Naive beginner's strategies. - page 4
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Even now I look at the history - here I would buy, and here I would sell, and I would definitely earn 500%.)
Another time I look and analyze losing trades of the TS. Here she has bought, and she is absolutely right, but the trade corresponds to the strategy and it is not TS's fault that the market went in the wrong direction. That is, the trade is correct.
I also analyze losing trades.
In general, such exercises based on history are very useful for strategy formation. Some of these hypotheses are subsequently confirmed and work quite well.
In my opinion, the main issue of strategy building is the author's attitude to losing trades. The attitude is simple - there is no income without expenses.
by manual analysis you won't find it.
by manual analysis you can only find something that makes profit 100% of the time, and there is no such thing.
If you trade with this strategy with 55% of correct trades you'll think that Pronosis only works half the time.
As the result, if I've been on the market, for example, for 7 years or so, the information for my needs becomes less and less, i.e. at a certain stage there is a variant to appear in informational vacuum and start to be engaged already in pure creativity, sometimes meeting lifeless bodies of conquerors of the same peaks who have run out of oxygen for further ascent :) At this stage the trader experiences an enormous overload purely physically, on the background of the lack of ideas and new information, and he is all alone, since he has gone too far and there is no one to help. And then someone may get a second wind and reach his goal, because he can't go back, too much is at stake. And someone will not be able to, and then he will be extremely frustrated, probably retire from business... but he will never forget about the market.
You know the answer about the infantile method of market analysis;)
Approximate chain of events of a trader's formation:
1. classical thechanalysis, fundamental analysis
2. indicators
3. martingale
4. analysis of volumes, different bulletins
5. correlation analysis, pair trading
6. arbitrage
7. hft, market depth analysis
8. leveraging machines in a broad sense (I am here now, having gone through all the previous ones)
9.... n... there may be new unexplored horizons
I used martingale calculations.
I used martingale to calculate volumes.
I have a correlation, but it is a question of how to make money on it.
i know that the indicators do not work for forex.
I do not want to get into arbitrage, brokers started to fight with it.
i don't want to get into arbitrage and i don't want to start to analyze it because brokers have started to fight with them.
i don't like black boxes. i have to know how and why it opens.
Conclusions: think about how to make money on correlations. to try pairs on the stock market. to study hft. to try trading on the news. To try to analyze volumes from futures.
Sometimes a strategy with 55% of correct entries will be very good.
you won't find it with manual analysis.With manual analysis you can only find something that makes a profit 100% of the time, and there's no such thing.
If you trade a strategy with 55% of correct trades, you will think that Pronosis only works half the time.
lerning machines - I don't like black boxes. I have to know how and why it opens.
Not really, that was a crude example. In fact, the whole process can be very well controlled. It's like comparing assembler with a high-level language like scala. In assembler you have to do everything, take into account all the nuances, do a lot of routine work, just like when you develop a conventional TS according to some strategy, as a result the solutions end up being cumbersome, inflexible and at the same time primitive. In machine learning, you're engaged in creating a high-level model that hides private implementations of certain primitive logics from you, allowing you to focus on the overall idea.
I can confidently say that most TCs that have been developed and tested for months can be tested in days with machine learning models. It's another level, you just have to figure out how to work with it. For example, you will never calculate millions of variants using primitive TC programming on an ordinary PC, with MO you can do it in a day on a video card. Do you feel all power of MO and primitiveness of usual approach to TC creation? I repeat, you just need to feel it once and there will be no more talks :)
in last variations of bots I don't look anywhere at all, only at tester and optimizer stats :) like, what haven't we seen there... Moreover, the last bots on some analogues of NS I don't even understand how they trade, the black boxes... it's really cool, I should note... I mean... it's really cool to make such systems, in general another approach to TS development, that's why I've marked the Lerning Machines as the highest level of algotrading
Yes, I agree, the first results of using DM are really impressive. However, I think, after some time you will be disappointed in this approach. Imho, of course, but it's the same eggs the same fitting, only in profile).
Imho, DM should complement the strategy, not replace it. This is connected, above all, with the fact that a full description of the strategy is unrealistic - there are too many factors. These factors would be good to entrust to DM. And the basis of the strategy is usually not a problem.
Yes, I agree, the first results of using DM are really impressive. However, I think, after some time you will be disappointed in this approach. Imho, of course, but it's the same eggs, same fitting, only in profile).
Imho, DM should complement the strategy, not replace it. This is connected, above all, with the fact that a full description of the strategy is unrealistic - there are too many factors. These factors would be good to entrust to DM. And the basis of the strategy is usually not difficult.
Well, yes, no one guarantees the result anyway, but the fact that it facilitates (speeds up) the development of the system is an indisputable fact for me
Well yes, no one guarantees the result anyway, but the fact that it facilitates (speeds up) the development of the system is an indisputable fact for me
That's all you need to know about whether you know TA.
Are you still working with one neuron, or have you already gone further?