks more to be able to start a live test.

 

[ValeryN]

+++ Monthly update +++

I paused the bot, as the general index trend is sideways.
Therefore, no change relative to the previous month.

In the meantime, that is for two years now since I started this journey, my experiments with neural nets and statistical learning have continued. I think I'm slowly getting onto something interesting, but it will take a few weeks more to be able to start a live test.

You're living many traders' dream. Markets keep paying you to continue your study of them

 

[_terminus_]

Thank you for your comment, Valery. I guess feelings of victory are relative hahaha.

For example, right now I can't feel like I'm living a many trader's dream. Not having been able to profit from movements in the last couple of months, annoys me. I have gained and lost like 5k in days, more than once. Yuck. My current strategy is not very well suited for this kind of market regime. The backtest tells me that if I had let it work in the last month or so, I wouldn't have gained or lost anything. In any case, I turned it off and saved on commissions.

The next generation of algos that I plan to roll out will all close the positions by the end of day, no exceptions. I don't want to be exposed longer than one day anymore.

I see you have been getting great returns lately, and this is very good. It shows once again that there are no right markets, only right strategies. Congratulations!

You're living many traders' dream. Markets keep paying you to continue your study of them