Fully Automated Stocks Trading

[flying_dutchman]

@ValeryN Interesting journal!

For the mean reversion strategies, my understanding is that you're looking for stocks that are overbought and oversold and these stocks may or may not be related to one another. Have you ever tried looking for *pairs* of stocks that are statistically correlated in some way via cointegration or some other test, ranking the pairs, and then selecting the best ones for your long/short trades? What are your thoughts on this?

Thank you.

 

[ValeryN]

@ValeryN Interesting journal!

For the mean reversion strategies, my understanding is that you're looking for stocks that are overbought and oversold and these stocks may or may not be related to one another. Have you ever tried looking for *pairs* of stocks that are statistically correlated in some way via cointegration or some other test, ranking the pairs, and then selecting the best ones for your long/short trades? What are your thoughts on this?

Thank you.

Re mean reversion strategies - yes, something like that. I don't expect those to make huge reversions too.

Re stat arb strategies - I haven't looked into them much. Personal reasons being - they tend to be quite capital intense (a lot more money needed to make same return), lots of work required to continuously asses if relationships hold, bigger risks associated with black swans and they tend to requirer lower resolution data = more work / more expensive.

One day, if I have more time I'll probably look into those more.

Those are arguably more suitable for pros with access to higher leverage and manpower, and there only if they are NOT Nobel Prize-winning economists.

Val

 

[ValeryN]

In anyone is wondering if I blew up yet, today's answer - not yet.

Quick update on performance: as I described in recent posts there were some strategies adjustments done recently. I feel like strategies are fairly balanced now and working as expected for their target market conditions. There were a few weeks when I felt a bit out of sync, and when I get that feeling I tend to revise and make adjustments. Last time like that was in August. That's where discretionary element comes into systematic and even automated trading.

My friends keep asking me to start selling my "app". I tried to ask - what would you be buying exactly? And got not much of an answer. I suggested that if I would to sell something where you press a button and it makes you money, that would be priceless. Anything less will probably not be enough.

Val

 

[murray t turtle]

Hope everyone is doing alright.

Seeing today's action I looked at recent closed short positions, and mix of "looking for justice" and curiosity, but there was no justice, not today.. Just look at GME and GSX. Whoever is trading them up doesn't care about market or tomorrow. I think GME wasn't on my list but it's probably more of a luck than doing something particularly smart.

View attachment 250065

Changes I made back in August to both short strategies paid off handsomely. If anyone remembers, my logic was - as market was hitting new highs despite the state of economy and so many people having nothing better to do than learn trading from home, we might as well see a bubble follow.

My pre-august versions were fine for 20 years, except bubble in 90-ies, where they encountered few, somewhat short 20-40% DD bursts. Which didn't seem unreasonable as that market type is pretty rare and rate of recovery was not too bad.

Basically what I did was reducing individual position sizes, average market exposure and used more bubble-friendly setups and ranking. All changes combined managed to keep ARR close to original but DD cut to ~12%. As of now old version is in ~34% DD (basically close to historical max during 90-ies), "after-august" version is around 15%. Sounds pretty bad but in a context - not really. This system is not expected to make money in such market.

Anyway, this post is getting too long. One more thing - good to have thought thru exposure management, probably not too much of a problem for discretionary traders. But with automation lots of things can go wrong if market moves fast and there are many anticipated positions hitting their entry points. Pretty much everything my long systems were willing to get into was triggered. Highlighted in red trades that were not taken. Not often I see list that long. Once again, some work put in July to assure correct exposure management automatically pays off.

View attachment 250064

Val

%%
I like charts like that GME/ as long as its an ETF + those are monthly candles or quarterly /LOL.
I like the way some of those system books help my discretion/LOL.
i would not even let a kid trade that GME in any size; even though a paper trade only would be ok, with it clearly labeled paper trade only......................................................

 

[themickey]

GME was ok, it was a bit of social media circus, the only problem is, what the patrons weren't aware of, outside the big tent when they exited were pens where they were fleeced.
So they paid twice, going in and going out.

"TAKE MY MONEY!"

 

[ValeryN]

It is pretty common to hear people say - doing what is comfortable doesn't make money. Or some variation of that.

Which is probably true. So, for some time now, I've been contemplating if I have became too comfortable with my type of uncomfortable. "falling knifes" and "stocks can shoot to the moon" don't worry me too much (remind me if I blow up on the next GME). And that brings good and bad. The bad part is - it decreases willingness to look at new styles/ideas.

For instance, in current market, trends and momentum make a lot of money. And I haven't had that covered in the past. Mostly because after years of testing ideas I just couldn't find anything that had sufficient number of trades or haven't looked like overfitting. I am also too comfortable with mean reversion. So over time my skepticism towards taking remotely consistent advantage of stocks that just keep going up, or down, have only increased.

The right thing to do it pretty obvious - keeping an open mind and trying to find something that takes advantage of that type of market, had sufficient number of trades, doesn't smell like overfit and doesn't blowup in different market types. Preferably doesn't stay flat for longer than a year.

I started thinking about this more and more in Aug of last year. As what I seen as illogical has became quite normal. And then just stayed this way for another month. Then another, then some more.

Anyway, good part is that my logical has a very strong override on emotions. So I did a bunch of right things that ended up paying off.

Emotionally, when I look at charts of recent big winners, I still feel like those trades shouldn't work. Removing such negativity is something to work on.

Val

 

[Poljot]

Hi Val,
I see you are running like 5 strategies simultaneously but they are of course not profitable all at the same time. Some are better suited to current market then others and later on it will change and others will outperform the first.
Would it be profitable to take the significantly loss making one temporarily offline, until it recovers and let it run in an off line mode (like demo mode).
Later turn it on again when it shows signs of recovery in demo?
Why keep on selling ice cream in winter?

 

[ValeryN]

Hi Val,
I see you are running like 5 strategies simultaneously but they are of course not profitable all at the same time. Some are better suited to current market then others and later on it will change and others will outperform the first.
Would it be profitable to take the significantly loss making one temporarily offline, until it recovers and let it run in an off line mode (like demo mode).
Later turn it on again when it shows signs of recovery in demo?
Why keep on selling ice cream in winter?

It is very tempting until you go thru that cycle

Market is not as predictable as the seasons. In the market there could be winter and summer and back to winter on the same day.

In the example you given - let's say you turn it off on 10% DD, then wait for recovery, then back on, then it loses another 10% next month, now you're are in 20% DD. According to backtest you never hit below 10 and it's all happened in the past many times just like that. Would you turn it off again again wait?

In last 2 years I truly killed only one strategy. And it was very experimental to start with. It gave me great insight into a specific market for the future but some assumptions on liquidity didn't hold which was visible from it not working when market conditions were right and some other stuff.

Otherwise fluctuations within expected boundaries is totally normal. None of them will perform as well as all together. Even considering ones in DD. The beauty of systematic approach is that I can formally test this and this is not just a guess.

But that only might make sense for systematic trading. Descretionary traders probably live or die on their abily to sniff out what's right for current market and adapt appropriately.

Val

 

[_terminus_]

Likely scenario:

1. You lose 10%.
2. You turn the strategy off, "for safety".
3. The strategy recovers 10%, but you can't profit from it because you disabled it.

End results: you lost 10%.

Whereas, if you had let it alone, you'd have lost nothing.

Hi Val,
I see you are running like 5 strategies simultaneously but they are of course not profitable all at the same time. Some are better suited to current market then others and later on it will change and others will outperform the first.
Would it be profitable to take the significantly loss making one temporarily offline, until it recovers and let it run in an off line mode (like demo mode).
Later turn it on again when it shows signs of recovery in demo?
Why keep on selling ice cream in winter?

 

[globalarbtrader]

Hi Val,
I see you are running like 5 strategies simultaneously but they are of course not profitable all at the same time. Some are better suited to current market then others and later on it will change and others will outperform the first.
Would it be profitable to take the significantly loss making one temporarily offline, until it recovers and let it run in an off line mode (like demo mode).
Later turn it on again when it shows signs of recovery in demo?
Why keep on selling ice cream in winter?

Because we don't know for sure that it's winter

GAT