Notice: Using capitalization for emphasis, not because I'm screaming. Stefan, Trading is a game of stats. A single trade sequence that seems to be a winner is not useful, in fact, it is misleading. One has to collect stats of the trading performance over relatively long periods of time. For crypto you need at least a couple of years given market changed a lot over this period of time.
You need to show a histogram with the distribution of your PL. Fees are not enough, slippage is about 2x the fees or more (depending on coins). Actually that is with active control of slippage so assume even more than that. You can ignore what I'm saying but it would be nice for people in the community to discuss in an intelligent manner the use of AI or any interesting algo in trading. I find that many people in computer science have no idea about this topic and the ones that know anything don't share ideas or information. But let me tell you in general about this idea "milking", "juicing" and so on. It doesn't work. I explained why. Scale invariance. When you decide on a given % change to use as a parameter to take profit you are basically choosing a time scale (in average), there is a precise correspondence between an average price change and time scales. The problem is there is nothing special about a particular time scale, financial data is scale invariant. Large % changes C are rare and small % changes are many. But the final gain is basically C^n (n is the number of trades), when C is expressed as a return (Price_close/Price_open if it is a long position). It turns out that THERE IS NOTHING TO OPTIMIZE because the final gain is the same it doesn't depend on a time scale. The curve final gain vs % of "juicing" is a flat curve (you asked yourself in a video what this curve looks like and this is the answer). I know it sounds counter-intuitive but it is the hard truth about financial markets in particular crypto given the market is 24/7 so it doesn't have the 24 h periodicity of stocks. The addition of a % loss, pullback % and so doesn't change the reality of this. The main parameter is the % you select to take a profit and that is a meaningless parameter because of scale invariance. What is the result of using this strategy of "juicing"? Without fees, it is 0. Why? Because there is no reason to expect your entry criterium is predicting CONSISTENTLY winning moves. Just using the fact there is a "trend" over a given, arbitrary time scale is not enough to predict what the market would do next. Unless you have found the holy grail that thousands of people have looked for centuries, any indicator I have tested (tried hundreds) doesn't do better than average. So you will get 50 % winning and 50 % losing trades. When you add fees then of course over time you will lose money. What you really want to find is a way to PREDICT the market DIRECTION. I repeat DIRECTION. I'm not sure if it can even be done at this point, at least consistently. Your juicing method is based on exploiting volatility AFTER THE FACT (not a prediction) but volatility is not enough to extract gains. It took me years to realize this and it is not to something that is often discussed in trading. It is a deep truth. The theoretical explanation above is actually the result of years of testing and thinking about these issues. Stefan, you can ignore all this and you should continue playing with your algo. I did the same thing (but also tried to educate myself by reading extensively books and scientific literature on this topic). It is a good learning experience but at least you should understand you need to get consistent stats before blasting publicly you have an interesting trading algo. You should take this as kindness from my side to point out this. Most people don't even bother to tell you and prefer to ignore what you are saying. Good luck anyway, Giovanni On Tue, Oct 4, 2022 at 2:13 PM stefan.reich.maker.of.eye via AGI < agi@agi.topicbox.com> wrote: > Simulation only, but yes we did factor in fees. So eager to use real USDT! > > http://botcompany.de/images/1103170 > > https://t.me/MoneyPrintingTradingBot > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + > delivery options <https://agi.topicbox.com/groups/agi/subscription> > Permalink > <https://agi.topicbox.com/groups/agi/Tc83fbaf97f0beb5a-M59cc63187462e0489a7fdd4a> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tc83fbaf97f0beb5a-M620cf38f108321459f5b461f Delivery options: https://agi.topicbox.com/groups/agi/subscription
