Programming language
It does not matter whichy technology and programming language you choose to write trading bot, you should focus on algorithms that will help your bots run better. A good bot should anticipate market behavior. Bring us the profit and guarantee a low risk of capital loss. Basic approach would be seting 'border value' and buying when red selling when the profit exceeds the assumed percentage. Such approach may works but probability of loss is high.
Markov chains
To understand the markov chains will present several situations in which they apply.
Student takes part in the class once a week. If in one week the student came prepared the probability that he will be prepared in the next week is 0.7 on the other hand, if in a given week the student is unprepared for classes, the next week the probability of being prepared is 0.2. Markov chains will help you answer the following questions, If a student is unprepared for a week(red), how long will it take for him to prepare(green)? Or If in a given week the student is prepared for the class, how long will we have to wait until he is imprepared?(how long will it be green?)
If it's so easy then why is it so hard?
The biggest problem is gathering the right amount of data to determine the probability of a particular event. In the beginning the bot should walk without turning real money. The later bot starts functioning on the real exchange, the better the results will be.
Additional resources
If you still do not understand how markov chains works, check http://setosa.io/ev/markov-chains/ for visual explenation or http://www.statslab.cam.ac.uk/~rrw1/markov/ for full course in this field.
The only issue with using Markov chains to trade is that they are vulnerable to events they have never seen. There are few methods that attempt to address this issue (back-off smoothing), but even still these models are fragile to low probability, high impact events
Thanks for your input. That's why it program should collect data as long as possible so the probability of occurrence of an unknown event will be as low as possible.
I agree, but I also add that you should probably add some risk-controls and monitor the model to avoid losing money during rare events.
I agree, the biggest problem in algos is still the uncexpected big fundamental news and sudden changes in market sentiment. Maybe more development in quantitative data gathering/analysis and model of neuron networks will bring up more and more flexible algos in the future.
This post has been ranked within the top 50 most undervalued posts in the second half of Jun 27. We estimate that this post is undervalued by $35.95 as compared to a scenario in which every voter had an equal say.
See the full rankings and details in The Daily Tribune: Jun 27 - Part II. You can also read about some of our methodology, data analysis and technical details in our initial post.
If you are the author and would prefer not to receive these comments, simply reply "Stop" to this comment.
IDC approves of this post.
Markov chains are also very useful in other domains, like particle physics when high-dimensional scans must be achieved! I actually didn't know some people were considering using them for designing a trading bot.
Google pagerank algorithm is using markov chains also. I thought it is normall to implement markov chains in trading bot as some predictive algorithm logic. Anyway in the time of crisis(red like now) runing bot is hazardous
Yeah, it may work better when the amount of waves is smaller ;)
Are you the creator of steemStem?
One of them :)
Thanks for the upvote!
;)
good post, great info. thanks for sharing
I'm glad you like it :)
Nice... I've been looking into trading bots for a few wks now... still just skimmed the surface of this monster. What do you think of bitconnect's trading bot?
I would try unknown bots only with small amounts.
good info! thx!
followed+upvoted+resteemed